Re-imagining Algorithmic Fairness in India and Beyond
Open accessAbstract
Conventional algorithmic fairness is West-centric, as seen in its subgroups, values, and methods. In this paper, we de-center algorithmic fairness and analyse AI power in India. Based on 36 qualitative interviews and a discourse analysis of algorithmic deployments in India, we find that several assumptions of algorithmic fairness are challenged. We find that in India, data is not always reliable due to socio-economic factors, ML makers appear to follow double standards, and AI evokes unquestioning aspiration. We contend that localising model fairness alone can be window dressing in India, where the distance between models and oppressed communities is large. Instead, we re-imagine algorithmic fairness in India and provide a roadmap to re-contextualise data and models, empower oppressed communities, and enable Fair-ML ecosystems.
AI snapshot
Written from the paper’s full text.
TL;DR
The paper de-centers West-centric algorithmic fairness by analyzing AI power in India through 36 expert interviews and discourse analysis, arguing that core fairness assumptions about reliable data, unbiased ML makers, and accountable ecosystems fail in the Indian context. It proposes a roadmap to re-contextualise data and models, empower oppressed communities, and enable Fair-ML ecosystems rather than only localising model fairness.
Key findings
- Data and model distortions: datasets are not always reliable due to socio-economic factors; half the Indian population lacks Internet access, with the excluded half primarily women, rural communities, and Adivasis; 67% of Internet users are male; models overfit digitally-rich profiles, typically middle-class men; caste, gender, and religion require different fairness implementations.
- Double standards and distance by ML makers: Indian users are perceived as “bottom billion” data subjects and “petri dishes” for intrusive models, with poor recourse; AI makers are described as distant and largely privileged class/caste males, e.g., Dalit sanitation workers tracked with microphones, GPS, cameras, and SIMs.
- Unquestioning AI aspiration: AI is aspirational in the Indian state, media, and legislation and is adopted early in high-stakes domains, while a weak ecosystem of journalists, researchers, and activists inhibits meaningful fairness; of 138 FAccT papers in 2019–2020, only one substantially engaged a non-Western context—Marda’s paper on New Delhi’s predictive policing.
- Proposed roadmap: re-contextualise data and models, empower oppressed communities through participatory action, and enable Fair-ML ecosystems; reservations are discussed as an Indic justice approach, with quotas ranging from 30% to 80% depending on policy.
Methods
- Qualitative interviews with 36 expert researchers, activists, and lawyers working closely with marginalised Indian communities: Computer Science (11), Activism (9), Law and Public Policy (6), Science and Technology Studies (5), Development Economics (2), Sociology (2), and Journalism (1); all had 10–30 years of experience; 24 were based in India, 2 in Europe, 1 in Southeast Asia, and the rest in the USA; 25 self-identified as male, 10 as female, and 1 as non-binary.
- Semi-structured one-hour interviews in English via video conferencing, captured through field notes and video recordings; recruitment used purposeful sampling until saturation; respondents were compensated except government employees; transcripts were inductively coded with seven top-level categories and analysed using feminist, decolonial, and anti-caste lenses.
- Discourse analysis of algorithmic deployments and emerging policies in India starting from Aadhaar (2009), including Indian news publications, policy documents, community media, and prior research.
Limitations
- Stated: the authors acknowledge that their interpretations of research ethics may not be universal, and that all authors come from privileged positions of class and/or caste; three authors are Indian and two are White.
- Stated: the concerns raised may also be true of other countries, and the broader goal should be global approaches to Fair-ML; the authors describe the challenges as large and open-ended.
- Evident from the text: the sample consists of 36 experts rather than directly affected community members, interviews were conducted in English, recruitment was purposive, 25 of 36 respondents were male, and the discourse analysis relies on secondary sources; findings are qualitative and interpretive rather than statistically generalizable.
Model-written summary. Check it against the paper before relying on it.
Integrity signals
Registry
Cites
Statistics
Claims the paper makes
0 of 129 cited sources checked · 0 supported · 0 flagged
- Findingown data
Proxies may not generalise even within India, as with asset ownership meaning different things in Mumbai and rural areas.
“Proxies may not generalise even within the country, e.g., asset ownership: “If you live in Mumbai, having a motorbike is a nuisance.If rural, you're the richest in town.”(P9, CS/IS researcher).”
Quote in the paperSubstantial claim - Findingreasoning
Infrastructures and social contracts in India challenge the assumption that datasets faithfully represent people and phenomena.
“Data and model distortions: Infrastructures and social contracts in India challenge the assumption that datasets are faithful representations of people and phenomena.”
Quote in the paperSubstantial claim - Findingown data
In India, data is not always reliable because of socio-economic factors.
“We find that in India, data is not always reliable due to socio-economic factors, ML makers appear to follow double standards, and AI evokes unquestioning aspiration.”
Quote in the paperSubstantial claim - Findingnone given
Ground truth on full names, location, contact details, biometrics, and usage patterns can be unstable, especially for marginalised groups in India.
“Ground truth on full names, location, contact details, biometrics, and their usage patterns can be unstable, especially for marginalised groups.”
Quote in the paperNot a substantial claim - Findingcitation
Data was reported missing because of artful user practices to manipulate algorithms, motivated by privacy, abuse, and reputation concerns.
“Data was reported to be 'missing' due to artful user practices to manipulate algorithms, motivated by privacy, abuse, and reputation concerns.”
Quote in the paperSubstantial claim
- Findingreasoning
The one-to-one correspondence between user data and people often fails in India.
“However the one-to-one correspondence in datasets often fails in India.”
Quote in the paperNot a substantial claim - Findingown data
Consent to a data worker stems from high interpersonal trust and respect, relationships that may not be transitive to downstream AI applications.
“Many discussed how consent to a data worker stemmed from high interpersonal trust and holding them in high respect-relationships which may not be transitive to the downstream AI applications.”
Quote in the paperSubstantial claim - Findingown data
Household dynamics affected data collection, for example male heads of households often answered surveys on behalf of women but responses were recorded as women's.
“A few respondents pointed to how household dynamics impacted data collection, especially when using the door-to-door data collection method, e.g., how heads of households, typically men, often answered data-gathering surveys on behalf of women, but responses were recorded as women's.”
Quote in the paperNot a substantial claim - Findingnone given
AI in India evokes unquestioning aspiration.
“We find that in India, data is not always reliable due to socio-economic factors, ML makers appear to follow double standards, and AI evokes unquestioning aspiration.”
Quote in the paperSubstantial claim - Findingcitation
Using phone numbers as a proxy for identity in account creation fails in the non-West because of device sharing, multiple SIM cards, and frequent number changes.
“However, this approach fails due to device sharing patterns [178], increased use of multiple SIM cards, and the frequency with which people change their numbers.”
Quote in the paperNot a substantial claim0 of 1 sources checked- cites [178]
Not checked in the free run.
- Findingreasoning
User identity can be mis-recorded by data collection instruments that assume static individual correspondence or expected behaviour.
“User identity can be mis-recorded by the data collection instrument, assuming a static individual correspondence or expected behaviour.”
Quote in the paperSubstantial claim - Findingown data
Machine-learning makers in India appear to follow double standards.
“We find that in India, data is not always reliable due to socio-economic factors, ML makers appear to follow double standards, and AI evokes unquestioning aspiration.”
Quote in the paperSubstantial claim - Findingown data
Location may not be permanent or tethered to a home, for example migrant workers regularly travel across porous nation-state boundaries.
“Several respondents mentioned how location may not be permanent or tethered to a home, e.g., migrant workers regularly travel across porous nation-state boundaries.”
Quote in the paperNot a substantial claim - Findingown data
Socioeconomic and demographic datasets at national, state, and municipal levels are frequently unavailable for public fairness research in India.
“Many respondents pointed to the frequent unavailability of socioeconomic and demographic datasets at national, state, and municipal levels for public fairness research.”
Quote in the paperSubstantial claim - Findingcitation
AI's technical illegibility was said to place recourse beyond groups disadvantaged by literacy, legal, and educational capital.
“Further, the high-tech illegibility of AI was noted to render recourse out of reach for groups marginalised by literacy, legal, and educational capital (see [208] for 'hiding behind a computer').”
Quote in the paperSubstantial claim0 of 1 sources checked- cites [208]
Not checked in the free run.
- Findingown data
AI models in India overfit to good data profiles of digitally-rich, privileged communities because of poor cultural understanding and exclusion by AI makers.
“Respondents described how AI models in India overfitted to 'good' data profiles of the digitally-rich, privileged communities, as a result of poor cultural understanding and exclusion on part of AI makers.”
Quote in the paperSubstantial claim - Findingcitation
Many respondents said human infrastructures such as street-level bureaucrats, administrative offices, and front-line workers, who provided recourse to marginalised Indian communities, were removed in AI systems.
“Many respondents shared how street-level bureaucrats, administrative offices, and front line workers-the human infrastructures [182] who played a crucial role in providing recourse to marginalised Indian communities-were removed in AI systems.”
Quote in the paperSubstantial claim0 of 1 sources checked- cites [182]
Not checked in the free run.
- Findingnone given
Model re-training left new room for bias because of a lack of Fair-ML standards for India.
“Model re-training left new room for bias, though, due to a lack of Fair-ML standards for India,”
Quote in the paperNot a substantial claim - Findingown data
Respondents linked the embrace of AI to earlier high-tech interventions like Aadhaar, MGNREGA payments, and the NRC.
“The AI embrace was attributed to follow the trajectory of recent high-tech interventions (such as Aadhaar, MGNREGA payments, and the National Register of Citizens (NRC)).”
Quote in the paperNot a substantial claim - Findingown data
Several interviewees said marginalised users faced an even worse absence of recourse.
“Several respondents described how lack of recourse was even more dire for marginalised users.”
Quote in the paperSubstantial claim - Findingown data
Many interviewees said AI makers, whether local or foreign, public or private, used Indian users as test beds for models.
“Several respondents discussed how AI developers, both regional and international-private and statetreated Indian user communities as 'petri dishes' for models.”
Quote in the paperSubstantial claim - Findingcitation
Popular lending apps equate good credit with whether the user called their parents daily, had over 58 contacts, played car-racing games, and could repay in 30 days.
“Popular lending apps equate 'good credit' with whether the user called their parents daily, had stored over 58 contacts, played car-racing games, and could repay in 30 days [56].”
Quote in the paperNot a substantial claim0 of 1 sources checked- cites [56]
Not checked in the free run.
- Findingreasoning
Model inputs in India appear disproportionately skewed because of large disparities in digital access.
“Model inputs in India appear to be disproportionately skewed due to large disparities in digital access.”
Quote in the paperSubstantial claim - Findingown data
Several respondents described AI makers as having a transactional mindset toward Indians, viewing them as agency-less data subjects whose behavioural traces improved models.
“Several respondents described how AI makers had a transactional mindset towards Indians, seeing them as agency-less data subjects that generated large-scale behavioural traces to improve models.”
Quote in the paperSubstantial claim - Findingown data
The sub-groups with access to underlying variables for data-rich profiles, such as money, mobility, and literacy, are typically middle-class men.
“Respondents noted that the sub-groups that had access to underlying variables for data-rich profiles, like money, mobility, and literacy, were typically middle-class men.”
Quote in the paperSubstantial claim - Findingown data
Almost every interviewee said AI creators and researchers, even local ones, were far removed from the Indian communities they served.
“Nearly all respondents described how AI makers and researchers, including regional makers, were heavily distant from the Indian communities they served.”
Quote in the paperSubstantial claim - Findingown data
Interviewees said the rare recourse available to Indians was frequently culturally insensitive or degrading.
“Respondents described that when recourse existed, it was often culturally-insensitive or dehumanising, e.g., a respondent was violently questioned about their clothing by staff of a ride-sharing application, during redressal for an assault faced in a taxi (also see [177] for poor recourse).”
Quote in the paperSubstantial claim0 of 1 sources checked- cites [177] source
Not checked in the free run.
- Findingcitation
Many interviewees said Indian public-sector AI was seen as a way to modernise and fix long-standing inefficiencies in resource delivery.
“Many respondents described how public sector AI projects in India were viewed as modernising efforts to overcome age-old inefficiencies in resource delivery (also in [175]).”
Quote in the paperNot a substantial claim0 of 1 sources checked- cites [175] source
Not checked in the free run.
- Findingown data
Almost every interviewee brought up reservations as a way to pursue algorithmic fairness in India.
“Nearly all respondents discussed reservations as a technique for algorithmic fairness in India.”
Quote in the paperSubstantial claim - Findingown data
Respondents described Indian technology journalists as covering app launches and investments more than algorithms and their impacts.
“Respondents described how technology journalists in India-a keystone species for public discourse-covered app launches and investments, and less on algorithms and their impacts.”
Quote in the paperNot a substantial claim - Findingown data
Many interviewees said new 'AI for good' projects often excluded minorities.
“Many respondents pointed to how emerging '4good' deployments tended to leave out minorities.”
Quote in the paperSubstantial claim - Findingreasoning
At the study's time, algorithmic bias was not a prominent topic in Indian public awareness.
“Issues of algorithmic bias were not widely raised in the public consciousness in India, at the time of the study.”
Quote in the paperNot a substantial claim - Findingown data
Many interviewees described opacity across inputs, model behaviour, and outcomes in India, beyond the usual black-box problem.
“In contrast to the 'black box AI problem', i.e., even the humans who design models do not always understand how variables are being combined to make inferences [171], many respondents discussed an end-to-end opacity of inputs, model behaviour, and outcomes in India.”
Quote in the paperSubstantial claim0 of 1 sources checked- cites [171]
Not checked in the free run.
- Findingown data
The authors showed that many assumptions behind algorithmic fairness and its supporting infrastructures do not hold in India.
“In Section 4, we demonstrated that several underlying assumptions about algorithmic fairness and its enabling infrastructures fail in the Indian context.”
Quote in the paperSubstantial claim- Rests on an assumption: The views and experiences reported by the interview respondents are representative enough of Indian AI practices to support general conclusions about India.
- Findingown data
Several interviewees said diverse stakeholders were excluded from public-sector AI decisions, laws, and policies.
“Several respondents discussed a lack of inclusion of diverse stakeholders in decision-making processes, laws, and policies for public sector AI.”
Quote in the paperSubstantial claim - Premisereasoning
Conventional algorithmic fairness is centered on Western concerns, including its subgroups, values, and methods.
“Conventional algorithmic fairness is West-centric, as seen in its subgroups, values, and methods.”
- Also stated in: § 1 INTRODUCTION · plain
“Despite the exponential growth of fairness in Machine Learning (AI) research, it remains centred on Western concerns and histories-the structural injustices (e.g., race and gender), the data (e.g., ImageNet), the measurement scales (e.g., Fitzpatrick scale), the legal tenets (e.g., equal opportunity), and the enlightenment values.”
Quote in the paperSubstantial claim - Also stated in: § 1 INTRODUCTION · plain
- Interpretationreasoning
Fairness being treated as the default moral standard and signifier of justice is itself evidence of cultural situatedness.
“Fairness' status as the de facto moral standard of choice and signifier of justice, is itself a sign of cultural situatedness.”
Quote in the paperSubstantial claim - Premisereasoning
Conventional measurements of algorithmic fairness rest on several assumptions drawn from Western institutions and infrastructures.
“Conventional measurements of algorithmic fairness make several assumptions based on Western institutions and infrastructures.”
Quote in the paperSubstantial claim - Interpretationreasoning
The assumptions behind conventional algorithmic fairness may not hold in much of the non-Western world.
“We argue that the above assumptions may not hold in much else of the world.”
Quote in the paperSubstantial claim - Premisecitation
The fairness community often repurposes US legal terms such as disparate impact, disparate treatment, and equal opportunity to compare technical fairness properties by analogy with legal concepts.
“The community also often repurposes terminology from US legal domains, such as “disparate impact”, “disparate treatment”, and “equal opportunity”, or use them as points of triangulation, in order to compare technical properties of fairness through analogy with legal concepts [81,82].”
Quote in the paperNot a substantial claim0 of 2 sources checked- cites [81] source
Not checked in the free run.
- cites [82]
Not checked in the free run.
- Premisereasoning
Biases relevant to other geographies and cultures, such as those affecting Adivasis and Dalits, are not explored in fairness research.
“While other axes of discrimination and injustices such as disability status [91], age [59], and sexual orientation [76] have gotten some attention, biases relevant to other geographies and cultures are not explored (e.g., Adivasis (indigeneous tribes of South Asia) and Dalits).”
Quote in the paperSubstantial claim0 of 3 sources checked- cites [91]
Not checked in the free run.
- cites [59]
Not checked in the free run.
- cites [76]
Not checked in the free run.
- Interpretationreasoning
Defining algorithmic fairness solely through a Western lens is dangerous.
“The above works point to the dangers in defining fairness of algorithmic systems based solely on a Western lens.”
Quote in the paperNot a substantial claim - Interpretationreasoning
Naively generalising fairness to the non-West risks failing to keep AI deployments in check.
“While algorithmic fairness keeps AI within ethical and legal boundaries in the West, there is a real danger that naive generalisation of fairness will fail to keep AI deployments in check in the non-West.”
Quote in the paperSubstantial claim - Premisecitation
Proxy categorisation in fairness analyses carries Western biases and origins, as with the Fitzpatrick skin type originally developed for UV sensitivity.
“Even the categorisation of proxies in fairness analyses have Western biases and origins; e.g., the Fitzpatrick skin type is often used by researchers as a phenotype [46], but was originally developed to categorise UV sensitivity) [73].”
Quote in the paperNot a substantial claim0 of 2 sources checked- cites [46]
Not checked in the free run.
- cites [73] source
Not checked in the free run.
- Premisecitation
Cultural relationships such as power distance and temporal orientation mediate how much importance is placed on fairness.
“Furthermore, cultural relationships such as power distance, and temporal orientation, are known to mediate the importance placed on fairness [122].”
Quote in the paperNot a substantial claim0 of 1 sources checked- cites [122]
Not checked in the free run.
- Premisecitation
Algorithmic fairness has been connected to Western philosophical concepts including egalitarianism, consequentialism, deontic justice, and Rawlsian distributive justice.
“Connections have been made between algorithmic fairness and Western concepts such as egalitarianism [38], consequentialism [142,167], deontic justice [38,142], and Rawls' distributive justice [99,142].”
Quote in the paperNot a substantial claim0 of 4 sources checked- cites [38]
Not checked in the free run.
- cites [142]
Not checked in the free run.
- cites [167]
Not checked in the free run.
- cites [99]
Not checked in the free run.
- Premisecitation
AI ethical principles are often interpreted, prioritised, contextualised, and implemented differently across the globe.
“The call for a global lens in AI accountability is not new [84,155], but the ethical principles in AI are often interpreted, prioritised, contextualised, and implemented differently across the globe [98].”
Quote in the paperSubstantial claim0 of 3 sources checked - Premisereasoning
Algorithmic fairness research framings often rely implicitly on US laws such as the Civil Rights Acts and Fair Housing Act, and on US legal concepts of discrimination.
“Since early inquiries into algorithmic fairness largely dealt with US law enforcement (predictive policing and recidivism risk assessment) as well as state regulations (e.g., in housing, loans, and education), the research framings often rely implicitly on US laws such as the Civil Rights Acts and Fair Housing Act, as well as on US legal concepts of discrimination.”
Quote in the paperSubstantial claim - Premisecitation
Categories such as race and gender in fairness research are culturally and historically situated.
“However these categories are culturally and historically situated [85].”
Quote in the paperSubstantial claim0 of 1 sources checked- cites [85] source
Not checked in the free run.
- Premisecitation
Traditional societies often value restorative justice, which focuses on repairing harms, rather than fairness.
“Traditional societies often value restorative justice, which emphasises repairing harms [44], rather than fairness, e.g., contemporary Australian Aboriginal leaders emphasise reconciliation rather than fairness in their political goals [94].”
