The DOI matches its Crossref record; 70 of the 70 references it cites were found in a registry or an index; none of the 81 sources its claims cite is checked for support yet.
Ethical principles in machine learning and artificial intelligence: cases from the field and possible ways forward
Open accessAbstract
Decision-making on numerous aspects of our daily lives is being outsourced to machine-learning (ML) algorithms and artificial intelligence (AI), motivated by speed and efficiency in the decision process. ML approaches—one of the typologies of algorithms underpinning artificial intelligence—are typically developed as black boxes. The implication is that ML code scripts are rarely scrutinised; interpretability is usually sacrificed in favour of usability and effectiveness. Room for improvement in practices associated with programme development have also been flagged along other dimensions, including inter alia fairness, accuracy, accountability, and transparency. In this contribution, the production of guidelines and dedicated documents around these themes is discussed. The following applications of AI-driven decision-making are outlined: (a) risk assessment in the criminal justice system, and (b) autonomous vehicles, highlighting points of friction across ethical principles. Possible ways forward towards the implementation of governance on AI are finally examined.
Overview
AI snapshot
Written from the paper’s full text.
TL;DR
The paper reviews ethical principles in machine learning and AI, maps the proliferation of guideline documents, and examines two applied cases—criminal justice recidivism-risk assessment and autonomous vehicles—to show trade-offs among fairness, accuracy, transparency, accountability, and privacy. It argues that AI ethics remains nascent, guidelines are mostly voluntary, and governance should include scrutiny tools, stakeholder inclusion, and possibly super partes standards.
Key findings
- Guideline production has skyrocketed since 2016; Floridi and Cowls identified 47 principles mapping onto five dimensions (beneficence, non-maleficence, autonomy, justice, explicability), while Jobin et al. reviewed 84 documents and classified values around transparency, justice and fairness, non-maleficence, responsibility, and privacy.
- In the COMPAS case, an investigative report examined a pool of cases with recidivism scores for >18,000 criminal defendants in Broward County, Florida and flagged potential racial bias; the company refused to disclose full proprietary code, and while about 140 variables exist, only core variables were disclosed, with race not among them. Interpretable alternatives reportedly achieved similar accuracy using only two variables—age and total number of previous convictions.
- Autonomous vehicles pose continuous decisions and a “statistical trolley dilemma”; reported failures include inability to perceive a pedestrian as such and acceleration when braking was required due to conflicting instructions from different algorithms, with millions of lines of code making causality practically impossible to scrutinise.
- The paper highlights points of friction across ethical dimensions, including transparency versus accountability/privacy/property, fairness versus accuracy, group versus individual fairness, and ex-ante assumptions versus ex-post consequences.
Methods
- Conceptual/review paper: it discusses the production of AI ethics guidelines and secondary literature, then examines two applied cases—recidivism-risk assessment in the criminal justice system and autonomous vehicles.
- Draws on cited secondary literature and reports, including Floridi and Cowls, Jobin et al., COMPAS/ProPublica, and the MIT Moral Machine; no primary empirical data collection or systematic review protocol is stated.
- Main analysis is a narrative synthesis of ethical dimensions, trade-offs, and possible governance ways forward, such as global sensitivity analysis, NUSAP, sensitivity auditing, extended peer communities, and a super partes authority for standards.
Limitations
- The paper states that AI ethics is at its infancy and that the development process of algorithms cannot be perfect; one must be open to negotiation and work with imperfections and clumsiness.
- It notes that guidelines are largely voluntary with no sanctionary scheme for non-compliance, accountability is mainly theoretical, transparency may not suffice, and proprietary code is often undisclosed.
- Evident from the text: there is no formal limitations section; the work is a narrative synthesis with two selected case studies, not a systematic review or empirical evaluation.
Model-written summary. Check it against the paper before relying on it.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
76 Not checked yet · 45 Own interpretation · 6 No basis given
In the paper
“which also lacks a full understanding of the consequences of the actions enacted by algorithms (Wallach and Allen, 2008).”
§ Ethical principles in machine learning and artificial intelligence: cases from the field and possible ways forward > Guidelines and secondary literature on AI ethics, its dimensions and stakes
What the sources say
- Not checked yetcites Wallach and Allen, 2008
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Loss in accuracy may be produced by the exclusion of politically critical features (such as gender, race, age, etc.) from the pool of training predictive variables.”