Quote in the paperNot a substantial claim0 of 2 sources checked- cites [44]
Not checked in the free run.
- cites [94] source
Not checked in the free run.
- Premisecitation
Most fairness research examines racial and gender biases in models, the two dimensions that dominate American public discourse.
“The majority of fairness research looks at racial [46,57,121,125,183] and gender biases [42,46,197,219] in models-two dimensions that dominate the American public discourse.”
Quote in the paperSubstantial claim0 of 8 sources checked- cites [46]
Not checked in the free run.
- cites [57]
Not checked in the free run.
- cites [121] source
Not checked in the free run.
- cites [125] source
Not checked in the free run.
- cites [183]
Not checked in the free run.
- cites [42]
Not checked in the free run.
- cites [197]
Not checked in the free run.
- cites [219]
Not checked in the free run.
- Premisecitation
In some cases, data workers have been shown to fudge data without actual conversations with affected people.
“In some cases, though, data workers have been shown to fudge data without actual conversations with affected people; efforts like jun sanwais (public hearings) and Janata Information Systems organized by the Mazdoor Kisan Shakti Sangatan are examples to secure representation through data.[97,193].”
Quote in the paperNot a substantial claim0 of 2 sources checked- cites [97] source
Not checked in the free run.
- cites [193]
Not checked in the free run.
- Premisecitation
NLP models disproportionately fail to detect names of people from non-Western backgrounds.
“[12] shows that NLP models disproportionately fail to even detect names of people from non-Western backgrounds.”
Quote in the paperNot a substantial claim0 of 1 sources checked- cites [12]
Not checked in the free run.
- Interpretationreasoning
Datasets derived from Internet connectivity will exclude a significant part of the Indian population.
“Datasets derived from Internet connectivity will exclude a significant population, e.g., many argued that India's mandatory covid-19 contact tracing app excluded hundreds of millions due to access constraints, pointing to the futility of digital nation-wide tracing (also see [112]).”
Quote in the paperSubstantial claim0 of 1 sources checked- cites [112]
Not checked in the free run.
- Interpretationcitation
Sub-groups such as caste, gender, and religion require different fairness implementations in India.
“Sub-groups like caste (endogamous, ranked social identities, assigned at birth [17,191]), 1 gender, and religion require different fairness implementations; but AI systems in India are under-analyzed for biases, mirroring the limited mainstream public discourse on oppression.”
Quote in the paperSubstantial claim0 of 2 sources checked- cites [17]
Not checked in the free run.
- cites [191]
Not checked in the free run.
- Interpretationreasoning
Barriers to responsibility in computer systems become more complicated when technology spans cultures.
“These barriers become more complicated when technology spans cultures: more, and more remote, hands are involved; intended behaviours may not be defined; computerblaming may meet computer-worship head on (see Section 4); and questions of ownership and liability become more complicated.”
Quote in the paperSubstantial claim - Premisecitation
There is limited political will in India to collect and share socio-economic indicators by caste or religion.
“Scholarship like Bond's caste report [184] argues that there is limited political will to collect and share socio-economic indicators by caste or religion.”
Quote in the paperSubstantial claim0 of 1 sources checked- cites [184]
Not checked in the free run.
- Interpretationreasoning
Localising model fairness alone can be window dressing in India, where the distance between models and oppressed communities is large.
“We contend that localising model fairness alone can be window dressing in India, where the distance between models and oppressed communities is large.”
Quote in the paperSubstantial claim - Interpretationreasoning
Conventional Fair-ML may be inappropriate, insufficient, or even inimical in India if it does not engage with local structures.
“In summary, we find that conventional Fair-ML may be inappropriate, insufficient, or even inimical in India if it does not engage with the local structures.”
Quote in the paperNot a substantial claim - Premisecitation
Open datasets such as ImageNet and Open Images have glaring gaps in geo-diversity.
“For instance, [190] highlights the glaring gaps in geo-diversity of open datasets such as ImageNet and Open Images that drive much of the computer vision research.”
Quote in the paperNot a substantial claim0 of 1 sources checked- cites [190] source
Not checked in the free run.
- Premisecitation
A 2019 global survey found Indians most likely to call AI exciting, futuristic, and mostly beneficial to society.
“In an international AI perceptions survey (2019), Indians ranked first in rating AI as 'exciting', 'futuristic' and 'mostly good for society' [110].”
Quote in the paperNot a substantial claim0 of 1 sources checked- cites [110]
Not checked in the free run.
- Premisecitation
Several studies indicate reservations successfully redistributed resources to oppressed sub-groups.
“Several studies have pointed to the successful redistribution of resources towards oppressed sub-groups, through reservations [43,68,152].”
Quote in the paperNot a substantial claim0 of 3 sources checked- cites [43]
Not checked in the free run.
- cites [68]
Not checked in the free run.
- cites [152]
Not checked in the free run.
- Interpretationown data
Indian justice traditions provide fresh, pertinent means to put algorithmic fairness into local practice.
“India's own justice approaches present new and relevant ways to operationalise algorithmic fairness locally.”
Quote in the paperSubstantial claim - Premisecitation
Equal opportunity and equal odds are fairness methods that come from US legal and epistemological traditions.
“Popular fairness techniques, such as equal opportunity and equal odds, stem from epistemological and legal systems of the US (e.g., [62,216]).”
Quote in the paperSubstantial claim0 of 2 sources checked- cites [62]
Not checked in the free run.
- cites [216]
Not checked in the free run.
- Interpretationown data
Inscrutability suppresses algorithmic fairness.
“Inscrutability suppresses algorithmic fairness.”
Quote in the paperSubstantial claim - Interpretationreasoning
Operationalising Western fairness approaches to Indian axes of historical injustice is often nontrivial.
“However, it is important to note that operationalising fairness approaches from the West to these axes is often nontrivial.”
Quote in the paperSubstantial claim - Premisecitation
A Newslaundry/Oxfam report described how privileged castes comprised Indian news media, invisibilising the oppressed majority in public discourse.
“A seminal report by Newslaundry/Oxfam described how privileged castes comprised Indian news media, invisibilising the vast oppressed majority in public discourse [145].”
Quote in the paperNot a substantial claim0 of 1 sources checked- cites [145]
Not checked in the free run.
- Interpretationreasoning
In the non-West, unequal standards, weak safeguards, and questionable AI uses can have catastrophic consequences.
“Similarly, unequal standards, inadequate safeguards, and dubious applications of AI in the non-West can lead to catastrophic effects (similar analogies have been made for content moderation [165,177]).”
Quote in the paperSubstantial claim0 of 2 sources checked- cites [165]
Not checked in the free run.
- cites [177] source
Not checked in the free run.
- Premisecitation
Indian undergraduate admissions run through joint allocation across many programmes and institutions, using exam scores, student preference rankings, and affirmative-action quotas.
“As an illustration, compared to the US undergraduate admission process of selection from a pool of candidates, the undergraduate admissions in India is done through various joint seat allocation processes, over hundreds of programmes, across dozens of universities and colleges that takes into account test scores, ordered preference lists provided by students, as well as various affirmative action”
Quote in the paperNot a substantial claim0 of 1 sources checked- cites [31] source
Not checked in the free run.
- Premisecitation
Earlier Indian disasters, such as the 1984 Union Carbide leak, show how faulty design and poor quality reflect double standards for the third world.
“Past disasters in India, like the fatal Union Carbide gas leak in 1984-one of the world's worst industrial accidents-point to faulty design and low quality, double standards for the 'third world' [206].”
Quote in the paperNot a substantial claim0 of 1 sources checked- cites [206] source
Not checked in the free run.
- Interpretationreasoning
Context-free assumptions in Fair-ML, whether in homegrown or international AI systems, can fail and produce harm inadvertently when applied to different infrastructural and cultural systems.
“Context-free assumptions in Fair-ML, whether in homegrown or international AI systems, can not just fail, but produce harm inadvertently when applied to different infrastructural and cultural systems.”
Quote in the paperSubstantial claim - Interpretationcitation
Even if models are not trained to identify sub-groups, proxy variables can still map disproportionately onto Dalits, Adivasis, and Muslims.
“While algorithms may not be trained on sub-group identification, proxies may correspond to Dalits, Adivasis, and Muslims disproportionately.”
Quote in the paperSubstantial claim - Interpretationown data
The idea that AI is neutral and free of humans gave its algorithmic authority credibility.
“AI's 'neutral' and 'human-free' associations lent credence to its algorithmic authority.”
Quote in the paperSubstantial claim - Premisecitation
In January 2020, facial recognition helped police arrest more than a thousand protesters in Delhi.
“In January 2020, over a thousand protestors were arrested during protests in Delhi, aided by FR.”
Quote in the paperNot a substantial claim0 of 1 sources checked- cites [5]
Not checked in the free run.
- May combine two claims. 56
- Classified as probably not a substantial claim. 51
May combine two claims.
FindingcitationAmong 138 FAccT papers from 2019 and 2020, only a few mention non-Western countries, and only one substantially engages a non-Western context.
“of the 138 papers published in 2019 and 2020, only a handful of papers even mention non-West countries, and only one of them-Marda's paper on New Delhi's predictive policing system [127]-substantially engages with a non-Western context.”
Quote in the paperSubstantial claim0 of 1 sources checked- cites [127]
Not checked in the free run.
Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based).
Findingnone givenTraditional occupations may correspond to caste, gender, or religion in India.
“• Occupation: traditional occupations may correspond to caste, gender, or religion; e.g., manual scavenging or butchery.”
Quote in the paperNot a substantial claimClassified as probably not a substantial claim. May combine two claims.
Findingown dataSafety apps populated by middle-class users tend to mark Dalit, Muslim, and slum areas as unsafe, potentially leading to hyper-patrolling in those areas.
“Many respondents were concerned that the safety apps were populated by middle-class users and tended to mark Dalit, Muslim, and slum areas as unsafe, potentially leading to hyper-patrolling in these areas.”
Quote in the paperSubstantial claimProbably not a substantial claim: may not pass S4 (based). May combine two claims.
Findingnone givenIndian users are perceived as bottom-billion data subjects and petri dishes for intrusive models, and are given poor recourse, effectively limiting their agency.
“Double standards and distance by ML makers: Indian users are perceived as 'bottom billion' data subjects, petri dishes for intrusive models, and given poor recourse-thus, effectively limiting their agency.”
Quote in the paperNot a substantial claimMay combine two claims.
FindingcitationSkin tone may indicate caste, ethnicity, and class in India, but its correspondence to Indian sub-groups is weaker than race-skin tone correlations, and dark skin tones can be discriminated against.
“• Skin tone: may indicate caste, ethnicity, and class; however, unlike correlations between race and skin tone, correspondences to Indian sub-groups is weaker.Dark skin tones can be discriminated against in India [60].”
Quote in the paperNot a substantial claim0 of 1 sources checked- cites [60]
Not checked in the free run.
Probably not a substantial claim: may not pass S4 (based). May combine two claims.
Findingnone givenAlthough Indians are part of the AI workforce, most work in services, and engineers do not fully represent marginalities, limiting remediation of distances.
“While Indians are part of the AI workforce, a majority work in services, and engineers do not entirely represent marginalities, limiting re-mediation of distances.”
Quote in the paperNot a substantial claimProbably not a substantial claim: may not pass S4 (based). May combine two claims.
Findingnone givenAI is readily adopted in high-stakes domains in India, often too early.
“AI is readily adopted in high-stakes domains, often too early.”
Quote in the paperNot a substantial claimMay combine two claims.
FindingcitationAI systems in India are under-analysed for biases, mirroring limited mainstream public discourse on oppression.
“Sub-groups like caste (endogamous, ranked social identities, assigned at birth [17,191]), 1 gender, and religion require different fairness implementations; but AI systems in India are under-analyzed for biases, mirroring the limited mainstream public discourse on oppression.”
- Also stated in: § Gender Identity & Sexual Orientation (No official · plain
“AI systems in India remain under-analysed for biases, mirroring the limited public discourse on oppression.”
Quote in the paperSubstantial claim0 of 2 sources checked- cites [17]
Not checked in the free run.
- cites [191]
Not checked in the free run.
- Also stated in: § Gender Identity & Sexual Orientation (No official · plain
Probably not a substantial claim: may not pass S2 (checkable), S4 (based).
Findingnone givenThe AI imaginary is aspirational in the Indian state, media, and legislation.
“The AI imaginary is aspirational in the Indian state, media, and legislation.”
Quote in the paperNot a substantial claimMay combine two claims.
FindingcitationZip codes in India can correspond to caste, religion, ethnicity, or class, and many Indian zip codes are heterogeneous, unlike rather homogeneous Western zip codes.
“• Zip codes: can correspond to caste, religion, ethnicity, or class. Many Indian zip codes are heterogeneous, with slums and posh houses abutting in the same area [20,36], unlike the rather homogeneous zip codes in the West from the segregated past [50].”
Quote in the paperNot a substantial claim0 of 3 sources checked- cites [20]
Not checked in the free run.
- cites [36]
Not checked in the free run.
- cites [50]
Not checked in the free run.
May combine two claims.
Findingown dataThe Indian state and industry collect and retain valuable large-scale data, but these datasets are not always publicly available because of infrastructure and non-transparency issues.
“Some respondents reported on how the state and industry apparati collected and retained valuable, large-scale data, but the datasets were not always made publicly available due to infrastructure and non-transparency issues.”
Quote in the paperSubstantial claimProbably not a substantial claim: may not pass S3 (load-bearing), S4 (based).
Findingnone givenExpenditure on dietary and lifestyle items may be proxies for religion, caste, or ethnicity in India.
“• Expenditure: on dietary and lifestyle items may be proxies for religion, caste, or ethnicity; e.g., expenses on pork or beef.”
Quote in the paperNot a substantial claimProbably not a substantial claim: may not pass S4 (based). May combine two claims.
Findingnone givenAI models in India are over-fitted for digitally-rich profiles, typically middle-class men, further excluding the half of the population without Internet access.
“Models are over-fitted for digitally-rich profiles-typically middle-class men-further excluding the 50% without Internet access.”
Quote in the paperNot a substantial claimClassified as probably not a substantial claim.
Findingown dataMany respondents observed a bias toward upper-middle-class problems, data, and deployments because of easier data access and revenue.
“Many respondents observed a bias towards upper-middle class problems, data, and deployments due to easier data access and revenue, as P8, CS/IS (computer science/information sciences researcher) put it, “rich people problems like cardiac disease and cancer, not poor people's Tuberculosis, prioritised in AI [in India]”, exacerbating inequities among those who benefit from AI and those who do not.”
Quote in the paperSubstantial claimProbably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.
Findingown dataDatasets under-collected darker skin tones, and measurement scales such as Fitzpatrick are not calibrated on diverse Indian skin tones.
“Many respondents described how datasets under-collected darker skin tones and measurement scales like Fitzpatrick scale are not calibrated on diverse Indian skin tones.”
Quote in the paperNot a substantial claimMay combine two claims.
FindingcitationNames are revelatory proxies for caste, religion, gender, and ethnicity in India and have contributed to discrimination.
“• Names: are revelatory proxies for caste, religion, gender, and ethnicity, and have contributed to discrimination in India [15,202], e.g., Banerjee or Khan connote to caste or religion location.”
Quote in the paperNot a substantial claim0 of 2 sources checked- cites [15]
Not checked in the free run.
- cites [202]
Not checked in the free run.
Probably not a substantial claim: may not pass S3 (load-bearing).
Findingown dataSome interviewees said AI surveillance tested on Dalit and Muslim bodies, such as efficiency trackers used on Dalit sanitation workers in Panchkula and Nagpur.
“Some respondents pointed to how Dalit and Muslim bodies were used as test subjects for AI surveillance, e.g., pointing to how human efficiency trackers were increasingly deployed among Dalit sanitation workers in cities like Panchkula and Nagpur.”
Quote in the paperNot a substantial claimMay combine two claims.
Findingown dataSeveral interviewees said enthusiasm for AI as socio-economic uplift came with strong trust in automation, little transparency, and no empowered Fair-ML ecosystem in India.
“Several respondents described how strong aspiration for AI for socio-economic upliftment was accompanied by high trust in automation, limited transparency, and the lack of an empowered Fair-ML ecosystem in India.”
Quote in the paperSubstantial claimMay combine two claims.
Findingown dataInterviewees saw no real recourse or redress for Indians, unlike the West's moderate response to user bias reports.
“In contrast to how AI industry and research were moderately responsive to user bias reports in the West, recourse and redress for Indians were perceived to be non-existent.”
Quote in the paperSubstantial claimMay combine two claims.
FindingcitationMany respondents said social audits and working with civil society improved understanding and accountability.
“Many described how social audits and working with civil society created a better understanding and accountability.8”
Quote in the paperNot a substantial claimMay combine two claims.
FindingcitationLending models imagined middle-class men as end-users, even though microfinance studies show women have high loan repayment rates.
“Many respondents described how lending models imagined middle-class men as end-users-even with many microfinance studies showing that women have high loan repayment rates [69,198].”
Quote in the paperSubstantial claim0 of 2 sources checked- cites [69]
Not checked in the free run.
- cites [198] source
Not checked in the free run.
May combine two claims.
Findingown dataSeveral respondents said automation solutions had fervent rhetoric while their accuracy and performance were low.
“Several respondents pointed to how automation solutions had fervent rhetoric; whereas in practice, accuracy and performance of systems were low.”
Quote in the paperNot a substantial claimProbably not a substantial claim: may not pass S3 (load-bearing).
Findingnone givenCritics said these systems gave dominant-caste supervisors more power.
“Such interventions were criticised for placing power in the hands of dominant-caste supervisors.”
Quote in the paperNot a substantial claimClassified as probably not a substantial claim. May combine two claims.
Findingown dataMany interviewees said Indians received poor designs, algorithms, and support, which they linked to weak technology policy and accountability enforcement in India.
“Many discussed how low quality designs, algorithms, and support were deployed for Indians, attributing to weak tech policy and enforcement of accountability in India.”
Quote in the paperSubstantial claimMay combine two claims.
FindingcitationLanguage can correspond to religion, class, ethnicity, and caste in India, and many AI systems serve in English, which only 10% of Indians understand.
“• Language: Language can correspond to religion, class, ethnicity, and caste.Many AI systems serve in English, which only 10% of Indians understand [174].”
Quote in the paperNot a substantial claim0 of 1 sources checked- cites [174] source
Not checked in the free run.
Classified as probably not a substantial claim.
FindingcitationMany respondents said the Indian technology sector claimed to be merit-based and open to anyone highly gifted in technical sciences.
“Many respondents described how the Indian technology sector claimed to be 'merit-based', open to anyone highly gifted in the technical sciences; but many have pointed to how merit is a function of caste privilege [196,207].”
Quote in the paperSubstantial claim0 of 2 sources checked- cites [196] source
Not checked in the free run.
- cites [207]
Not checked in the free run.
May combine two claims.
Findingown dataSeveral interviewees said most Indian AI engineers were men from privileged castes and classes.
“Several respondents discussed how Indian AI engineers were largely”
Quote in the paperSubstantial claimMay combine two claims.
FindingcitationSeveral AI applications require state-issued documentation such as Aadhaar or birth certificates, and the economically poor are also document-poor.
“• Documentation: Several AI applications require state-issued documentation like Aadhaar or birth certificates, e.g., in finance.The economically poor are also reported to be document-poor [173].”
Quote in the paperNot a substantial claim0 of 1 sources checked- cites [173]
Not checked in the free run.
May combine two claims.
FindingcitationMobility can correlate to gender and disability in India: women travel shorter distances and depend on public transport, and persons with disabilities often have lower mobility due to inaccessible infrastructure.
“• Mobility: can correlate to gender and disability.Indian women travel shorter distances and depend on public transport, for safety and cost [138].Persons with disabilities often have lower mobility in India, due to a lack of ramps and accessible infrastructure [188].”
Quote in the paperNot a substantial claim0 of 2 sources checked- cites [138]
Not checked in the free run.
- cites [188]
Not checked in the free run.
Classified as probably not a substantial claim. May combine two claims.
FindingcitationMany criticised neoliberal AI, including state systems, for treating Indians as peripheral 'bottom billion' data subjects: intrusive models, non-consensual automation, weak tech policy, poor user research, low-quality cheap or free products, and 'unsaturated markets'.
“Many criticised how neo-liberal AI (including those from the state) tended to treat Indians as 'bottom billion data subjects' in the periphery [211]-being subject to intrusive models, non-consensual automation, poor tech policies, inadequate user research, low-cost or free products that are low standard, and considered 'unsaturated markets'.”
Quote in the paperSubstantial claim0 of 1 sources checked- cites [211]
Not checked in the free run.
May combine two claims.
FindingcitationIn financial lending, women overwhelmingly availed and paid back loans in the names of male relatives to avoid perceived gender bias in credit scoring.
“In some cases, those with 'poor' data profiles subverted model predictions-as in P23's (STS researcher) research on financial lending, where women overwhelmingly availed and paid back loans in the names of male relatives to avoid perceived gender bias in credit scoring.”
Quote in the paperNot a substantial claimMay combine two claims.
Findingown dataInterviewees said feedback channels were often not adapted to India, and problems went unrecognised unless an activist reached company staff directly.
“Even when feedback mechanisms were included, respondents shared that they were not always localised for India, and incidents were not always recognised unless an activist contacted the company staff directly.”
Quote in the paperSubstantial claimMay combine two claims.
FindingcitationInternet access in India corresponds to gender, caste, and class, with 67% of Internet users being male.
“• Devices and infrastructure: Internet access corresponds to gender, caste, and class, with 67% Internet users being males [102].”
Quote in the paperNot a substantial claim0 of 1 sources checked- cites [102]
Not checked in the free run.
Classified as probably not a substantial claim.
Findingown dataLending apps in India instantly determine creditworthiness through alternative credit histories built from users' SMS messages, calls, and social networks.
“For instance, P11, tech policy researcher, illustrated how lending apps instantly determined creditworthiness through alternative credit histories built based on the user's SMS messages, calls, and social networks (due to limited credit or banking history).”
Quote in the paperSubstantial claimProbably not a substantial claim: may not pass S3 (load-bearing).
Findingown dataSome interviewees described a colonial desire for tight control over automation law, producing narrow, cautious views from the judiciary and state.
“Some talked about a colonial mindset of tight control in decision-making on automation laws, leading to reticent and monoscopic views by the judiciary and state.”
Quote in the paperNot a substantial claimClassified as probably not a substantial claim. May combine two claims.
Findingown dataP29 said AI framed as public good, like smart-city surveillance or women's safety, erased transgender people or intensified monitoring of Dalit and Muslim areas; an FR outside women's restrooms to catch non-female entrants could expose trans people to violence.
“e.g., P29 (LGBTQ+ activist) discussed how AI was justified in the public domain, e.g., surveillance for smart cities, 11 as women's safety measures, but tended to invisibilise transgender members or increase monitoring of Dalit and Muslim areas, e.g., a FR was deployed outside women's restrooms to detect intrusion by non-female entrants, potentially leading to violence against transgender members.”
Quote in the paperSubstantial claimClassified as probably not a substantial claim.
FindingcitationMany respondents expressed concern that AI advances in detecting sexuality, criminality, or terrorism could be exported to India and harm minorities.
“Many respondents expressed concern over AI advances in detecting sexuality, criminality, or terrorism (e.g., [187,212]) potentially being exported to India and harming minorities.”
Quote in the paperSubstantial claim0 of 2 sources checked- cites [187]
Not checked in the free run.
- cites [212] source
Not checked in the free run.
Probably not a substantial claim: may not pass S3 (load-bearing).
Findingown dataThe study's interviewees criticised neoliberal AI for continuing a wider extraction from 'bottom billion' data subjects and workers.
“Critiques were raised in our study on how neo-liberal AI followed a broader pattern of extraction from the 'bottom billion' data subjects and labourers.”
Quote in the paperNot a substantial claimMay combine two claims.
FindingcitationMany interviewees said overconfidence in technology plus weak policy engagement by decision-makers produced poor high-stakes systems, such as Delhi Police's FR service, reported to be very inaccurate and unable to tell boys from girls.
“Many described how disproportionate confidence in high-tech solutions, combined with limited technology policy engagement among decision-makers, appeared to lead to sub-par high-stakes solutions, e.g., the FR service used by Delhi Police was reported to have very low accuracy and failed to distinguish between boy and girl children [5]. 10”
Quote in the paperSubstantial claim0 of 1 sources checked- cites [5]
Not checked in the free run.
Probably not a substantial claim: may not pass S3 (load-bearing).
Findingown dataP15 said reporters can be criticised for challenging dominant narratives.
“P15 (journalist) pointed out that journalists may face disapproval for questioning certain narratives.”
Quote in the paperNot a substantial claimProbably not a substantial claim: may not pass S4 (based).
Findingnone givenThe study discusses how caste, religion, and tribe are eluded even within Indian technology discourse and policy.
“Our study discusses how caste, religion, and tribe are eluded even within the Indian technology discourse and policy.”
Quote in the paperNot a substantial claimMay combine two claims.
Findingown dataThe study found unreliable data from socio-economic factors, double standards in ML for Indian users, and unquestioning enthusiasm for AI.
“We found that data was not always reliable due to socioeconomic factors, ML products for Indian users sufffer from double standards, and AI was seen with unquestioning aspiration.”
Quote in the paperSubstantial claim- Rests on an assumption: The views and experiences reported by the interview respondents are representative enough of Indian AI practices to support general conclusions about India.
Probably not a substantial claim: may not pass S3 (load-bearing).
FindingcitationInterviewees noted major scope creep in high-stakes AI, like an FR tender that began with finding criminals, moved to missing children, and ended up arresting protesters.
“Additionally, respondents pointed to significant scope creep in high-stakes AI, e.g., a few mentioned how the tender for an FR system was initially motivated by detecting criminals, later missing children, to then arresting protestors [5].”
Quote in the paperNot a substantial claim0 of 1 sources checked- cites [5]
Not checked in the free run.
Probably not a substantial claim: may not pass S3 (load-bearing).
FindingcitationSome interviewees said certain automation projects, announced in response to public feeling, became surveillance, including Delhi predictive policing and railway-station FR after the 2012 Nirbhaya gangrape and women's safety incidents.
“Some respondents mentioned how a few automation solutions were announced following public sentiment, but turned into surveillance e.g., how predictive policing in Delhi and FR in train stations was announced after Nirbhaya's gruesome gangrape in 2012 and women's safety incidents [28,33].”
Quote in the paperNot a substantial claim0 of 2 sources checked- cites [28]
Not checked in the free run.
- cites [33]
Not checked in the free run.
Classified as probably not a substantial claim.
Findingown dataMany interviewees said scarce public debate and analysis of AI in India produced weak Fair-ML implementation.
“Many respondents described how limited debate and analysis of AI in India led to a weak implementation of”
Quote in the paperSubstantial claimMay combine two claims.
Findingown dataRespondents said fairness work in India lacked access to source datasets, APIs, and documentation, making high-stakes AI hard for researchers and civil society to assess.
“Fairness in India was reported to suffer from a lack of access to contributing datasets, APIs, and documentation, with several respondents describing how challenging it was for researchers and civil society to assess the high-stakes AI systems.”
Quote in the paperSubstantial claimMay combine two claims.
Findingown dataThe authors conducted a qualitative study and discourse analysis of algorithmic power in India, finding that fairness assumptions are challenged there.
“We presented a qualitative study and discourse analysis of algorithmic power in India, and found that algorithmic fairness assumptions are challenged in the Indian context.”
Quote in the paperSubstantial claim- Rests on an assumption: The views and experiences reported by the interview respondents are representative enough of Indian AI practices to support general conclusions about India.
Probably not a substantial claim: may not pass S3 (load-bearing).
PremisecitationNotions of algorithmic fairness fit within a broad arc of enlightenment and post-enlightenment thinking, including actuarial risk assessment.
“Indeed, notions of algorithmic fairness seem to fit within a broad arc of enlightenment and post-enlightenment thinking, including in actuarial risk assessment [147].”
Quote in the paperNot a substantial claim0 of 1 sources checked- cites [147] source
Not checked in the free run.
May combine two claims.
PremisecitationDistributive justice notions are not universal: children from a partially hunter-gatherer egalitarian African culture distributed spoils more equally than children from two other cultures, with merit playing only a limited role.
“Similarly, a study of children in three different cultures found that notions of distributive justice are not universal: “children from a partially hunter-gatherer, egalitarian African culture distributed the spoils more equally than did the other two cultures, with merit playing only a limited role” [185].”
Quote in the paperNot a substantial claim0 of 1 sources checked- cites [185] source
Not checked in the free run.
Classified as probably not a substantial claim.
PremisecitationMore collectivist and hierarchical Eastern societies differ from more individualistic and egalitarian Western cultures in how distributive, procedural, and retributive justice are conceptualised and achieved.
“Summarising decades worth of research, [120] says that the more collectivist and hierarchical societies in the East differs from the more individualistic and egalitarian cultures of the West in how different forms of justicedistributive, procedural, and retributive-are conceptualised and achieved.”
Quote in the paperSubstantial claim0 of 1 sources checked- cites [120] source
Not checked in the free run.
May combine two claims.
PremisecitationWestern AI fairness is becoming a universal ethical framework, while AI strategies in India, Tunisia, Mexico, and Uruguay espouse fairness but pay less attention to local fairness.
“Conventional western AI fairness is becoming a universal ethical framework for AI; consider the AI strategies from India [1], Tunisia [205], Mexico [129], and Uruguay [13] that espouse fairness and transparency, but pay less attention to what is fair in local contexts.”
Quote in the paperSubstantial claim0 of 3 sources checked- cites [1]
Not checked in the free run.
- cites [205]
Not checked in the free run.
- cites [129] source
Not checked in the free run.
Probably not a substantial claim: may not pass S3 (load-bearing).
PremisecitationResearchers since the late 1960s have attempted to translate US antidiscrimination law into statistical metrics.
“Indeed, researchers since the late 1960s have tried to translate US antidiscrimination law into statistical metrics [90].”
Quote in the paperNot a substantial claim0 of 1 sources checked- cites [90]
Not checked in the free run.
May combine two claims.
PremisecitationMost empirical work on social justice has been conducted in the US and Western Europe, grounding understandings of justice in the Western cultural context.
“While the concern for justice has a long history in the West (e.g., Aristotle, Rawls) and the East (e.g., Confucius, Chanakya), they show that the majority of empirical work on social justice has been situated in the US and Western Europe, grounding the understanding of justice in the Western cultural context.”
Quote in the paperSubstantial claim0 of 1 sources checked- cites [120] source
Not checked in the free run.
May combine two claims.
InterpretationreasoningWithout engaging non-Western conditions, values, politics, and histories, AI fairness can be tokenistic at best and harmful at worst for communities.
“Without engagement with the conditions, values, politics, and histories of the non-West, AI fairness can be a tokenism, at best-pernicious, at worst-for communities.”
Quote in the paperNot a substantial claimMay combine two claims.
PremisecitationChildren across studied cultures developed aversion to disadvantageous inequity, while advantageous inequity aversion was more prevalent in the West.
“For instance, [40] compared the acquisition of fairness behaviour in seven different societies: Canada, India, Mexico, Peru, Senegal, Uganda, and the USA, and found that while children from all cultures developed aversion towards disadvantageous inequity (avoid receiving less than a peer), advantageous inequity aversion (avoid receiving more than a peer) was more prevalent in the West.”
Quote in the paperNot a substantial claim0 of 1 sources checked- cites [40] source
Not checked in the free run.
May combine two claims.
PremisecitationScholars and activists have documented societal biases in Indian predictive policing, job applications, call-center callbacks, caste wage gaps, agricultural loan decisions, and online matrimonial ads.
“Researchers have documented societal biases in predictive policing in New Delhi [127], caste [202] and ethnicity [15] biases in job applications, call-center job callbacks [27], castebased wage-gaps [124], caste discrimination in agricultural loans decisions [116], and even in online matrimonial ads [157].”
Quote in the paperSubstantial claim0 of 3 sources checked- cites [127]
Not checked in the free run.
- cites [202]
Not checked in the free run.
- cites [15]
Not checked in the free run.
Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.
Study design and conductcitationThe study recruited experts through purposeful sampling, iteratively until saturation.
“We recruited respondents via a combination of reaching out directly and personal contacts, using purposeful sampling [151]-i.e., identifying and selecting experts with relevant experience-iterative until saturation.”
Quote in the paperNot a substantial claim0 of 1 sources checked- cites [151] source
Not checked in the free run.
Probably not a substantial claim: may not pass S2 (checkable), S3 (load-bearing).
PremisecitationAccountability is fundamentally about answerability for actions.
“Accountability is fundamentally about answerability for actions [114], and central to these are three phases by which an actor is made answerable to a forum: information-sharing, deliberation and discussion, and the imposition of consequences [214].”
Quote in the paperNot a substantial claim0 of 2 sources checked- cites [114] source
Not checked in the free run.
- cites [214]
Not checked in the free run.
Probably not a substantial claim: may not pass S4 (based).
InterpretationreasoningThe lack of an ecosystem of tools, policies, and stakeholders such as journalists, researchers, and activists to interrogate high-stakes AI inhibits meaningful fairness in India.
“Lack of an ecosystem of tools, policies, and stakeholders like journalists, researchers, and activists to interrogate high-stakes AI inhibits meaningful fairness in India.”
Quote in the paperNot a substantial claimProbably not a substantial claim: may not pass S3 (load-bearing).
PremisecitationRepresentational gaps in data are documented as a major challenge for responsible AI from a global perspective.
“Representational gaps in data is documented as one of the major challenges in achieving responsible AI from a global perspective [21,190].”
Quote in the paperNot a substantial claim0 of 2 sources checked- cites [21]
Not checked in the free run.
- cites [190] source
Not checked in the free run.
May combine two claims.
PremisecitationHalf of India's population lacks Internet access, and the excluded half is primarily women, rural communities, and Adivasis.
“Half the Indian population lacks access to the Internet-the excluded half is primarily women, rural communities, and Adivasis [96,106,153,168,189].”
Quote in the paperSubstantial claim0 of 5 sources checked- cites [96]
Not checked in the free run.
- cites [106]
Not checked in the free run.
- cites [153]
Not checked in the free run.
- cites [168]
Not checked in the free run.
- cites [189]
Not checked in the free run.
Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.
PremisecitationThe IEEE Standards Association highlighted the monopoly of Western ethical traditions in AI ethics and asked how Buddhist, Ubuntu, and Shinto-inspired traditions might change responsible AI.
“Recently, the IEEE Standards Association highlighted the monopoly of Western ethical traditions in AI ethics, and inquired how incorporating Buddhist, Ubuntu, and Shinto-inspired ethical traditions might change the processes of responsible AI [93].”
Quote in the paperNot a substantial claim0 of 1 sources checked- cites [93]
Not checked in the free run.
Probably not a substantial claim: may not pass S3 (load-bearing).
PremisecitationAn actor is made answerable to a forum through information-sharing, deliberation and discussion, and imposition of consequences.
“Accountability is fundamentally about answerability for actions [114], and central to these are three phases by which an actor is made answerable to a forum: information-sharing, deliberation and discussion, and the imposition of consequences [214].”
Quote in the paperNot a substantial claim0 of 1 sources checked- cites [214]
Not checked in the free run.
Probably not a substantial claim: may not pass S3 (load-bearing).
Study design and conductreasoningThe study used a research ethics protocol to protect respondent privacy and safety.
“We took great care to create a research ethics protocol to protect respondent privacy and safety, especially due to the sensitive nature of our inquiry.”
Quote in the paperNot a substantial claimMay combine two claims.
PremisecitationConventional gender roles in Indian society give men better access to devices, documentation, and mobility, so women often borrowed phones.
“conventional gender roles in Indian society lead to men having better access to devices, documentation, and mobility (see [64,178]), women often borrowed phones.”
Quote in the paperSubstantial claim0 of 2 sources checked- cites [64] source
Not checked in the free run.
- cites [178]
Not checked in the free run.
Probably not a substantial claim: may not pass S2 (checkable), S4 (based).
InterpretationreasoningWhere the distance between models and dis-empowered communities is large, a myopic focus on localising fair model outputs alone can backfire.
“In a societal context where the distance between models and dis-empowered communities is large-via technical distance, social distance, ethical distance, temporal distance, and physical distance-a myopic focus on localising fair model outputs alone can backfire.”
Quote in the paperNot a substantial claimProbably not a substantial claim: may not pass S3 (load-bearing).
PremisecitationNissenbaum distinguishes four barriers to responsibility in computer systems: many hands, bugs, blaming the computer, and ownership without liability.
“Nissenbaum distinguishes four barriers to responsibility in computer systems: (1) the problem of many hands, (2) bugs, (3) blaming the computer, and (4) ownership without liability [146].”
Quote in the paperNot a substantial claim0 of 1 sources checked- cites [146]
Not checked in the free run.
Classified as probably not a substantial claim.
Study design and conductcitationThe study coded and analysed transcripts for patterns using an inductive approach.
“Transcripts were coded and analyzed for patterns using an inductive approach [201].”
Quote in the paperSubstantial claim0 of 1 sources checked- cites [201]
Not checked in the free run.
May combine two claims.
PremisecitationMany, though not all, IIT graduates, founders of pioneering Indian software companies, and nearly all Nobel prize winners of Indian origin come from privileged castes and class backgrounds.
“Many, though not all, graduates of Indian Institutes of Technology, founders of pioneering Indian software companies and nearly all Nobel prize winners of Indian origin have come from privileged castes and class backgrounds [71,196].”
Quote in the paperNot a substantial claim0 of 2 sources checked- cites [71]
Not checked in the free run.
- cites [196] source
Not checked in the free run.
Probably not a substantial claim: may not pass S3 (load-bearing).
PremisecitationResearchers have pointed to technology's aspirational role in India, signifying modernity and progress via technocracy.
“Researchers have pointed to the aspirational role played by technology in India, signifying symbolic meanings of modernity and progress via technocracy [149,150,176].”
Quote in the paperNot a substantial claim0 of 3 sources checked- cites [149] source
Not checked in the free run.
- cites [150]
Not checked in the free run.
- cites [176]
Not checked in the free run.
Probably not a substantial claim: may not pass S3 (load-bearing).
PremisecitationReservations set aside places for marginalised groups at both national and state levels.
“Reservations assigns quotas for marginalised communities at the national and state levels.7”
Quote in the paperNot a substantial claim0 of 1 sources checked- cites [164] source
Not checked in the free run.
Probably not a substantial claim: may not pass S3 (load-bearing).
PremisecitationIndia's diversity of languages, scripts, and populace has been reported as attractive for improving model robustness and training data.
“reported to be attractive for improving model robustness and training data [14].”
Quote in the paperNot a substantial claim0 of 1 sources checked- cites [14]
Not checked in the free run.
Probably not a substantial claim: may not pass S3 (load-bearing).
PremisecitationMany have pointed out that merit in the Indian technology sector is a function of caste privilege.
“many have pointed to how merit is a function of caste privilege [196,207].”
Quote in the paperNot a substantial claim0 of 2 sources checked- cites [196] source
Not checked in the free run.
- cites [207]
Not checked in the free run.
May combine two claims.
PremisecitationUsing AI for social good is a central development priority in India, spanning health, farming, education, smart cities, and transport, and shaping public imaginaries of AI.
“AI for societal benefit is a pivotal development thrust in India, with a focus on healthcare, agriculture, education, smart cities, and mobility [1]-influencing citizen imaginaries of AI.”