§ Ethical principles in machine learning and artificial intelligence: cases from the field and possible ways forward > Points of friction between ethical dimensions.
hedgedconditional- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: unclear — It is unclear why the paper says it holds.
In the paper
“Furthermore, issues of garbage-in-garbage-out (Saltelli and Funtowicz, 2014) may be prone to emerge in contexts when external control is entirely removed.”
§ Ethical principles in machine learning and artificial intelligence: cases from the field and possible ways forward > Introduction
What the sources say
- Not checked yetcites Saltelli and Funtowicz, 2014
hedgedconditional- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Rests on an assumption: Decision-making on numerous aspects of daily lives is being outsourced to ML algorithms and AI, motivated by speed and efficiency in the decision process.
In the paper
“However, publishing scripts expose their developers to the public scrutiny of professional programmers, who may find shortcomings in the development of the code (Sonnenburg, 2007).”
§ Ethical principles in machine learning and artificial intelligence: cases from the field and possible ways forward > Points of friction between ethical dimensions.
What the sources say
- Not checked yetcites Sonnenburg, 2007
plaindirect- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“This aspect points again to possible political issues of a trade-off between fairness, demanded by society, and algorithmic accuracy, demanded by, e.g., a private actor.”
§ Ethical principles in machine learning and artificial intelligence: cases from the field and possible ways forward > Points of friction between ethical dimensions.
hedgedcorrelational- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Finally, the accountability of professionals or institutions is at this stage mainly theoretical, having the vast majority of these guidelines been defined on a merely voluntary basis and hence with the total lack of a sanctionary scheme for non-compliance.”
§ Ethical principles in machine learning and artificial intelligence: cases from the field and possible ways forward > Guidelines and secondary literature on AI ethics, its dimensions and stakes
plaindirect- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
In the paper
“This issue may be further exacerbated by the offer of new services of auto-ML (Chin, 2019), where the entire algorithm development workflow is automatised and the residual human control practically removed.”
§ Ethical principles in machine learning and artificial intelligence: cases from the field and possible ways forward > Introduction
What the sources say
- Not checked yetcites Chin, 2019
hedgedconditional- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Rests on an assumption: Decision-making on numerous aspects of daily lives is being outsourced to ML algorithms and AI, motivated by speed and efficiency in the decision process.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“However, this process appears problematic due to four main differences between medicine and the medical professionals on one side, and AI and its developers on the other.”
§ Ethical principles in machine learning and artificial intelligence: cases from the field and possible ways forward > Guidelines and secondary literature on AI ethics, its dimensions and stakes
hedgeddirect- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Thirdly, AI faces the difficulty of translating overarching principle into practices.”
§ Ethical principles in machine learning and artificial intelligence: cases from the field and possible ways forward > Guidelines and secondary literature on AI ethics, its dimensions and stakes
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Even its current setting of seeking maximum speed, efficiency and profit clashes with the resource and time requirements of an ethical assessment and/or counselling.”
§ Ethical principles in machine learning and artificial intelligence: cases from the field and possible ways forward > Guidelines and secondary literature on AI ethics, its dimensions and stakes
plaindirect- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Probably not a substantial claim. 72
- It may state two separate claims as one. 28
In the paper
“Hence, he argues against full transparency along four main lines of reasoning: (i) leaking of privacy sensitive data into the open; (ii) backfiring into an implicit invitation to game the system; (iii) harming of the company property rights with negative consequences on their competitiveness (and on the developers reputation as discussed above); (iv) inherent opacity of algorithms, whose interpret”
§ Ethical principles in machine learning and artificial intelligence: cases from the field and possible ways forward > Points of friction between ethical dimensions.
What the sources say
- Not checked yetcites de Laat (2018) source
plaindirect- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim. It may state two separate claims as one.
In the paper
“Even articles 21 and 22 of the recent European Union General Data Protection Regulation include passages functional to AI governance, although further action has been recently demanded from the European Parliament (De Sutter, 2019).”
§ Ethical principles in machine learning and artificial intelligence: cases from the field and possible ways forward > Guidelines and secondary literature on AI ethics, its dimensions and stakes
What the sources say
- Not checked yetcites De Sutter, 2019
plaincorrelational- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: fails — The paper’s conclusions may not depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim. The paper’s conclusions may not depend on it. It may state two separate claims as one.