Quote in the paperNot a substantial claim0 of 1 sources checked- cites [1]
Not checked in the free run.
Probably not a substantial claim: may not pass S3 (load-bearing).
Premisenone givenQuota shares under reservations vary by policy, from 30% to 80%.
“Depending on the policy, reservations can allocate quotas from 30% to 80%.”
Quote in the paperNot a substantial claimMay combine two claims.
PremisecitationThe trackers, equipped with microphones, GPS, cameras, and a SIM, enabled detailed surveillance of movement and work, causing some women workers to avoid restrooms, sensitive conversations, and to wait for the tracker to die before going home.
“Equipped with microphones, GPS, cameras, and a SIM, the trackers allowed detailed surveillance of movement and work, leading to some women workers avoiding restrooms for fear of camera capture, avoiding sensitive conversations for fear of snooping, and waiting for the tracker to die before going home [111].”
Quote in the paperNot a substantial claim0 of 1 sources checked- cites [111]
Not checked in the free run.
May combine two claims.
PremisecitationIndian film stars are mostly fair-skinned, conventionally attractive, and able-bodied, so they do not fully represent the larger Indian society.
“Indian film stars are overwhelmingly fair-skinned, conventionally attractive, and able-bodied [108], not fully representative of the larger society.”
Quote in the paperSubstantial claim0 of 1 sources checked- cites [108] source
Not checked in the free run.
May combine two claims.
PremisecitationThey began for Dalits and Adivasis and now cover other backward castes, women, disabled people, and religious minorities.
“Originally designed for Dalits and Adivasis, reservations have expanded to include other backward castes, women, persons with disabilities, and religious minorities.”
Quote in the paperNot a substantial claim0 of 1 sources checked- cites [164] source
Not checked in the free run.
May combine two claims.
Interpretationown dataEngineers and researchers are privileged worldwide, but the sharp socio-economic gap between Indian engineers and marginalised groups may widen the distance further.
“While engineers and researchers are mostly privileged everywhere, the stark socio-economic disparities between Indian engineers and the marginalised communities may further amplify the distances.”
Quote in the paperSubstantial claimMay combine two claims.
PremisecitationReservations are a restorative justice measure and a form of affirmative action enshrined in the Indian constitution.
“One of the restorative justice measures to repair centuries of subordination-reservations are a type of affirmative action enshrined in the Indian constitution [164].”
Quote in the paperNot a substantial claim0 of 1 sources checked- cites [164] source
Not checked in the free run.
Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based). May combine two claims.
Premisenone givenIn contrast to the West, where a large active stakeholder ecosystem is AI-literate and has access to open APIs and data.
“Contrast with the West, where a large, active stakeholder ecosystem (of civil society, journalists, and law makers) is AI-literate and has access to open APIs and data.”
Quote in the paperNot a substantial claimProbably not a substantial claim: may not pass S3 (load-bearing).
PremisecitationCommentators have called India's reservation system among the most radical policies globally.
“Reservations in India have been described as one of the world's most radical policies [25] (see [164] for more).”
Quote in the paperNot a substantial claim0 of 2 sources checkedProbably not a substantial claim: may not pass S2 (checkable), S3 (load-bearing), S4 (based). May combine two claims.
Interpretationnone givenAI is more aspirational and consequential in non-Western nations, where it performs an efficiency function in distributing scarce socio-economic resources and differentiating economies.
“AI is even more aspirational and consequential in non-Western nations, where it performs an 'efficiency' function in distributing scarce socio-economic resources and differentiating economies.”
Quote in the paperNot a substantial claimProbably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.
PremisecitationAI is increasingly perceived as a masculine, hype-filled, technoutopian enterprise, with nations turning into AI superpowers.
“AI is increasingly perceived as a masculine, hype-filled, technoutopian enterprise, with nations turning into AI superpowers [170].”
Quote in the paperNot a substantial claim0 of 1 sources checked- cites [170]
Not checked in the free run.
Classified as probably not a substantial claim.
Interpretationnone givenThe considerations identified are not limited to India.
“The considerations we identified are certainly not limited to India; likewise, we call for inclusively evolving global approaches to Fair-ML.”
Quote in the paperSubstantial claimProbably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.
PremisecitationCheap labour, big diverse populations, and weak policy have been cited as reasons for double standards in India, including non-consensual human trials and waste dumping.
“Low costs, large and diverse populations, and policy infirmities have been cited as reasons for following double standards in India, e.g., in non-consensual human trials and waste dumping [123] (also see [137]).”
Quote in the paperNot a substantial claim0 of 2 sources checked- cites [123]
Not checked in the free run.
- cites [137]
Not checked in the free run.
Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based). May combine two claims.
Premisenone givenIn India, names signal many socio-demographic traits, yet no large, easily available Indian-name datasets exist for fairness testing.
“For instance, personal names act as a signifier for various socio-demographic attributes in India, however there are no large datasets of Indian names (like the US Census data, or the SSA data) that are readily available for fairness evaluations.”
Quote in the paperNot a substantial claimMay combine two claims.
PremisecitationAccording to NCRB 2015, 34% of undertrials were Dalits and Adivasis (25% of population); 20% were Muslims (14% of population); and 70% were non-literate (26% of population).
“e.g., according to the National Crime Records Bureau (NCRB) in 2015, 34% of undertrials were Dalits and Adivasis (25% of the population); 20% were Muslims (14% of population); and 70% were non-literate (26% of the population) [203].”
Quote in the paperNot a substantial claim0 of 1 sources checked- cites [203]
Not checked in the free run.
Probably not a substantial claim: may not pass S2 (checkable).
InterpretationreasoningRecontextualising data and models has limits unless communities are empowered to help define problems, set fairness expectations, and design systems.
“Recontextualising data and models can only go so far without participatorily empowering communities in identifying problems, specifying fairness expectations, and designing systems.”
Quote in the paperNot a substantial claimMay combine two claims.
PremisecitationThe quota system gives rise to matching under distributional constraints, a known economics problem that has not received attention within the FAccT community.
“The quota system gives rise to the problem of matching under distributional constraints, a known problem in economics [23,79,105], but has not received attention within the FAccT community (although [51] is related).”
Quote in the paperNot a substantial claim0 of 3 sources checked- cites [23] source
Not checked in the free run.
- cites [79]
Not checked in the free run.
- cites [105]
Not checked in the free run.
May combine two claims.
InterpretationreasoningWhen data and models omit Indic factors and values, and ML has double standards and disjuncture amid unquestioning AI enthusiasm, there is a risk of entrenching injustice and harm.
“Missing Indic factors and values in data and models, compounded by double standards and disjuncture in ML, deployed in an environment of unquestioning AI aspiration, face the risk of reinforcing injustices and harms.”
Quote in the paperNot a substantial claim
The paper relies on this without support.
There is a lack of substantial policy or research on advancing algorithmic fairness for India's large AI-interfacing population.
“Despite the momentum on high-stakes AI systems, currently there is a lack of substantial policy or research on advancing algorithmic fairness for such a large population interfacing with AI.”
The paper relies on this without support.
India is on a unique path to AI characterised by pluralism, socio-economic development, technocratic nationbuilding, and uneven AI capital.
“We contend that India is on a unique path to AI, characterised by pluralism, socio-economic development, technocratic nationbuilding, and uneven AI capital-which requires us to confront many assumptions made in algorithmic fairness.”
The paper relies on this without support.
The views and experiences reported by the interview respondents are representative enough of Indian AI practices to support general conclusions about India.
“We presented a qualitative study and discourse analysis of algorithmic power in India, and found that algorithmic fairness assumptions are challenged in the Indian context.”
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
Fairness is often contextual: a seemingly 'fair' model designed for one context might be misleading, inaccurate, or harmful when applied to a different context.
“However, as existing literature makes clear, fairness is often contextual [57,90,91,131,136]: a seemingly”
From Out of Context: Investigating the Bias and Fairness Concerns of “Artificial Intelligence as a Service” (2023) · cites this paper as [131]
Transparency which supports broader accountability processes on training sets, models and internal processes behind these services, as well as acknowledgement of known limitations, past failures, and lessons learnt, can help move AIaaS from opaque, unknown and potentially harmful 'black-boxes' to more understandable, accepting of their own limitations, building blocks for affordable and widely acc
“Transparency which supports broader accountability processes on training sets, models and internal processes behind these services, as well as in the widersense of acknowledgement of known-limitations, past failures, and lessons learnt [27,131] can help move AIaaS from the opaque, unknown and potentially harmful 'black-boxes' of today, to a more understandable, accepting of their own limitations,”
From Out of Context: Investigating the Bias and Fairness Concerns of “Artificial Intelligence as a Service” (2023) · cites this paper as [131]
Transparency can assist in facilitating algorithmic fairness.
“As a part of the internal and external accountability processes, transparency can assist in facilitating algorithmic fairness.”
From Out of Context: Investigating the Bias and Fairness Concerns of “Artificial Intelligence as a Service” (2023) · cites this paper as [131]
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
Each contradiction between two passages listed here was confirmed by an independent check.
SEWA citation points to an unrelated reference
MinorThe text cites reference [10] for SEWA, but reference 10 is an article about cow-related violence. SEWA appears as reference 9.
“Initiatives like Design Beku [4] and SEWA [10] are excellent decolonial examples of participatorily co-designing with under-served communities.”
“10. Delna Abraham and Ojaswi Rao. [n.d.]. 84% Dead In Cow-Related Vi- olence Since 2010 Are Muslim; 97% Attacks After 2014 | IndiaSpend.”
Cites
A General Inductive Approach for Analyzing Qualitative Evaluation Data
Thomas, D. R. (2006) · American Journal of Evaluation · doi:10.1177/1098214005283748 · Cited by 10,858 · cited 1× · see citing works
“Analysis and coding Transcripts were coded and analyzed for patterns using an inductive approach [201].”
MethodFanon, F. (2000) · Princeton University Press eBooks · doi:10.1007/978-1-137-05194-3_4 · Cited by 7,509 · see citing works
The global landscape of AI ethics guidelines
Jobin, A., Ienca, M., & Vayena, E. (2019) · Nature Machine Intelligence · doi:10.1038/s42256-019-0088-2 · Cited by 6,220 · cited 1× · see citing works
“The call for a global lens in AI accountability is not new [84, 155], but the ethical principles in AI are often interpreted, prioritised, contextualised, and implemented differently across the globe [98].”
A Survey on Bias and Fairness in Machine Learning
Mehrabi, N., Morstatter, F., Saxena, N., Lerman, K., & Galstyan, A. (2022) · ACM Computing Surveys · doi:10.1145/3457607 · Cited by 5,812 · see citing works
Encountering development: the making and unmaking of the Third World
(1995) · Choice Reviews Online · doi:10.5860/choice.32-5764 · Cited by 5,445 · see citing works
The validity and practicality of sun-reactive skin types I through VI
Fitzpatrick, T. B. (1988) · Archives of Dermatology · doi:10.1001/archderm.1988.01670060015008 · Cited by 4,339 · cited 4× · see citing works
“often used by researchers as a phenotype [46], but was originally developed to categorise UV sensitivity) [73].”
BackgroundInfluential(2018) · Cited by 3,510 · see citing works
Decolonising the Mind: The Politics of Language in African Literature
Kamoche, J. G., & Thiong’o, N. W. (1987) · World Literature Today · doi:10.2307/40143257 · Cited by 3,087 · cited 1× · see citing works
“A decolonial perspective (borrowed from Latin American and African scholars like [66, 70, 72, 134, 210]) helps us confront inequalities from colonisation in India, providing new openings for knowledge and justice in AI fairness research.”
BackgroundBarocas, S., & Selbst, A. D. (2014) · doi:10.2139/ssrn.2477899 · Cited by 3,048 · see citing works
Model Cards for Model Reporting
Mitchell, M., Wu, S., Zaldivar, A., Barnes, P., Vasserman, L., Hutchinson, B., Spitzer, E., Raji, I. D., & Gebru, T. (2019) · Proceedings of the Conference on Fairness, Accountability, and Transparency · doi:10.1145/3287560.3287596 · Cited by 2,681 · see citing works
Self-citationGebru, T., Morgenstern, J., Vecchione, B., Vaughan, J. W., Wallach, H., Daumé, H., & Crawford, K. (2018) · arXiv · doi:10.48550/arxiv.1803.09010 · Cited by 2,474
Gebru, T., Morgenstern, J., Vecchione, B., Vaughan, J. W., Wallach, H., III, H. D., & Crawford, K. (2021) · Communications of the ACM · doi:10.1145/3458723 · Cited by 2,469 · see citing works
Feminism without Borders: Decolonizing Theory, Practicing Solidarity
Mohanty, C. T. (2003) · Project Muse (Johns Hopkins University) · Cited by 2,374 · see citing works
Adversarial machine learning at scale
Kurakin, A., Goodfellow, I., & Bengio, S. (2016) · arXiv · doi:10.48550/arxiv.1611.01236 · Cited by 2,325 · cited 1× · see citing works
“perturbation testing [156], data augmentation [220], adversarial testing [117], and adherence to terminology guidelines for oppressed groups, such as SIGACCESS.”
Background(2011) · Cited by 2,277 · see citing works
Man is to Computer Programmer as Woman is to Homemaker? Debiasing Word Embeddings
Bolukbasi, T., Chang, K.-W., Zou, J., Saligrama, V., & Kalai, A. (2016) · arXiv · doi:10.48550/arxiv.1607.06520 · Cited by 1,615 · cited 1× · see citing works
“, [42, 218]), alongside testing strategies, e.”
Mitigating Unwanted Biases with Adversarial Learning
Zhang, B. H., Lemoine, B., & Mitchell, M. (2018) · Proceedings of the 2018 AAAI/ACM Conference on AI, Ethics, and Society · doi:10.1145/3278721.3278779 · Cited by 1,387 · cited 1× · see citing works
“, [42, 218]), alongside testing strategies, e.”
Normal Life: Administrative Violence, Critical Trans Politics, and the Limits of Law
Spade, D. (2011) · Project Muse (Johns Hopkins University) · Cited by 1,049 · see citing works
Decolonial AI: Decolonial Theory as Sociotechnical Foresight in Artificial Intelligence
Mohamed, S., Png, M.-T., & Isaac, W. (2020) · Philosophy & Technology · doi:10.1007/s13347-020-00405-8 · Cited by 912 · cited 2× · see citing works
“, in non-consensual human trials and waste dumping [123] (also see [137]).”
BackgroundVirginia Held,The ethics of care: Personal, political, and global(Oxford University Press, 2006)
Mahowald, M. B. (2009) · International Journal of Feminist Approaches to Bioethics · doi:10.2979/fab.2009.2.1.177 · Cited by 909 · see citing works
Bender, E. M., & Friedman, B. (2018) · Transactions of the Association for Computational Linguistics · doi:10.1162/tacl_a_00041 · Cited by 854 · cited 1× · see citing works
“, [34, 77, 136]), openly discussing limitations, failure cases, and lessons learnt can help move from the ‘magic pill’ role of fairness as a checklist for ethical issues in India—to a more pragmatic, flawed, and evolving scientific function.”
BackgroundMen Also Like Shopping: Reducing Gender Bias Amplification using Corpus-level Constraints
Zhao, J., Wang, T., Yatskar, M., Ordonez, V., & Chang, K.-W. (2017) · Proceedings of the 2017 Conference on Empirical Methods in Natural Language Processing · doi:10.18653/v1/d17-1323 · Cited by 829 · see citing works
The Risk of Racial Bias in Hate Speech Detection
Sap, M., Card, D., Gabriel, S., Choi, Y., & Smith, N. A. (2019) · Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics · doi:10.18653/v1/p19-1163 · Cited by 817 · cited 1× · see citing works
“The majority of fairness research looks at racial [46, 57, 121, 125, 183] and gender biases [42, 46, 197,”
Background“Everyone wants to do the model work, not the data work”: Data Cascades in High-Stakes AI
Sambasivan, N., Kapania, S., Highfill, H., Akrong, D., Paritosh, P., & Aroyo, L. M. (2021) · Proceedings of the 2021 CHI Conference on Human Factors in Computing Systems · doi:10.1145/3411764.3445518 · Cited by 781 · see citing works
Self-citationDigital colonialism: US empire and the new imperialism in the Global South
Kwet, M. (2019) · Race & Class · doi:10.1177/0306396818823172 · Cited by 779 · cited 1× · see citing works
“Scholars have pointed to how neoliberal AI follows the technical architecture of classic colonialism through data extraction, impairing indigenous innovation, and shipping manufactured services back to the data subjects—among communities already prone to exploitation, under-development, and inequality from centuries o…”
BackgroundImproving Fairness in Machine Learning Systems: What Do Industry Practitioners Need?
Holstein, K., Wortman Vaughan, J., Daumé, H., Dudik, M., & Wallach, H. (2019) · Proceedings of the 2019 CHI Conference on Human Factors in Computing Systems · doi:10.1145/3290605.3300830 · Cited by 777 · cited 1× · see citing works
“Biases in models are frequently attributed to biased datasets, presupposing the possibility of achieving fairness by “fixing” the data [89].”
BackgroundWang, Y., & Kosinski, M. (2018) · Journal of Personality and Social Psychology · doi:10.1037/pspa0000098 · Cited by 766 · cited 1× · see citing works
“, [187, 212]) potentially being exported to India and harming minorities.”
BackgroundSocial Data: Biases, Methodological Pitfalls, and Ethical Boundaries
Olteanu, A., Castillo, C., Diaz, F., & Kıcıman, E. (2019) · Frontiers in Big Data · doi:10.3389/fdata.2019.00013 · Cited by 763 · cited 1× · see citing works
“5This issue is somewhat related to what [148] call “Non-individual agents”.”
BackgroundRed tape: bureaucracy, structural violence, and poverty in India
Gupta, A. (2013) · Choice Reviews Online · doi:10.5860/choice.50-3376 · Cited by 676 · see citing works
Fairness in Machine Learning: Lessons from Political Philosophy
Binns, R. (2017) · arXiv · doi:10.48550/arxiv.1712.03586 · Cited by 676 · cited 1× · see citing works
“Connections have been made between algorithmic fairness and Western concepts such as egalitarianism [38], consequentialism [142, 167], deontic justice [38, 142], and Rawls’ distributive justice [99, 142].”
BackgroundPande, R. (2003) · American Economic Review · doi:10.1257/000282803769206232 · Cited by 658 · cited 1× · see citing works
“studies have pointed to the successful redistribution of resources towards oppressed sub-groups, through reservations [43, 68, 152].”
BackgroundLum, K., & Isaac, W. (2016) · Significance · doi:10.1111/j.1740-9713.2016.00960.x · Cited by 611 · see citing works
Spectral Housing and Urban Cleansing: Notes on Millennial Mumbai
Appadurai, A. (2000) · Public Culture · doi:10.1215/08992363-12-3-627 · Cited by 586 · cited 1× · see citing works
“Many Indian zip codes are heterogeneous, with slums and posh houses abutting in the same area [20, 36], unlike the rather homogeneous zip codes in the West from the segregated past [50].”
BackgroundThe ontogeny of fairness in seven societies
Blake, P. R., McAuliffe, K., Corbit, J., Callaghan, T. C., Barry, O., Bowie, A., Kleutsch, L., Kramer, K. L., Ross, E., Vongsachang, H., et al. (2015) · Nature · doi:10.1038/nature15703 · Cited by 534 · cited 1× · see citing works
“For instance, [40] compared the acquisition of fairness behaviour in seven different societies: Canada, India, Mexico, Peru, Senegal, Uganda, and the USA, and found that while children from all cultures developed aversion towards disadvantageous inequity (avoid receiving less than a peer), advantageous inequity aversi…”
BackgroundWomen and Repayment in Microfinance: A Global Analysis
D’Espallier, B., Guérin, I., & Mersland, R. (2010) · World Development · doi:10.1016/j.worlddev.2010.10.008 · Cited by 498 · cited 1× · see citing works
“models imagined middle-class men as end-users—even with many microfinance studies showing that women have high loan repayment rates [69, 198].”
BackgroundDoes microfinance empower women? Evidence from self‐help groups in India
Swain, R. B., & Wallentin, F. Y. (2009) · International Review of Applied Economics · doi:10.1080/02692170903007540 · Cited by 435 · cited 1× · see citing works
“models imagined middle-class men as end-users—even with many microfinance studies showing that women have high loan repayment rates [69, 198].”
BackgroundRudin, C., & Radin, J. (2019) · Harvard Data Science Review · doi:10.1162/99608f92.5a8a3a3d · Cited by 432 · cited 1× · see citing works
“, even the humans who design models do not always understand how variables are being combined to make inferences [171], many respondents discussed an end-to-end opacity of inputs, model behaviour, and outcomes in India.”
BackgroundRacial Bias in Hate Speech and Abusive Language Detection Datasets
Davidson, T., Bhattacharya, D., & Weber, I. (2019) · Proceedings of the Third Workshop on Abusive Language Online · doi:10.18653/v1/w19-3504 · Cited by 414 · see citing works
Scaling of End-To-End Governance Risk Assessments for AI Systems (Practitioner Track)
Weimer, D., Gensch, A., & Koller, K. (2025) · OASIcs : OpenAccess Series in Informatics · doi:10.4230/oasics.saia.2024.4 · Cited by 398 · see citing works
The case for technology in developing regions
Brewer, E., Demmer, M., Du, B., Ho, M., Kam, M., Nedevschi, S., Pal, J., Patra, R. K., Surana, S., & Fall, K. (2005) · Computer · doi:10.1109/mc.2005.204 · Cited by 384 · see citing works
Algorithmic Colonization of Africa
Birhane, A. (2020) · SCRIPTed: A Journal of Law, Technology & Society · doi:10.2966/scrip.170220.389 · Cited by 349 · cited 1× · see citing works
“Scholars have pointed to how neoliberal AI follows the technical architecture of classic colonialism through data extraction, impairing indigenous innovation, and shipping manufactured services back to the data subjects—among communities already prone to exploitation, under-development, and inequality from centuries o…”
BackgroundAccountability in a computerized society
Nissenbaum, H. (1996) · Science and Engineering Ethics · doi:10.1007/bf02639315 · Cited by 346 · cited 1× · see citing works
“Nissenbaum distinguishes four barriers to responsibility in computer systems: (1) the problem of many hands, (2) bugs, (3) blaming the computer, and (4) ownership without liability [146].”
BackgroundAccounts and accountability: Theoretical implications of the right-to-information movement in India
Jenkins, R., & Goetz, A. M. (1999) · Third World Quarterly · doi:10.1080/01436599913712 · Cited by 345 · see citing works
Geek Heresy: Rescuing Social Change from the Cult of Technology
Toyama, K. (2017) · doi:10.13021/g8itlcp.9.2017.1853 · Cited by 333 · see citing works
Wieringa, M. (2020) · Proceedings of the 2020 Conference on Fairness, Accountability, and Transparency · doi:10.1145/3351095.3372833 · Cited by 321 · cited 1× · see citing works
“made answerable to a forum: information-sharing, deliberation and discussion, and the imposition of consequences [214].”
BackgroundOneto, L., & Chiappa, S. (2020) · Studies in computational intelligence · doi:10.1007/978-3-030-43883-8_7 · Cited by 315 · see citing works
50 Years of Test (Un)fairness: Lessons for Machine Learning
Hutchinson, B., & Mitchell, M. (2019) · Proceedings of the Conference on Fairness, Accountability, and Transparency · doi:10.1145/3287560.3287600 · Cited by 297 · see citing works
Self-citationTowards a critical race methodology in algorithmic fairness
Hanna, A., Denton, R., Smart, A., & Smith-Loud, J. (2020) · Proceedings of the 2020 Conference on Fairness, Accountability, and Transparency · doi:10.1145/3351095.3372826 · Cited by 295 · cited 2× · see citing works
“However these categories are culturally and historically situated [85].”
BackgroundManzini, T., Chong, L. Y., Black, A. W., & Tsvetkov, Y. (2019) · Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies (NAACL-HLT) · doi:10.18653/v1/n19-1062 · Cited by 284 · see citing works
Social Biases in NLP Models as Barriers for Persons with Disabilities
Hutchinson, B., Prabhakaran, V., Denton, E., Webster, K., Zhong, Y., & Denuyl, S. (2020) · Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics · doi:10.18653/v1/2020.acl-main.487 · Cited by 274 · cited 1× · see citing works
“While other axes of discrimination and injustices such as disability status [91], age [59], and sexual orientation [76] have gotten some attention, biases relevant to other geographies and cultures are not explored (e.”
BackgroundSelf-citationMaking Merit: The Indian Institutes of Technology and the Social Life of Caste
Subramanian, A. (2015) · Comparative Studies in Society and History · doi:10.1017/s0010417515000043 · Cited by 242 · cited 2× · see citing works
“winners of Indian origin have come from privileged castes and class backgrounds [71, 196].”
BackgroundText-Free User Interfaces for Illiterate and Semiliterate Users
Medhi, I., Sagar, A., & Toyama, K. (2007) · Information Technologies and International Development · doi:10.1162/itid.2007.4.1.37 · Cited by 241 · see citing works
Counterfactual Data Augmentation for Mitigating Gender Stereotypes in Languages with Rich Morphology
Zmigrod, R., Mielke, S. J., Wallach, H., & Cotterell, R. (2019) · Proceedings of the conference - Association for Computational Linguistics. Meeting · doi:10.18653/v1/p19-1161 · Cited by 241 · cited 1× · see citing works
“perturbation testing [156], data augmentation [220], adversarial testing [117], and adherence to terminology guidelines for oppressed groups, such as SIGACCESS.”
BackgroundEfficient Matching under Distributional Constraints: Theory and Applications
Kamada, Y., & Kojima, F. (2015) · American Economic Review · doi:10.1257/aer.20101552 · Cited by 239 · cited 1× · see citing works
“The quota system gives rise to the problem of matching under distributional constraints, a known problem in economics [23, 79, 105], but has not received attention within the FAccT community (although [51] is related).”
BackgroundIntermediated technology use in developing communities
Sambasivan, N., Cutrell, E., Toyama, K., & Nardi, B. (2010) · Proceedings of the SIGCHI Conference on Human Factors in Computing Systems · doi:10.1145/1753326.1753718 · Cited by 236 · cited 1× · see citing works
“, via delay-tolerant connectivity, low cost devices, text-free interfaces, intermediaries, and NGO partnerships (see [45, 74, 86, 115, 132, 179, 209]).”
BackgroundSelf-citationGovernment of India Ministry of Home Affairs
Commissioner, I. R. G. C. (2016) · Census Library, India · Cited by 235 · see citing works
Digital Green: Participatory video for agricultural extension
Gandhi, R., Veeraraghavan, R., Toyama, K., & Ramprasad, V. (2007) · 2007 International Conference on Information and Communication Technologies and Development · doi:10.1109/ictd.2007.4937388 · Cited by 221 · cited 1× · see citing works
“, via delay-tolerant connectivity, low cost devices, text-free interfaces, intermediaries, and NGO partnerships (see [45, 74, 86, 115, 132, 179, 209]).”
BackgroundSchäfer, M., Haun, D. B. M., & Tomasello, M. (2015) · Psychological Science · doi:10.1177/0956797615586188 · Cited by 215 · cited 1× · see citing works
“Similarly, a study of children in three different cultures found that notions of distributive justice are not universal: “children from a partially hunter-gatherer, egalitarian African culture distributed the spoils more equally than did the other two cultures, with merit playing only a limited role” [185].”
BackgroundRoy, A. (2014) · Medical Entomology and Zoology · Cited by 210 · see citing works
Castes in India: Their Mechanism, Genesis and Development
Ambedkar, B. R. (2015) · Medical Entomology and Zoology · Cited by 204 · see citing works
Labor market discrimination in Delhi: Evidence from a field experiment
Banerjee, A., Bertrand, M., Datta, S., & Mullainathan, S. (2008) · Journal of Comparative Economics · doi:10.1016/j.jce.2008.09.002 · Cited by 200 · cited 1× · see citing works
“predictive policing in New Delhi [127], caste [202] and ethnicity [15] biases in job applications, call-center job callbacks [27], castebased wage-gaps [124], caste discrimination in agricultural loans decisions [116], and even in online matrimonial ads [157].”
BackgroundCounterfactual Fairness in Text Classification through Robustness
Garg, S., Perot, V., Limtiaco, N., Taly, A., Ed, H., & Beutel, A. (2019) · Proceedings of the AAAI/ACM Conference on AI Ethics and Society · doi:10.1145/3306618.3317950 · Cited by 200 · cited 1× · see citing works
“While other axes of discrimination and injustices such as disability status [91], age [59], and sexual orientation [76] have gotten some attention, biases relevant to other geographies and cultures are not explored (e.”
BackgroundShankar, S., Halpern, Y., Breck, E. J., Atwood, J. R., Wilson, J., & Sculley, D. (2017) · arXiv (Cornell University) · doi:10.48550/arxiv.1711.08536 · Cited by 175 · cited 2× · see citing works
“For instance, [190] highlights the glaring gaps in geo-diversity of open datasets such as ImageNet and Open Images that drive much of the computer”
Background"They Don't Leave Us Alone Anywhere We Go": Gender and Digital Abuse in South Asia
Sambasivan, N., Batool, A., Ahmed, N., Matthews, T., Thomas, K., Gaytán-Lugo, L. S., Nemer, D., Bursztein, E., Churchill, E., & Consolvo, S. (2019) · Proceedings of the 2019 CHI Conference on Human Factors in Computing Systems · doi:10.1145/3290605.3300232 · Cited by 175 · see citing works
Self-citationAnnihilation of Caste: The Annotated Critical Edition
Ambedkar, B. R., Anand, S. S., & Roy, A. (2014) · Medical Entomology and Zoology · Cited by 168 · see citing works
DOUBLE STANDARDS IN MEDICAL RESEARCH IN DEVELOPING COUNTRIES
Smith, P. W. F. (2004) · SCRIPTed A Journal of Law Technology & Society · doi:10.2966/scrip.010404.639 · Cited by 161 · see citing works
Marda, V. (2018) · Philosophical Transactions of the Royal Society A: Mathematical, Physical and Engineering Sciences · doi:10.1098/rsta.2018.0087 · Cited by 161 · cited 11× · see citing works
“opportunities for AI in India [104], proposed policy deliberation frameworks that take into account the unique policy landscape of India [126], and questioned the intrusive data collection practices through Aadhaar (biometric-based unique identity for Indians) in India [158, 159].”
BackgroundInfluentialAlgorithmic realism: expanding the boundaries of algorithmic thought
Green, B., & Viljoen, S. (2020) · Proceedings of the 2020 Conference on Fairness, Accountability, and Transparency · doi:10.1145/3351095.3372840 · Cited by 161 · cited 1× · see citing works
“The community also often repurposes terminology from US legal domains, such as “disparate impact”, “disparate treatment”, and “equal opportunity”, or use them as points of triangulation, in order to compare technical properties of fairness through analogy with legal concepts [81, 82].”
BackgroundToward situated interventions for algorithmic equity: lessons from the field
Katell, M., Young, M., Dailey, D., Herman, B., Guetler, V., Tam, A., Bintz, C., Raz, D., & Krafft, P. M. (2020) · Proceedings of the 2020 Conference on Fairness, Accountability, and Transparency · doi:10.1145/3351095.3372874 · Cited by 153 · cited 1× · see citing works
“Since outcomes of ML deployments can be difficult to predict, proposals for accountability include participatory design [109] and participatory problem formulation [128], sharing the responsibility for design-”
BackgroundDuflo, E. (2005) · Journal of the European Economic Association · doi:10.1162/jeea.2005.3.2-3.668 · Cited by 151 · cited 1× · see citing works
“studies have pointed to the successful redistribution of resources towards oppressed sub-groups, through reservations [43, 68, 152].”
BackgroundMobile Phones for Maternal Health in Rural India
Kumar, N., & Anderson, R. J. (2015) · Proceedings of the 33rd Annual ACM Conference on Human Factors in Computing Systems · doi:10.1145/2702123.2702258 · Cited by 146 · cited 1× · see citing works
“, via delay-tolerant connectivity, low cost devices, text-free interfaces, intermediaries, and NGO partnerships (see [45, 74, 86, 115, 132, 179, 209]).”
BackgroundAddressing age-related bias in sentiment analysis,
Diaz, M., Johnson, I., Lazar, A., Piper, A. M., & Gergle, D. (2018) · Proceedings of the 2018 CHI Conference on Human Factors in Computing Systems · doi:10.1145/3173574.3173986 · Cited by 138 · see citing works
After Access: Inclusion, Development, and a More Mobile Internet
Donner, J. (2015) · Cited by 136 · see citing works
Men Also Like Shopping: Reducing Gender Bias Amplification using Corpus-level ConstraintsPreprint
Zhao, J., Wang, T., Yatskar, M., et al. (2017) · arXiv (Cornell University) · doi:10.48550/arxiv.1707.09457 · Cited by 122 · see citing works
Bioethics and Human Rights: A Historical Perspective
BAKER, R. (2001) · Cambridge Quarterly of Healthcare Ethics · doi:10.1017/s0963180101003048 · Cited by 119 · cited 1× · see citing works
“Reservations in India have been described as one of the world’s most radical policies [25] (see [164] for more).”
BackgroundBanaji, S., Bhat, R., Agarwal, A., Passanha, N. K. J., & Pravin, M. (2019) · London School of Economics and Political Science Research Online (London School of Economics and Political Science) · Cited by 119 · see citing works
Employment, Exclusion and 'Merit' in the Indian IT Industry
Upadhya, C. (2007) · Cited by 118 · see citing works
50 Years of Test (Un)fairness: Lessons for Machine LearningPreprint
Hutchinson, B., & Mitchell, M. (2018) · arXiv (Cornell University) · doi:10.48550/arxiv.1811.10104 · Cited by 118 · see citing works
Self-citationBottom of the Data Pyramid: Big Data and the Global South
Arora, P. (2016) · International journal of communication · Cited by 115 · see citing works
Perturbation Sensitivity Analysis to Detect Unintended Model Biases
Prabhakaran, V., Hutchinson, B., & Mitchell, M. (2019) · Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing (EMNLP-IJCNLP) · doi:10.18653/v1/d19-1578 · Cited by 115 · cited 1× · see citing works
“perturbation testing [156], data augmentation [220], adversarial testing [117], and adherence to terminology guidelines for oppressed groups, such as SIGACCESS.”
BackgroundSelf-citationThis Thing Called Fairness: Disciplinary Confusion Realizing a Value in Technology
Mulligan, D. K., Kroll, J. A., Kohli, N., & Wong, R. Y. (2019) · Proceedings of the ACM on Human-Computer Interaction · doi:10.1145/3359221 · Cited by 111 · see citing works
AI Ethics in Predictive Policing: From Models of Threat to an Ethics of Care
Asaro, P. (2019) · IEEE Technology and Society Magazine · doi:10.1109/mts.2019.2915154 · Cited by 110 · cited 1× · see citing works
“, perhaps ethics of care [22, 87, 217]) may be relevant to take into account these social relations.”
BackgroundBig Data and Its ExclusionsPreprint
Lerman, J. (2013) · SSRN Electronic Journal · doi:10.2139/ssrn.2293765 · Cited by 110 · see citing works
Text-Free User Interfaces for Illiterate and Semi-Literate Users
Medhi, I., Sagar, A., & Toyama, K. (2006) · International Conference on Information and Communication Technologies and Development (ICTD) · doi:10.1109/ictd.2006.301841 · Cited by 110 · cited 1× · see citing works
“, via delay-tolerant connectivity, low cost devices, text-free interfaces, intermediaries, and NGO partnerships (see [45, 74, 86, 115, 132, 179, 209]).”
BackgroundMitigating Gender Bias in Natural Language Processing: Literature ReviewPreprint
Sun, T., Gaut, A., Tang, S., Huang, Y., ElSherief, M., Zhao, J., Mirza, D., Belding, E., Chang, K.-W., & Wang, W. Y. (2019) · arXiv (Cornell University) · doi:10.48550/arxiv.1906.08976 · Cited by 109 · see citing works
Does Access to Formal Agricultural Credit Depend on Caste?
Kumar, S. M. (2012) · World Development · doi:10.1016/j.worlddev.2012.11.001 · Cited by 108 · cited 1× · see citing works
“predictive policing in New Delhi [127], caste [202] and ethnicity [15] biases in job applications, call-center job callbacks [27], castebased wage-gaps [124], caste discrimination in agricultural loans decisions [116], and even in online matrimonial ads [157].”
BackgroundGlobal AI Ethics: A Review of the Social Impacts and Ethical Implications of Artificial IntelligencePreprint
Hagerty, A., & Rubinov, I. (2019) · arXiv (Cornell University) · doi:10.48550/arxiv.1907.07892 · Cited by 108 · cited 1× · see citing works
“The call for a global lens in AI accountability is not new [84, 155], but the ethical principles in AI are often interpreted, prioritised, contextualised, and implemented differently across the globe [98].”
BackgroundThe human infrastructure of ICTD
Sambasivan, N., & Smyth, T. N. (2010) · International Conference on Information and Communication Technologies and Development (ICTD) · doi:10.1145/2369220.2369258 · Cited by 107 · cited 2× · see citing works
“Many respondents shared how street-level bureaucrats, administrative offices, and front line workers—the human infrastructures [182] who played a crucial role in providing recourse to marginalised Indian communities—were removed in AI systems.”
BackgroundSelf-citationData Voids: Where Missing Data Can Easily Be Exploited
boyd, D., & Golebiewski, M. (2018) · Cited by 107 · see citing works
The Effectiveness of Jobs Reservation: Caste, Religion and Economic Status in India
Borooah, V. K., Dubey, A., & Iyer, S. (2007) · Development and Change · doi:10.1111/j.1467-7660.2007.00418.x · Cited by 103 · cited 1× · see citing works
“studies have pointed to the successful redistribution of resources towards oppressed sub-groups, through reservations [43, 68, 152].”
BackgroundCrawford, K. (2013) · Harvard business review · Cited by 102 · see citing works
Science, Jews, and Secular Culture: Studies in Mid-Twentieth-Century American Intellectual History
Pells, R. H., & Hollinger, D. A. (1997) · History of Education Quarterly · doi:10.2307/369468 · Cited by 102 · see citing works
Local, sustainable, small-scale cellular networks
Heimerl, K., Hasan, S., Ali, K., Brewer, E., & Parikh, T. (2013) · Proceedings of the Sixth International Conference on Information and Communication Technologies and Development: Full Papers - Volume 1 · doi:10.1145/2516604.2516616 · Cited by 83 · cited 1× · see citing works
“, via delay-tolerant connectivity, low cost devices, text-free interfaces, intermediaries, and NGO partnerships (see [45, 74, 86, 115, 132, 179, 209]).”
BackgroundResidual unfairness in fair machine learning from prejudiced data
Kallus, N., & Zhou, A. (2018) · arXiv · doi:10.48550/arxiv.1806.02887 · Cited by 83 · cited 1× · see citing works
“Entire communities may be missing or misrepresented in datasets, exacerbated by digital divides, leading to wrong conclusions [30, 52, 53, 119] and residual unfairness [103].”
BackgroundBanalities Turned Viral: Narendra Modi and the Political Tweet
Pal, J. (2015) · Television & New Media · doi:10.1177/1527476415573956 · Cited by 80 · see citing works
The false promise of risk assessments: epistemic reform and the limits of fairness
Green, B. (2020) · Proceedings of the 2020 Conference on Fairness, Accountability, and Transparency · doi:10.1145/3351095.3372869 · Cited by 80 · cited 1× · see citing works
“The community also often repurposes terminology from US legal domains, such as “disparate impact”, “disparate treatment”, and “equal opportunity”, or use them as points of triangulation, in order to compare technical properties of fairness through analogy with legal concepts [81, 82].”
BackgroundEngaging Solidarity in Data Collection Practices for Community Health
Ismail, A., & Kumar, N. (2018) · Proceedings of the ACM on Human-Computer Interaction · doi:10.1145/3274345 · Cited by 78 · see citing works
From Margins to Seams: Imbrication, Inclusion, and Torque in the Aadhaar Identification Project
Singh, R., & Jackson, S. J. (2017) · Proceedings of the 2017 CHI Conference on Human Factors in Computing Systems · doi:10.1145/3025453.3025910 · Cited by 72 · cited 2× · see citing works
“• Language: Language can correspond to religion, class, ethnicity, • Devices and infrastructure: Internet access corresponds to gender, • Documentation : Several AI applications require state-issued documentation like Aadhaar or birth certificates, e.g., in finance.”
BackgroundWorld System versus World-Systems: A Critique
Wallerstein, I. M. (1991) · Critique of Anthropology · doi:10.1177/0308275x9101100207 · Cited by 70 · cited 2× · see citing works
“Scholars have pointed to how neoliberal AI follows the technical architecture of classic colonialism through data extraction, impairing indigenous innovation, and shipping manufactured services back to the data subjects—among communities already prone to exploitation, under-development, and inequality from centuries o…”
BackgroundThe Living Dead: Time to Integrate Scavenging into Ecological Teaching
Moleón, M., & Sánchez‐Zapata, J. A. (2015) · BioScience · doi:10.1093/biosci/biv101 · Cited by 66 · see citing works
Negotiating Women’s Agency through ICTs: A Comparative Study of Uganda and India
Masika, R., & Bailur, S. (2015) · Gender, Technology and Development · doi:10.1177/0971852414561615 · Cited by 61 · cited 1× · see citing works
“, studies have shown how women users have ‘confused’ algorithms, motivated by privacy needs [130, 178]).”
BackgroundCulture's Impact on the Importance of Fairness in Interorganizational Relationships
Lund, D. J., Scheer, L. K., & Kozlenkova, I. V. (2013) · Journal of International Marketing · doi:10.1509/jim.13.0020 · Cited by 61 · cited 1× · see citing works
“to mediate the importance placed on fairness [122].”
BackgroundOpportunities and Challenges for Artificial Intelligence in India
Kalyanakrishnan, S., Panicker, R. A., Natarajan, S., & Rao, S. (2018) · Proceedings of the AAAI/ACM Conference on AI Ethics and Society · doi:10.1145/3278721.3278738 · Cited by 61 · cited 11× · see citing works
“opportunities for AI in India [104], proposed policy deliberation frameworks that take into account the unique policy landscape of India [126], and questioned the intrusive data collection practices through Aadhaar (biometric-based unique identity for Indians) in India [158, 159].”
BackgroundInfluentialDesigning Matching Mechanisms under General Distributional Constraints
Goto, M., Kojima, F., Kurata, R., Tamura, A., & Yokoo, M. (2017) · American Economic Journal Microeconomics · doi:10.1257/mic.20160124 · Cited by 59 · see citing works
Cotter, A., Jiang, H., Wang, S., Narayan, T., Gupta, M. R., You, S., & Sridharan, K. (2018) · arXiv (Cornell University) · doi:10.48550/arxiv.1809.04198 · Cited by 57 · see citing works
Data in New Delhi's predictive policing system
Marda, V., & Narayan, S. (2020) · ACM Conference on Fairness, Accountability, and Transparency (FAccT) · doi:10.1145/3351095.3372865 · Cited by 55 · cited 2× · see citing works
“, in the premier FAccT conference, of the 138 papers published in 2019 and 2020, only a handful of papers evenmention non-West countries, and only one of them—Marda’s paper on New Delhi’s predictive policing system[127]—substantially engages with a non-Western context.”
BackgroundShin, D. (. (2019) · Javnost - The Public · doi:10.1080/13183222.2019.1589249 · Cited by 53 · cited 1× · see citing works
“, India, Brazil and Nigeria [180], and China and Korea [192].”
BackgroundRawlsian Fairness for Machine Learning.Preprint
Joseph, M., Kearns, M. J., Morgenstern, J., Neel, S., & Roth, A. (2016) · arXiv (Cornell University) · Cited by 52 · see citing works
Toward responsible AI for the next billion users
Sambasivan, N., & Holbrook, J. (2018) · interactions · doi:10.1145/3298735 · Cited by 51 · see citing works
Self-citationBéteille, A. (1990) · Man · doi:10.2307/2803715 · Cited by 50 · cited 1× · see citing works
“Dalits (’broken men’ in Marathi) are the most inferiorised category of the people, are within the social hierarchy, but excluded in caste categories [17, 35, 184, 191].”
BackgroundThe Biometric Imaginary: Bureaucratic Technopolitics in Post-Apartheid Welfare
Donovan, K. P. (2015) · Journal of Southern African Studies · doi:10.1080/03057070.2015.1049485 · Cited by 46 · cited 1× · see citing works
“6National IDs are contested in the non-West, where they are used to deliver welfare in an ‘objective’ manner, but lead to populations falling through the cracks (see research on IDs in Aadhaar [161, 194], post-apartheid South Africa [65] and Sri Lankan Tamils and Côte d’Ivoirians [24]).”
BackgroundDealing with the digital panopticon: the use and subversion of ICT in an Indian bureaucracy
Veeraraghavan, R. (2013) · Proceedings of the Sixth International Conference on Information and Communication Technologies and Development: Full Papers - Volume 1 · doi:10.1145/2516604.2516631 · Cited by 44 · cited 1× · see citing works
“Further, the high-tech illegibility of AI was noted to render recourse out of reach for groups marginalised by literacy, legal, and educational capital (see [208] for ’hiding behind a computer’).”
BackgroundObsessions with Fair Skin: Color Discourses in Indian Advertising
Karan, K. (2008) · Advertising & Society Review · doi:10.1353/asr.0.0004 · Cited by 43 · cited 1× · see citing works
“, an FR service used by police stations in eight Indian states retrained a western FR model on photos of Bollywood and regional film stars to mitigate the bias [61]—but Indian film stars are overwhelmingly fair-skinned, conventionally attractive, and able-bodied [108], not fully representative of the larger society.”
MethodCentralized Admissions for Engineering Colleges in India
Baswana, S., Chakrabarti, P. P., Chandran, S., Kanoria, Y., & Patange, U. (2019) · INFORMS Journal on Applied Analytics · doi:10.1287/inte.2019.1007 · Cited by 41 · see citing works
The Unexpected Entry and Exodus of Women in Computing and HCI in India
Thakkar, D., Sambasivan, N., Kulkarni, P., Kalenahalli Sudarshan, P., & Toyama, K. (2018) · Proceedings of the 2018 CHI Conference on Human Factors in Computing Systems · doi:10.1145/3173574.3173926 · Cited by 38 · cited 1× · see citing works
“9India fares slightly better than the US in gender representation in the tech workforce; however, gender roles and safety concerns lead to nearly 80% of women leaving computing by their thirties (coinciding with family/parenting responsibilities) [200].”
BackgroundSelf-citationA Broader View on Bias in Automated Decision-Making: Reflecting on Epistemology and DynamicsPreprint
Dobbe, R., Dean, S., Gilbert, T. K., & Kohli, N. (2018) · arXiv (Cornell University) · doi:10.48550/arxiv.1807.00553 · Cited by 36 · see citing works
Chaudhuri, M. (2005) · Cited by 34 · see citing works
On the Legal Compatibility of Fairness Definitions
Xiang, A., & Raji, I. D. (2019) · arXiv · doi:10.48550/arxiv.1912.00761 · Cited by 34 · see citing works
Are technology-enabled cash transfers really 'direct'?
Vivek, S., Narayanan, R., Chakraborty, D., Veeraraghavan, R., & Vardhan, V. (2018) · Cited by 32 · see citing works
Social Justice from a Cultural Perspective.
Leung, K., & Stephan, W. G. (2001) · Cited by 30 · see citing works
Martin, D., Prabhakaran, V., Kuhlberg, J., Smart, A., & Isaac, W. S. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2005.07572 · Cited by 28 · cited 1× · see citing works
“Since outcomes of ML deployments can be difficult to predict, proposals for accountability include participatory design [109] and participatory problem formulation [128], sharing the responsibility for design-”
BackgroundSelf-citationRajadesingan, A., Mahalingam, R., & Jurgens, D. (2019) · International Conference on Web and Social Media · doi:10.1609/icwsm.v13i01.3239 · Cited by 27 · cited 1× · see citing works
“predictive policing in New Delhi [127], caste [202] and ethnicity [15] biases in job applications, call-center job callbacks [27], castebased wage-gaps [124], caste discrimination in agricultural loans decisions [116], and even in online matrimonial ads [157].”
BackgroundBray, S., & Madonna (2013) · Cited by 27 · see citing works
Agarwal, O., Yang, Y., Wallace, B. C., & Nenkova, A. (2020) · arXiv.org · doi:10.48550/arxiv.2004.04123 · Cited by 25 · cited 1× · see citing works
“[12] shows that NLP models disproportionately fail to even detect names of people from non-Western backgrounds.”
BackgroundIsolated by Caste: Neighbourhood-Scale Residential Segregation in Indian MetrosPreprint
Bharathi, N., Malghan, D. V., & Rahman, A. (2018) · SSRN Electronic Journal · doi:10.2139/ssrn.3195672 · Cited by 20 · cited 1× · see citing works
“Many Indian zip codes are heterogeneous, with slums and posh houses abutting in the same area [20, 36], unlike the rather homogeneous zip codes in the West from the segregated past [50].”
BackgroundPartially Generative Neural Networks for Gang Crime Classification with Partial Information
Seo, S., Chan, H., Brantingham, P. J., Leap, J., Vayanos, P., Tambe, M., & Liu, Y. (2018) · Proceedings of the 2018 AAAI/ACM Conference on AI, Ethics, and Society · doi:10.1145/3278721.3278758 · Cited by 17 · cited 1× · see citing works
“, [187, 212]) potentially being exported to India and harming minorities.”
BackgroundCrawford, K. (2013) · Foreign Policy · Cited by 14 · see citing works
Nathan, L. P., Kaczmarek, M., castor, M., Cheng, S. K., & Mann, R. (2017) · International Conference on Communities and Technologies (C&T) · doi:10.1145/3083671.3083685 · Cited by 13 · cited 1× · see citing works
“tise and experience necessary to identify what is of benefit to all [144].”
BackgroundAssignment Mechanisms Under Distributional Constraints
Ashlagi, I., Saberi, A., & Shameli, A. (2020) · Operations Research · doi:10.1287/opre.2019.1887 · Cited by 11 · see citing works
Ramanathan, U. (2015) · IEEE Technology and Society Magazine · doi:10.1109/mts.2015.2396113 · Cited by 10 · cited 11× · see citing works
“opportunities for AI in India [104], proposed policy deliberation frameworks that take into account the unique policy landscape of India [126], and questioned the intrusive data collection practices through Aadhaar (biometric-based unique identity for Indians) in India [158, 159].”
BackgroundInfluentialImagined Connectivities: Synthesized Conceptions of Public Wi-Fi in Urban India
Sambasivan, N., & Aoki, P. M. (2017) · Proceedings of the 2017 CHI Conference on Human Factors in Computing Systems · doi:10.1145/3025453.3025545 · Cited by 10 · cited 1× · see citing works
“Researchers have pointed to the aspirational role played by technology in India, signifying symbolic meanings of modernity and progress via technocracy [149, 150, 176].”
BackgroundSelf-citationThe remarkable illusions of technology for social good
Sambasivan, N. (2019) · interactions · doi:10.1145/3319378 · Cited by 10 · see citing works
Self-citationKelley, P. G., Yang, Y., Heldreth, C. M., Moessner, C., Sedley, A., Kramm, A., Newman, D. T., & Woodruff, A. (2020) · Cited by 10 · see citing works
Expected UtilitarianismPreprint
Roff, H. M. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2008.07321 · Cited by 7 · see citing works
Justice as the Lens: Interrogating Rawls through Sen and Ambedkar
Rodrigues, V. (2011) · Indian Journal of Human Development · doi:10.1177/0973703020110109 · Cited by 6 · cited 1× · see citing works
“Rawl’s distributive justice [166] (also see Sen’s critique of Rawl’s idea of original position and inadequacies of impartiality-driven justice and fairness [186]).”
Background
Flagged first, then by citations
6 of the 141 references couldn’t be matched to a known paper.
Research works printed but not linked: 31
- Solon Barocas, Moritz Hardt, and Arvind Narayanan. 2017. Fairness in machine learning. NIPS Tutorial 1 (2017).
- Alexander Bogner, Beate Littig, and Wolfgang Menz. 2009. Interviewing experts. Springer.
- C Boyes-Watson. 2014. Suffolk University, College of Arts & Sciences. Center for Restorative Justice. Retrieved on November 28 (2014), 2015.
- Uma Chakravarti. 1993. Conceptualising Brahmanical patriarchy in early India: Gender, caste, class and state. Economic and Political Weekly (1993), 579--585.
- Susan Dray, Ann Light, A Dearden, Vanessa Evers, Melissa Densmore, D Ramachandran, M Kam, G Marsden, N Sambasivan, T Smyth, et al. 2012. Human--Computer Interaction for Development: Changing Human--Computer Interaction to Change the World. In The Human-Computer Interaction Handbook: Fundamentals, Evolving Technologies, and Emerging Applications, Third Edition. CRC press, 1369--1394.
- Jesse Graham, Jonathan Haidt, Sena Koleva, Matt Motyl, Ravi Iyer, Sean P Wojcik, and Peter H Ditto. 2013. Moral foundations theory: The pragmatic validity of moral pluralism. In Advances in experimental social psychology. Vol. 47. Elsevier, 55--130.
- IEEE. 2019. The IEEE Global Initiative on Ethics of Autonomous and Intelligent Systems. "Classical Ethics in A/IS". In Ethically Aligned Design: A Vision for Prioritizing Human Well-being with Autonomous and Intelligent Systems, First Edition. 36--67.
- S Inayatullah. 2006. Culture and Fairness: The Idea of Civilization Fairness. In Fairness, Globalization and Public Institutions. University of Hawaii Press, 31--33.
- Mayank Jain. 2016. India's internet population is exploding but women are not logging in. Scroll.in (26 9 2016). https://scroll.in/article/816892/indias-internet-population-is-exploding-but-women-are-not-logging-inia
- Yuvraj Joshi. 2018. Racial Indirection. UCDL Rev. 52 (2018), 2495.
- Srinivas Kodali. 2020. Aarogya Setu: A bridge too far? | Deccan Herald. https://www.deccanherald.com/specials/sunday-spotlight/aarogya-setu-a-bridge-too-far-835691.html. (Accessed on 08/01/2020).
- Nitin Kohli, Renata Barreto, and Joshua A Kroll. 2018. Translation tutorial: a shared lexicon for research and practice in human-centered software systems. In 1st Conference on Fairness, Accountability, and Transparancy. New York, NY, USA, Vol. 7.
- Subramaniam Madheswaran and Paul Attewell. 2007. Caste discrimination in the Indian urban labour market: Evidence from the National Sample Survey. Economic and political Weekly (2007), 4146--4153.
- Emma. Martinho-Truswell, Hannah. Miller, Isak Nti Asare, Andre Petheram, Richard (Oxford Insights) Stirling, Constanza Gómez Mont, and Cristina (C Minds) Martinez. 2018. Towards an AI strategy in Mexico: Harnessing the AI revolution. In AI whitepaper.
- Newslaundry and Oxfam India. 2019. Who Tells Our Stories Matters: Representation of Marginalised Caste Groups in Indian Newsrooms. (8 2019).
- Rodrigo Ochigame. 2020. The Long History of Algorithmic Fairness. Phenomenal World (2020).
- Lawrence A Palinkas, Sarah M Horwitz, Carla A Green, Jennifer P Wisdom, Naihua Duan, and Kimberly Hoagwood. 2015. Purposeful sampling for qualitative data collection and analysis in mixed method implementation research. Administration and policy in mental health and mental health services research 42, 5 (2015), 533--544.
- Amy Paul, C Jolley, and Aubra Anthony. 2018. Reflecting the Past, Shaping the Future: Making AI Work for International Development. USAID. gov (2018).
- Usha Ramanathan. 2014. Biometrics use for social protection programmes in India-Risk: Violating human rights of the poor. United Nations Research Institute for Social Development 2 (2014).
- Sharmila Rege. 1998. Dalit women talk differently: A critique of 'difference' and towards a Dalit feminist standpoint position. Economic and Political Weekly (1998), WS39-WS46.
- Sarah T Roberts. 2016. Digital refuse: Canadian garbage, commercial content moderation and the global circulation of social media's waste. Wi: journal of mobile media (2016).
- Oliver Rowntree. 2020. The mobile gender gap report 2020.
- Nithya Sambasivan, Garen Checkley, Amna Batool, Nova Ahmed, David Nemer, Laura Sanely Gaytán-Lugo, Tara Matthews, Sunny Consolvo, and Elizabeth Churchill. 2018. "Privacy is not for me, it's for those rich women": Performative Privacy Practices on Mobile Phones by Women in South Asia. In Fourteenth Symposium on Usable Privacy and Security ({SOUPS} 2018). 127--142.
- Nadia Saracini and Murali Shanmugavelan. 2019. BOND: Caste and Development. (2019).
- Amartya Kumar Sen. 2009. The idea of justice. Harvard University Press.
- Murali Shanmugavelan. 2018. Everyday Communicative Practices of Arunthathiyars: The Contribution of Communication Studies to the Analysis of Caste Exclusion and Subordination of a Dalit Community in Tamil Nadu, India. (2018).
- Ranjit Singh. 2018. 'The Living Dead'. Whispers from the Field: Ethnographic Poetry and Creative Prose (2018), 29--31.
- Nisha Tamang. 2020. Section 377: Challenges and Changing Perspectives in the Indian Society. Changing Trends in Human Thoughts and Perspectives: Science, Humanities and Culture Part I (2020), 68.
- Sukhadeo Thorat and Paul Attewell. 2007. The legacy of social exclusion: A correspondence study of job discrimination in India. Economic and political weekly (2007), 4141--4145.
- Tunisia. 2018. National AI Strategy: Unlocking Tunisia's capabilities potential. http://www.anpr.tn/national-ai-strategy-unlocking-tunisias-capabilities-potential/. In AI workshop.
- Virginius Xaxa. 2011. Tribes and social exclusion. CSSSC-UNICEF Social Inclusion Cell, An Occasional Paper 2 (2011), 1--18.
Cited by
Sorted by citations
PaLM: Scaling Language Modeling with Pathways
Chowdhery, A., Narang, S., Devlin, J., Bosma, M., Mishra, G., Roberts, A., Barham, P., Chung, H. W., Sutton, C., Gehrmann, S., et al. (2022) · arXiv · doi:10.48550/arxiv.2204.02311 · Cited by 2,590 · see citing works
“, 2020), or biases that are salient in non-Western socio-cultural contexts, such as caste (Sambasivan et al., 2021).”
BackgroundSelf-citation“Everyone wants to do the model work, not the data work”: Data Cascades in High-Stakes AI
Sambasivan, N., Kapania, S., Highfill, H., Akrong, D., Paritosh, P., & Aroyo, L. M. (2021) · Proceedings of the 2021 CHI Conference on Human Factors in Computing Systems · doi:10.1145/3411764.3445518 · Cited by 781 · see citing works
“, [31, 98, 107], understanding socio-cultural factors e.”
BackgroundSelf-citationTaxonomy of Risks posed by Language Models
Weidinger, L., Uesato, J., Rauh, M., Griffin, C., Huang, P.-S., Mellor, J., Glaese, A., Cheng, M., Balle, B., Kasirzadeh, A., et al. (2022) · 2022 ACM Conference on Fairness Accountability and Transparency · doi:10.1145/3531146.3533088 · Cited by 764 · see citing works
“The axes along which unfair bias is encoded in the LM can be rooted in localised social hierarchies such as the Hindu caste system, making it harder to anticipate harmful social stereotypes across contexts [163].”
BackgroundFairness perceptions of algorithmic decision-making: A systematic review of the empirical literature
Starke, C., Baleis, J., Keller, B., & Marcinkowski, F. (2022) · Big Data & Society · doi:10.1177/20539517221115189 · Cited by 316 · see citing works
“In one of the few studies conducted in a non-Western context, Sambasivan et al. (2021) further found that missing data and misrepresentation of subgroups in the existing data were critical reasons for algorithmic unfairness perceived.”
BackgroundSociotechnical Harms of Algorithmic Systems: Scoping a Taxonomy for Harm Reduction
Shelby, R., Rismani, S., Henne, K., Moon, A., Rostamzadeh, N., Nicholas, P., Yilla-Akbari, N., Gallegos, J., Smart, A., Garcia, E., et al. (2023) · Proceedings of the 2023 AAAI/ACM Conference on AI, Ethics, and Society · doi:10.1145/3600211.3604673 · Cited by 279 · see citing works
“proliferating false ideas about cultural groups [78, 189].”
BackgroundInfluentialAssessing the Fairness of AI Systems: AI Practitioners' Processes, Challenges, and Needs for Support
Madaio, M., Egede, L., Subramonyam, H., Wortman Vaughan, J., & Wallach, H. (2022) · Proceedings of the ACM on Human-Computer Interaction · doi:10.1145/3512899 · Cited by 156 · see citing works
“This may involve developing AI systems (and disaggregated evaluations) in ways that are responsive to local values and norms, as in recent research to re-imagine the fairness of AI systems in India [67], research that has highlighted the risks of algorithmic colonization [11] and efforts toward decolonial AI [49], or,…”
ResultRethinking Fairness: An Interdisciplinary Survey of Critiques of Hegemonic ML Fairness Approaches
Weinberg, L. (2022) · Journal of Artificial Intelligence Research · doi:10.1613/jair.1.13196 · Cited by 127 · see citing works
“…to further consolidate and centralize power, assets, control, and/or rights in the hands of a minority, such as in the context of international AI governance standards and AI for international social development that often privilege Western interests (Mohamed et al., 2020; Sambasivan et al. 2021).”
BackgroundResultInfluentialSocietal Biases in Language Generation: Progress and Challenges
Sheng, E., Chang, K.-W., Natarajan, P., & Peng, N. (2021) · Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing (Volume 1: Long Papers) · doi:10.18653/v1/2021.acl-long.330 · Cited by 131 · see citing works
“Bias Evaluations It is difficult to evaluate societal biases in NLG tasks because NLG can be open-domain, and there are many different notions of biases from various backgrounds and cultures (Sambasivan et al., 2021).”
BackgroundWhat governs attitudes toward artificial intelligence adoption and governance?
O’Shaughnessy, M. R., Schiff, D. S., Varshney, L. R., Rozell, C. J., & Davenport, M. A. (2022) · Science and Public Policy · doi:10.1093/scipol/scac056 · Cited by 142 · see citing works
Seeing Like a Toolkit: How Toolkits Envision the Work of AI Ethics
Wong, R. Y., Madaio, M. A., & Merrill, N. (2023) · Proceedings of the ACM on Human-Computer Interaction · doi:10.1145/3579621 · Cited by 103 · see citing works
“[59] their information content away from the social and political contexts where AI systems are deployed and governed [56, 70], and from the organizational contexts in which those toolkits may be used [72].”
BackgroundDetecting shortcut learning for fair medical AI using shortcut testing
Brown, A. P. Y., Tomašev, N., Freyberg, J., Liu, Y., Karthikesalingam, A., & Schrouff, J. (2023) · Nature Communications · doi:10.1038/s41467-023-39902-7 · Cited by 112 · see citing works
A toolbox for surfacing health equity harms and biases in large language models
Pfohl, S. R., Cole-Lewis, H., Sayres, R., Neal, D., Asiedu, M., Dieng, A., Tomasev, N., Rashid, Q. M., Azizi, S., Rostamzadeh, N., et al. (2024) · Nature Medicine · doi:10.1038/s41591-024-03258-2 · Cited by 113 · see citing works
Exploring How Machine Learning Practitioners (Try To) Use Fairness Toolkits
Deng, W. H., Nagireddy, M., Lee, M. S. A., Singh, J., Wu, Z. S., Holstein, K., & Zhu, H. (2022) · 2022 ACM Conference on Fairness Accountability and Transparency · doi:10.1145/3531146.3533113 · Cited by 99 · see citing works
“When faced with a fairness-related challenge, no single developer is likely to have all of the cultural and domain knowledge relevant to understanding or addressing the issue [9, 52, 92], and the appropriateness of different fairness definitions may be socially contested [73, 74, 79, 108].”
BackgroundStakeholder roles in artificial intelligence projects
Miller, G. J. (2022) · Project Leadership and Society · doi:10.1016/j.plas.2022.100068 · Cited by 91 · see citing works
”Because AI is 100% right and safe”: User Attitudes and Sources of AI Authority in India
Kapania, S., Siy, O., Clapper, G., SP, A. M., & Sambasivan, N. (2022) · CHI Conference on Human Factors in Computing Systems · doi:10.1145/3491102.3517533 · Cited by 90 · see citing works
“This is in contrast with India, which currently lacks substantial research on responsible AI development [107].”
BackgroundResultSelf-citationArtificial Intelligence (AI) in Islamic Ethics: Towards Pluralist Ethical Benchmarking for AI
Elmahjub, E. (2023) · Philosophy & Technology · doi:10.1007/s13347-023-00668-x · Cited by 200 · see citing works
AI’s Regimes of Representation: A Community-centered Study of Text-to-Image Models in South Asia
Qadri, R., Shelby, R., Bennett, C. L., & Denton, R. (2023) · 2023 ACM Conference on Fairness Accountability and Transparency · doi:10.1145/3593013.3594016 · Cited by 85 · see citing works
“First, we respond to prior calls to re-orient algorithmic fairness outside Western contexts [105, 106] by offering the first empirical study of T2I performance in South Asia.”
BackgroundDíaz, M., Kivlichan, I., Rosen, R., Baker, D., Amironesei, R., Prabhakaran, V., & Denton, R. (2022) · 2022 ACM Conference on Fairness Accountability and Transparency · doi:10.1145/3531146.3534647 · Cited by 72 · see citing works
“In particular, [47] points out that the technical, social, ethical, and physical distance between the builders of a technology and the communities it is meant to serve is large, in such settings.”
BackgroundSelf-citationEvaluation Gaps in Machine Learning Practice
Hutchinson, B., Rostamzadeh, N., Greer, C., Heller, K., & Prabhakaran, V. (2022) · ACM Conference on Fairness, Accountability, and Transparency (FAccT) · doi:10.1145/3531146.3533233 · Cited by 75 · see citing works
“Also overlooked are inference-time externalities of energy use [28, 60], cultural aspects of the ecosystem [145], and long term impacts [29].”
BackgroundSelf-citationShahid, F., & Vashistha, A. (2023) · CHI Conference on Human Factors in Computing Systems · doi:10.1145/3544548.3581538 · Cited by 68 · see citing works
“…fairness approaches, especially since the mainstream notion of fairness and universal ethics based on the Western standards can reenact coloniality [1], lead to unethical outcomes for marginalized groups [104, 123], and benefit those already controlling the power and computing resources [116].”
BackgroundYou reap what you sow: On the Challenges of Bias Evaluation Under Multilingual Settings
Talat, Z., Névéol, A., Biderman, S., Clinciu, M., Dey, M., Longpre, S., Luccioni, A. S., Masoud, M., Mitchell, M., Radev, D., et al. (2022) · doi:10.18653/v1/2022.bigscience-1.3 · Cited by 61 · see citing works
“Sambasivan et al. (2021) and Chan et al. (2021) contend there remains a significant gap between the communities governing and governed by AI, and advocate for a redistribution of powers and responsibilities in developing responsible AI.”
BackgroundMaking AI Explainable in the Global South: A Systematic Review
Okolo, C. T., Dell, N. L., & Vashistha, A. (2022) · ACM SIGCAS/SIGCHI Conference on Computing and Sustainable Societies (COMPASS) · doi:10.1145/3530190.3534802 · Cited by 59 · see citing works
“[126] analyzed challenges surrounding ML deployments in India.”
BackgroundChanging the World by Changing the Data
Rogers, A. (2021) · Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing (Volume 1: Long Papers) · doi:10.18653/v1/2021.acl-long.170 · Cited by 51 · see citing works
“…sub-groups, their values, and underlying social constructs may also differ across communities: e.g. both in India and US there is discrimination based on skin tone, but in the US context it stands for race, and in India it is a proxy for ethnicity, caste and class (Sambasivan et al., 2021a).”
BackgroundHow Platform-User Power Relations Shape Algorithmic Accountability
Ramesh, D., Kameswaran, V., Wang, D., & Sambasivan, N. (2022) · ACM Conference on Fairness, Accountability, and Transparency (FAccT) · doi:10.1145/3531146.3533237 · Cited by 53 · see citing works
“Unfortunately, these preconditions may be unique to Global North contexts [109].”
BackgroundInfluentialSelf-citationOptimizing human-centered AI for healthcare in the Global South
Okolo, C. T. (2022) · Patterns · doi:10.1016/j.patter.2021.100421 · Cited by 48 · see citing works
“Research done by Sambasi-van et al. show that typical Western takes on algorithmic fairness are not portable in the context of the cultural values and standards in India.”
Backgroundvan Berkel, N., Sarsenbayeva, Z., & Gonçalves, J. (2022) · International Journal of Human-Computer Studies · doi:10.1016/j.ijhcs.2022.102954 · Cited by 41 · see citing works
Berman, G., Goyal, N., & Madaio, M. (2024) · CHI Conference on Human Factors in Computing Systems · doi:10.1145/3613904.3642398 · Cited by 45 · see citing works
“Additionally, recent work has explored how existing algorithmic fairness paradigms originating in Western contexts may not be applicable in the global contexts of AI systems’ use [e.g., 24, 81, 83, 160, 175].”
BackgroundMethodEthical Considerations in AI Development
Singh, B. P., & Joshi, A. (2024) · Advances in computational intelligence and robotics book series · doi:10.4018/979-8-3693-2964-1.ch010 · Cited by 41 · see citing works
Humans, AI, and Context: Understanding End-Users’ Trust in a Real-World Computer Vision Application
Kim, S. S. Y., Watkins, E. A., Russakovsky, O., Fong, R., & Monroy-Hernández, A. (2023) · 2023 ACM Conference on Fairness, Accountability, and Transparency · doi:10.1145/3593013.3593978 · Cited by 37 · see citing works
“Work around those principal interests has begun to mature, as scholars begin to conduct more empirical research on factors and human perceptions of fairness [7, 13, 21], and going even further, to recognize the influence of sociotechnical context on factors of fairness, ethics, and accountability [12, 31, 37, 40, 42].”
Background(Re)Politicizing Digital Well-Being: Beyond User Engagements
Docherty, N., & Biega, A. J. (2022) · CHI Conference on Human Factors in Computing Systems · doi:10.1145/3491102.3501857 · Cited by 37 · see citing works
“[97], using India as a case study, reveals how value judgements im-”
MethodFairness in Language Models Beyond English: Gaps and Challenges
Ramesh, K., Sitaram, S., & Choudhury, M. (2023) · Findings of the Association for Computational Linguistics: EACL · doi:10.18653/v1/2023.findings-eacl.157 · Cited by 46 · see citing works
Rismani, S., Shelby, R. M., Smart, A., Jatho, E. W., Kroll, J. A., Moon, A., & Rostamzadeh, N. (2023) · CHI Conference on Human Factors in Computing Systems · doi:10.1145/3544548.3581407 · Cited by 38 · see citing works
“Similarly, scholars have developed contextual methods of assessing fairness by focusing attention on the situated power dynamics of where systems are deployed [107, 127] and transparency [123] of ML systems.”
BackgroundFATE in AI: Towards Algorithmic Inclusivity and Accessibility
Inuwa-Dutse, I. (2023) · ACM Conference on Equity and Access in Algorithms, Mechanisms, and Optimization (EAAMO) · doi:10.1145/3617694.3623233 · Cited by 34 · see citing works
“Because of the diverse audience affected by AI technology, ensuring effective transparency cannot be monolithic [23] or dominated by less encompassing viewpoints that can disproportionately affect different communities [43, 60].”
BackgroundPublic Health Calls for/with AI: An Ethnographic Perspective
Ismail, A., Thakkar, D., Madhiwalla, N., & Kumar, N. (2023) · Proceedings of the ACM on Human-Computer Interaction · doi:10.1145/3610203 · Cited by 36 · see citing works
“Researchers have studied how AI practitioners think about fairness, and considered the role of checklists and visualization tools in informing AI practice [34, 57, 73, 97].”
BackgroundEthical Reasoning over Moral Alignment: A Case and Framework for In-Context Ethical Policies in LLMs
Rao, A., Khandelwal, A., Tanmay, K., Agarwal, U., & Choudhury, M. (2023) · Findings of the Association for Computational Linguistics: EMNLP (Findings of EMNLP) · doi:10.18653/v1/2023.findings-emnlp.892 · Cited by 32 · see citing works
Applied artificial intelligence for global child health: Addressing biases and barriers
Muralidharan, V., Schamroth, J., Youssef, A., Celi, L. A., & Daneshjou, R. (2024) · PLOS Digital Health · doi:10.1371/journal.pdig.0000583 · Cited by 33 · see citing works
Lewicki, K., Lee, M. S. A., Cobbe, J., & Singh, J. (2023) · Proceedings of the 2023 CHI Conference on Human Factors in Computing Systems · doi:10.1145/3544548.3581463 · Cited by 30 · see citing works
“supports broader accountability processes on training sets, models and internal processes behind these services, as well as in the widersense of acknowledgement of known-limitations, past failures, and lessons learnt [27, 128] can help move AIaaS from the opaque, unknown and potentially harmful ‘black-boxes’ of today,…”
BackgroundInfluentialLearning about Responsible AI On-The-Job: Learning Pathways, Orientations, and Aspirations
Madaio, M., Kapania, S., Qadri, R., Wang, D., Zaldivar, A., Denton, E., & Wilcox, L. G. (2024) · ACM Conference on Fairness, Accountability, and Transparency (FAccT) · doi:10.1145/3630106.3658988 · Cited by 31 · see citing works
“Others wanted customized trainings for different geographic or cultural contexts [cf. 108] to help support AI practitioners who are “ looking [at RAI issues] in this specific country, here’s a process for fairness or how to test your models ” (E22).”
BackgroundInterrupting Merit, Subverting Legibility: Navigating Caste In ‘Casteless’ Worlds of Computing
Vaghela, P., Jackson, S. J., & Sengers, P. (2022) · CHI Conference on Human Factors in Computing Systems · doi:10.1145/3491102.3502059 · Cited by 31 · see citing works
“Asia as they might exacerbate existing caste and class inequalities [73, 116].”
BackgroundBehl, A., Sampat, B. H., Pereira, V. E., & Jabbour, C. J. C. (2023) · Annals of Operations Research · doi:10.1007/s10479-023-05624-8 · Cited by 33 · see citing works
“Lack of fairness in AI algorithms has led to bans and moratoriums (Sambasivan et al., 2021).”
Ethical Considerations from Child-Robot Interactions in Under-Resourced Communities
Kumar, M., Kumar, M., Fosch‐Villaronga, E., Singh, D., & Shukla, J. (2022) · International Journal of Social Robotics · doi:10.1007/s12369-022-00882-1 · Cited by 29 · see citing works
“Again, these researches come from a collaboration of scholars, including researchers working for industry and academia, and are primarily based out of India and the Indian context [66].”
BackgroundDetecting Cross-Geographic Biases in Toxicity Modeling on Social Media
Ghosh, S., Baker, D., Jurgens, D., & Prabhakaran, V. (2021) · Workshop on Noisy User-generated Text (W-NUT) · doi:10.18653/v1/2021.wnut-1.35 · Cited by 29 · see citing works
“However, this reliance on lexicons and annotated data limit the focus to a handful of social injustices for which resources/lexicons are available, and largely ignores biases in non-Western contexts (Sambasivan et al., 2021) and online environments containing creative or noisy language variation.”
BackgroundSelf-citationFMEA-AI: AI fairness impact assessment using failure mode and effects analysis
Li, J. J., & Chignell, M. (2022) · AI and Ethics · doi:10.1007/s43681-022-00145-9 · Cited by 29 · see citing works
“…to human rights laws that prohibit discrimination against “protected” groups [31] on the grounds of race, sex, age, religion, pregnancy and other characteristics, sometimes referred to as “prohibited grounds” [32]; these grounds depend on sociocultural context (e.g., caste in India; cf. [33]).”
BackgroundCulturally Aware and Adapted NLP: A Taxonomy and a Survey of the State of the Art
Liu, C. C., Gurevych, I., & Korhonen, A. (2025) · Transactions of the Association for Computational Linguistics · doi:10.1162/tacl_a_00760 · Cited by 51 · see citing works
Global Digital Data Governance
Aguerre, C., Campbell‐Verduyn, M., & Scholte, J. A. (2024) · Routledge eBooks · doi:10.4324/9781003388418 · Cited by 24 · see citing works
Reliability Testing for Natural Language Processing Systems
Tan, S., Joty, S., Baxter, K., Taeihagh, A., Bennett, G. A., & Kan, M. (2021) · doi:10.18653/v1/2021.acl-long.321 · Cited by 27 · see citing works
“Crucially, reliability tests can be used to highlight fairness issues in NLP systems by including sensitive attributes for the target population, but it is paramount these requirements reflect local concerns rather than any prescriptivist perspective (Sambasivan et al., 2021).”
BackgroundPopova, K., Figueras, C., Höök, K., & Lampinen, A. (2024) · Proceedings of the ACM on Human-Computer Interaction · doi:10.1145/3637434 · Cited by 23 · see citing works
“Another line of criticism relates to the claims that these tools are also limited in their attention to structural inequality [10, 26, 68] and the need for greater participation of marginalized, impacted communities, and the Global South, at large [26, 58, 81, 86].”
BackgroundIndian-BhED: A Dataset for Measuring India-Centric Biases in Large Language Models
Khandelwal, K., Tonneau, M., Bean, A. M., Kirk, H. R., & Hale, S. A. (2024) · International Conference on Information Technology for Social Good (GoodIT) · doi:10.1145/3677525.3678666 · Cited by 21 · see citing works
“Overwhelmingly, this body of work is situated in Western contexts, relying on Western data, values and historical associations, and thus is not directly portable to non-Western contexts [54].”
Utilizing subjectivity level to mitigate identity term bias in toxic comments classification
Zhao, Z., Zhang, Z., & Hopfgartner, F. (2022) · Online Social Networks and Media · doi:10.1016/j.osnem.2022.100205 · Cited by 24 · see citing works
Choi, D., Lee, U., & Hong, H. (2022) · CHI Conference on Human Factors in Computing Systems · doi:10.1145/3491102.3517574 · Cited by 21 · see citing works
“Researchers have made multiple eforts to understand how people perceive fairness in algorithmic decision making [10, 40, 47, 56, 61, 73, 83, 90, 94].”
BackgroundUrman, A., & Makhortykh, M. (2022) · New Media & Society · doi:10.1177/14614448221099536 · Cited by 21 · see citing works
“…are important for the ongoing debate about the algorithmic fairness and its applicability to the Global South and East (Mohamed et al., 2020; Sambasivan et al., 2021), in particular in the context of the responsibility of transnational tech companies for preventing systems they design from…”
BackgroundSalac, J., Oleson, A., Armstrong, L., Meur, A. L., & Ko, A. J. (2023) · ACM Conference on International Computing Education Research (ICER) · doi:10.1145/3568813.3600110 · Cited by 21 · see citing works
“In contrast, a similar study in India found that AI evoked an unquestioning aspiration in their participants due to the lack of an ecosystem to interrogate high-stakes AI [58].”
ResultOkolo, C. T., Agarwal, D., Dell, N. L., & Vashistha, A. (2024) · Proceedings of the ACM on Human-Computer Interaction · doi:10.1145/3637348 · Cited by 21 · see citing works
“…but growing body of work has started to explore how concepts of AI fairness and ex-plainability differ across Western and non-Western contexts [85, 103], little is known about what end users like CHWs, who have limited AI/digital literacy, need to know to cooperatively work with AI-driven tools…”
Intelligent Machine Learning Solutions for Cybersecurity
Singh, B., & Kaunert, C. (2024) · Advances in computational intelligence and robotics book series · doi:10.4018/979-8-3693-5380-6.ch014 · Cited by 22 · see citing works
Trustworthy AI and the Logics of Intersectional Resistance
Knowles, B., Fledderjohann, J., Richards, J. T., & Varshney, K. R. (2023) · ACM Conference on Fairness, Accountability, and Transparency (FAccT) · doi:10.1145/3593013.3593986 · Cited by 22 · see citing works
“Our recommendations below build on insights generated through examinations of power dynamics which negatively shape perceptions of AI [1, 7, 82, 88, 93], and have deep resonance with feminist approaches to AI (e.g. [52, 95]) insofar as they call for appreciating intersection-ality, actively…”
BackgroundUsing ChatGPT in HCI Research—A Trioethnography
Desai, S., Sharma, T., & Saha, P. (2023) · ACM Conference on Conversational User Interfaces (CUI) · doi:10.1145/3571884.3603755 · Cited by 19 · see citing works
“Considering the AI field's focus on research, development, and design is predominantly centered on the "West" or "Global North" [47], AI systems such as ChatGPT have a disproportionate effect on the marginalized segments of the society [6].”
BackgroundPredicting Treatment Interruption Among People Living With HIV in Nigeria: Machine Learning Approach
Ogbechie, M.-D., Walker, C. L. F., Lee, M., Gana, A. A., Oduola, A., Idemudia, A., Edor, M., Harris, E. L., Stephens, J., Gao, X., et al. (2023) · JMIR AI · doi:10.2196/44432 · Cited by 18 · see citing works
Beyond "Fairness:" Structural (In)justice Lenses on AI for EducationPreprint
Madaio, M., Blodgett, S. L., Mayfield, E., & Dixon‐Román, E. J. (2021) · arXiv (Cornell University) · doi:10.48550/arxiv.2105.08847 · Cited by 34 · see citing works
Tacheva, Z. (2022) · Big Data & Society · doi:10.1177/20539517221112901 · Cited by 17 · see citing works
“It should come as no surprise that historically marginalized persons often end up on the wrong side of the “ hyperplane, ” but also that such algorithms — trained on Western-centric datasets — get deployed in the Global South where they perform dis-mally and produce harm (Sambasivan et al., 2021).”
BackgroundRepresentation in AI Evaluations
Bergman, A. S., Hendricks, L. A., Rauh, M., Wú, B., Agnew, W. S., Kunesch, M., Duan, I., Gabriel, I., & Isaac, W. S. (2023) · ACM Conference on Fairness, Accountability, and Transparency (FAccT) · doi:10.1145/3593013.3594019 · Cited by 17 · see citing works
“In order to shift power in the right direction we need to (1) interrogate the values behind datasets [124] and (2) explore how data subjects can (re)gain control over their data, armed with the information to decide whether they wish to be represented [114].”
BackgroundInfluentialEvaluating the Diversity, Equity, and Inclusion of NLP Technology: A Case Study for Indian Languages
Khanuja, S., Ruder, S., & Talukdar, P. P. (2023) · Findings of the Association for Computational Linguistics: EACL · doi:10.18653/v1/2023.findings-eacl.131 · Cited by 21 · see citing works
“As disparities exist along multiple axes such as caste, gender, religion and so on (Sambasivan et al., 2021), it is imperative to go beyond the language level.”
BackgroundSociotechnical Implications of Generative Artificial Intelligence for Information Access
Mitra, B., Cramer, H. S. M., & Gurevich, O. (2024) · The information retrieval series · doi:10.1007/978-3-031-73147-1_7 · Cited by 20 · see citing works
Anoop, K. R., Gangan, M. P., Deepak, P., & Lajish, V. L. (2022) · Lecture notes in electrical engineering · doi:10.1007/978-981-19-4453-6_2 · Cited by 15 · see citing works
“caste and religion based discrimination [96]).”
Bennett, S., Claisse, C., Luger, E. A., & Durrant, A. C. (2023) · Proceedings of the AAAI/ACM Conference on AI Ethics and Society · doi:10.1145/3600211.3604684 · Cited by 16 · see citing works
“We contribute to a growing body of literature in the AI Ethics and Society (AIES) and Fairness, Accountability and Transparency (FAccT) communities that investigate socio-political dimensions of AI and suggests conceptual tools to inform equitable design practice [4] [6] [12][48] [67].”
BackgroundInstance-Consistent Fair Face Recognition
Li, Y., Sun, Y., Cui, Z., Shen, P., & Shan, S. (2025) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2025.3545781 · Cited by 15 · see citing works
“Morerecentlyanumber ofworkshavebeendevelopedtointroducethefairnessinto machinelearningpipelinesandmitigatethedataoralgorithm bias[9],[34],[35].”
Zufall, F., Hamacher, M., Kloppenborg, K., & Zesch, T. (2022) · Natural Legal Language Processing Workshop (NLLP) · doi:10.18653/v1/2022.nllp-1.5 · Cited by 19 · see citing works
“As group detection can work with any set of group categories, it can also be adapted to cover non-Western groups (Sambasivan et al., 2021).”
BackgroundHonor Ethics: The Challenge of Globalizing Value Alignment in AI
Wu, S., Demetriou, D., & Husain, R. A. (2023) · ACM Conference on Fairness, Accountability, and Transparency (FAccT) · doi:10.1145/3593013.3594026 · Cited by 13 · see citing works
“We note that in practice, many non-WEIRD societies currently see AI with unguarded optimism [47].”
BackgroundMadaio, M., Chen, J., Wallach, H., & Vaughan, J. W. (2024) · Proceedings of the ACM on Human-Computer Interaction · doi:10.1145/3653705 · Cited by 13 · see citing works
“…may involve understanding what marginalization or fairness means in the particular sociocultural contexts in which AI systems are deployed [e.g., 8, 69]; understanding the public’s expectations’ for fairness in algorithmic pricing models [e.g., 27] or employee management algorithms [e.g., 67], as…”
BackgroundHussain, B., Aslam, S., & Imran, A. (2025) · Acta Psychologica · doi:10.1016/j.actpsy.2025.105734 · Cited by 12 · see citing works
Seeing like a dataset from the global south
Sambasivan, N. (2021) · interactions · doi:10.1145/3466160 · Cited by 12 · see citing works
“AI-based solutions are readily adopted in highstakes domains, often too early [2].”
BackgroundSelf-citationStakeholder-accountability model for artificial intelligence projects
Maxmetrics, H., & Miller, G. J. (2022) · Journal of Economics and Management · doi:10.22367/jem.2022.44.18 · Cited by 13 · see citing works
“Those data, typically gathered from individuals and the public, may be incomplete, biased by past practices, or otherwise unavailable (Chasalow & Levy, 2021; Sambasivan et al., 2021).”
BackgroundMethodInfluentialLazy Data Practices Harm Fairness Research
Simson, J., Fabris, A., & Kern, C. A. (2024) · ACM Conference on Fairness, Accountability, and Transparency (FAccT) · doi:10.1145/3630106.3658931 · Cited by 13 · see citing works
“Overall, fairness research has neglected this important axis of discrimination and its intersections with other vulnerable identities [45, 73, 85].”
BackgroundSocietal Biases in Language Generation: Progress and ChallengesPreprint
Sheng, E., Chang, K.-W., Natarajan, P., & Peng, N. (2021) · arXiv (Cornell University) · doi:10.48550/arxiv.2105.04054 · Cited by 12 · see citing works
“Bias Evaluations It is difficult to evaluate societal biases in NLG tasks because NLG can be open-domain, and there are many different notions of biases from various backgrounds and cultures (Sambasivan et al., 2021).”
BackgroundComparing Biases and the Impact of Multilingual Training across Multiple Languages
Levy, S., John, N. A., Liu, L., Vyas, Y., Ma, J., Fujinuma, Y., Ballesteros, M., Castelli, V., & Roth, D. (2023) · Conference on Empirical Methods in Natural Language Processing (EMNLP) · doi:10.18653/v1/2023.emnlp-main.634 · Cited by 18 · see citing works
Towards Geographic Inclusion in the Evaluation of Text-to-Image Models
Hall, M., Bell, S. J., Ross, C., Williams, A., Drozdzal, M., & Soriano, A. R. (2024) · The 2024 ACM Conference on Fairness Accountability and Transparency · doi:10.1145/3630106.3658927 · Cited by 11 · see citing works
“Similarly, Prabhakaran et al. [39] call for “culturally situated” evaluations [p. 3], while Sambasivan et al. [50] argue that fairness evaluations must account for non-Western perspectives on fairness itself.”
BackgroundBiju, P. R., & Gayathri, O. (2025) · AI & Society · doi:10.1007/s00146-025-02381-z · Cited by 13 · see citing works
Sitaram, S., Choudhury, M., Patra, B., Chaudhary, V., Ahuja, K., & Bali, K. (2023) · Proceedings of the conference - Association for Computational Linguistics. Meeting · doi:10.18653/v1/2023.acl-tutorials.3 · Cited by 10 · see citing works
Language, artificial education, and future-making in indigenous language education
Pradhan, U., & Dey, J. (2023) · Learning Media and Technology · doi:10.1080/17439884.2023.2278111 · Cited by 10 · see citing works
“This problem is even more exacerbated and di ffi cult to detect, given that Indian Language NLP currently has no ‘ benchmarks ’ , or a set of tasks, by which the functionality of a software is evaluated (Sambasivan et al. 2021).”
BackgroundWhat’s Impossible about Algorithmic Fairness?
Sahlgren, O. (2024) · Philosophy & Technology · doi:10.1007/s13347-024-00814-z · Cited by 10 · see citing works
“In contrast, emerging proposals suggest designing and evaluating predictive instruments from more contex-tualized perspectives and local justice viewpoints (see, e.g., Green & Viljoen, 2020; Sambasivan et al., 2021; see also Boyd & Crawford, 2012).”
Rifat, M. R., Safir, A. H., Saha, S., Junaed, J. A., Saleki, M., Amin, M. R., & Ahmed, S. I. (2024) · ACM Conference on Fairness, Accountability, and Transparency (FAccT) · doi:10.1145/3630106.3659030 · Cited by 9 · see citing works
“…process has been criticized in emerging literature [51] by acknowl-edging the interplay of various social identities reflected on the sense-making processes involved in annotation work [19, 30, 39, 66], how FRS values are entangled with such processes has largely remained understudied till date.”
BackgroundMethodInfluentialAI4Bangladesh: AI Ethics for Bangladesh - Challenges, Risks, Principles, and Suggestions
Mayeesha, T. T., Islam, F., & Ahmed, N. (2024) · International Conference on Information and Communication Technologies and Development (ICTD) · doi:10.1145/3700794.3700820 · Cited by 10 · see citing works
“As majority of the non-technical stakeholders did not have adequate knowledge of fairness related issues even if a significant amount of work has already been done in the global south and the north [23, 47, 87, 106].”
Asiedu, M., Haykel, I., Dieng, A., Kauer, K., Ahmed, T., Ofori, F., Chan, C., Pfohl, S., Rostamzadeh, N., & Heller, K. (2025) · ACM Conference on Fairness, Accountability, and Transparency (FAccT) · doi:10.1145/3715275.3732160 · Cited by 8 · see citing works
Yuan, X., Yuan, X., He, Y., Wang, Z., Ren, J., & Bryan–Kinns, N. (2025) · Designing Interactive Systems Conference · doi:10.1145/3715336.3735808 · Cited by 9 · see citing works
“Research should also examine the issue of technological accessibility–the disadvantages that local communities face in terms of computing resources and professional maintenance capacity may restrict their long-term participation in AI-driven collaborative innovation [74, 75].”
In her Shoes: Gendered Labelling in Crowdsourced Safety Perceptions Data from India
Sengupta, N., Vaidya, A., & Evans, J. A. (2023) · ACM Conference on Fairness, Accountability, and Transparency (FAccT) · doi:10.1145/3593013.3593987 · Cited by 8 · see citing works
“We demonstrate that there is an urgent need to analyse potential bias in crowdsourced data [10], particularly from the perspective of the Global South, which has received woefully less attention in the scientific discourse around algorithmic fairness [25].”
BackgroundFlickr Africa: Examining Geo-Diversity in Large-Scale, Human-Centric Visual Data
Naggita, K., LaChance, J., & Xiang, A. (2023) · Proceedings of the AAAI/ACM Conference on AI Ethics and Society · doi:10.1145/3600211.3604659 · Cited by 8 · see citing works
“While prior works have explored diversity beyond Western nations [24, 28], studies of access to, and applicability of, AI systems in Africa remain limited [1, 2].”
Sambasivan, N. (2022) · interactions · doi:10.1145/3516515 · Cited by 8 · see citing works
Self-citationMeasurement as governance in and for responsible AIPreprint
Jacobs, A. Z. (2021) · arXiv (Cornell University) · doi:10.48550/arxiv.2109.05658 · Cited by 14 · see citing works
On Evaluating and Mitigating Gender Biases in Multilingual Settings
Vashishtha, A., Ahuja, K., & Sitaram, S. (2023) · Findings of the Association for Computational Linguistics: ACL (Findings of ACL) · doi:10.18653/v1/2023.findings-acl.21 · Cited by 9 · see citing works
Gebru, T., & Denton, E. (2024) · Foundations and Trends® in Computer Graphics and Vision · doi:10.1561/0600000102 · Cited by 7 · see citing works
Algorithmic solutions, subjectivity and decision errors: a study of AI accountability
Biju, P. R., & Gayathri, O. (2024) · Digital Policy Regulation and Governance · doi:10.1108/dprg-05-2024-0090 · Cited by 7 · see citing works
DBNLP: detecting bias in natural language processing system for India-centric languages
G, K. K. T., Mendke, S., Parihar, R., Mayya, S., Venkatesh, S., & Koolagudi, S. G. (2025) · International Journal of Information Technology · doi:10.1007/s41870-025-02437-9 · Cited by 7 · see citing works
“[17] caters to the Indian context and aims to “re-contextualize data and models, empower oppressed communities, and enable Fair-ML ecosystems”.”
Algorithmic Bias and Discrimination in India: A Looming Crisis
David, D., Rajeshwari, B., & Timhna, S. (2025) · Journal of Development Policy and Practice · doi:10.1177/24551333251343358 · Cited by 7 · see citing works
“AI systems can perpetuate religious biases, particularly in a politically charged environment (Flowers, 2019; Sambasivan et al., 2021).”
Beyond the ML Model: Applying Safety Engineering Frameworks to Text-to-Image Development
Rismani, S., Shelby, R. M., Smart, A., Santos, R. D., Moon, A., & Rostamzadeh, N. (2023) · Proceedings of the AAAI/ACM Conference on AI Ethics and Society · doi:10.1145/3600211.3604685 · Cited by 10 · see citing works
“Similarly, Responsible AI scholars have examined data collection processes and pointed out clear areas for improvement [18, 31, 70].”
BackgroundMethodBuilding Stereotype Repositories with Complementary Approaches for Scale and Depth
Dev, S., Jha, A., Goyal, J., Tewari, D., Dave, S., & Prabhakaran, V. (2023) · Workshop on Cross-Cultural Considerations in NLP (C3NLP) · doi:10.18653/v1/2023.c3nlp-1.9 · Cited by 7 · see citing works
Self-citationA Global South perspective for ethical algorithms and the State
Sengupta, N., Subramanian, V., Mukhopadhyay, A., & Scaria, A. G. (2023) · Nature Machine Intelligence · doi:10.1038/s42256-023-00621-9 · Cited by 7 · see citing works
Harnessing Machine Learning Intelligence Against Cyber Threats
Singh, B., Kaunert, C., & Gautam, R. (2024) · Advances in business strategy and competitive advantage book series · doi:10.4018/979-8-3693-7658-4.ch016 · Cited by 7 · see citing works
Building Bridges: Generative Artworks to Explore AI EthicsPreprint
Srinivasan, R., & Parikh, D. (2021) · arXiv (Cornell University) · doi:10.48550/arxiv.2106.13901 · Cited by 7 · see citing works
“Most works concerning AI ethics are influenced heavily by western cultures in their definitions, axes of discrimination, and philosophical roots [Sambasivan et al., 2021].”
What governs attitudes toward artificial intelligence adoption and governance?Preprint
O’Shaughnessy, M. R., Schiff, D. S., Varshney, L. R., Rozell, C. J., & Davenport, M. A. (2021) · doi:10.31219/osf.io/pkeb8 · Cited by 7 · see citing works
Addressing Algorithmic Bias in India: Ethical Implications and PitfallsPreprint
Sood, Y. (2023) · SSRN Electronic Journal · doi:10.2139/ssrn.4466681 · Cited by 7 · see citing works
Bias in the Algorithm: Issues Raised Due to Use of Facial Recognition in India
Basheer, I. P. (2024) · Journal of Development Policy and Practice · doi:10.1177/24551333241283992 · Cited by 6 · see citing works
“Half the Indian population lacks access to the Internet—the excluded half is primarily women, rural communities, and Adivasis [96, 106, 153, 168, 189] (Sambasivan et al, 2021).”
- Authors
- Sambasivan, N., Arnesen, E., Hutchinson, B., Doshi, T., & Prabhakaran, V.
- Affiliations
- Google (United States) (US)
- Venue
- Proceedings of the 2021 ACM Conference on Fairness, Accountability, and Transparency
- Publisher
- ACM
- Published
- 2021-03
- Pages
- 315-328
- OpenAlex
- W3123871787
- Semantic Scholar
- 187608bf94b2dccd25d1266ed925abf7b55dbb2e
- Cited by
- 228
- References
- 141
- Citations by year
- 2026: 60 · 2025: 45 · 2024: 40 · 2023: 40 · 2022: 30 · 2021: 13
- FWCI
- 27.85
- Type
- conference paper
- Language
- en
- Open access
- hybrid
- doi.org
- Licence
- public-domain
- creativecommons.org/licenses/by-nc-sa/4.0/ (vor)
- Locations
- ACM Conference on Fairness, Accountability, and Transparency (FAccT) (publishedVersion, public-domain)
- arXiv (Cornell University) (submittedVersion)
- arXiv (Cornell University)
- Topics
- Social Sciences › Safety Research › Ethics and Social Impacts of AI
- Physical Sciences › Computer Science › Artificial Intelligence › Explainable Artificial Intelligence (XAI)
- Social Sciences › Law › Digitalization, Law, and Regulation
- Keywords
- algorithmic fairness, decolonial AI, model fairness, data reliability
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Claims read by deepseek-v4.1-flash under definition version 3 (extract-v3) on 2026-10-09.
Classified by jev-1.13-20260917 (gold-guideline-v0.2) on 2026-10-09.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