In the paper
“To this end, Watson and Floridi (2019) defined a formal framework for interpretable ML, where explanatory accuracy can be assessed against algorithmic simplicity and relevance.”
§ Ethical principles in machine learning and artificial intelligence: cases from the field and possible ways forward > Points of friction between ethical dimensions.
What the sources say
- Not checked yetcites Watson and Floridi (2019)
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: fails — The paper’s conclusions may not depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim. The paper’s conclusions may not depend on it.
In the paper
“The group produced a deliverable on the required criteria for AI trustworthiness (Daly, 2019).”
§ Ethical principles in machine learning and artificial intelligence: cases from the field and possible ways forward > Guidelines and secondary literature on AI ethics, its dimensions and stakes
What the sources say
- Not checked yetcites Daly, 2019
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: fails — The paper’s conclusions may not depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim. The paper’s conclusions may not depend on it.
In the paper
“Jobin et al. (2019) reviewed 84 documents, which were produced by several actors of the field, almost half of which from private companies or governmental agencies.”
§ Ethical principles in machine learning and artificial intelligence: cases from the field and possible ways forward > Guidelines and secondary literature on AI ethics, its dimensions and stakes
What the sources say
- Not checked yetcites Jobin et al. (2019)
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: fails — The paper’s conclusions may not depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim. The paper’s conclusions may not depend on it. It may state two separate claims as one.
In the paper
“These have been documented, for instance, in the case of the Amazon website, for which errors, such as the quotation of plain items (often books) up to 10,000 dollars (Smith, 2018) have been reported.”
§ Ethical principles in machine learning and artificial intelligence: cases from the field and possible ways forward > Introduction
What the sources say
- Not checked yetcites Smith, 2018
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: fails — The paper’s conclusions may not depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim. The paper’s conclusions may not depend on it.
In the paper
“Raji et al. (2020) suggest that a process of algorithmic auditing within the software-development company could help in tackling some of the ethical issues raised.”
§ Ethical principles in machine learning and artificial intelligence: cases from the field and possible ways forward > Points of friction between ethical dimensions.
What the sources say
- Not checked yetcites Raji et al. (2020)
hedgedconditional- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: fails — The paper’s conclusions may not depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim. The paper’s conclusions may not depend on it.
In the paper
“Ananny and Crawford (2018) comment that resorting to full algorithmic transparency may not be an adequate means to address their ethical dimensions; opening up the black-box would not suffice to disclose their modus operandi.”
§ Ethical principles in machine learning and artificial intelligence: cases from the field and possible ways forward > Points of friction between ethical dimensions.
What the sources say
- Not checked yetcites Ananny and Crawford (2018) source
hedged- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim. It may state two separate claims as one.
The paper states this without a source or data behind it here, so there is nothing to check it against. The quote shows it is in the paper.
In the paper
“An international European initiative is the multi-stakeholder European Union High-Level Expert Group on Artificial Intelligence, which is composed by 52 experts from academia, civil society, and industry.”
§ Ethical principles in machine learning and artificial intelligence: cases from the field and possible ways forward > Guidelines and secondary literature on AI ethics, its dimensions and stakes
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: fails — The paper’s conclusions may not depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim. The paper’s conclusions may not depend on it.
In the paper
“The classification proposed by Jobin et al. (2019) is around a slightly different set of values: transparency, justice and fairness, non-maleficience, responsibility and privacy.”
§ Ethical principles in machine learning and artificial intelligence: cases from the field and possible ways forward > Guidelines and secondary literature on AI ethics, its dimensions and stakes
What the sources say
- Not checked yetcites Jobin et al. (2019)
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: fails — The paper’s conclusions may not depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim. The paper’s conclusions may not depend on it.
In the paper
“ML algorithms are autonomous and self-sufficient when performing their learning function.”
§ Ethical principles in machine learning and artificial intelligence: cases from the field and possible ways forward > Introduction
What the sources say
- Not checked yetcites Samuel, 1959 source
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: fails — The paper’s conclusions may not depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim. The paper’s conclusions may not depend on it.
In the paper
“A set of 47 principles has been identified, which mapped onto five overarching dimensions (Floridi and Cowls, 2019): beneficence, non-maleficence, autonomy, justice and, explicability.”
§ Ethical principles in machine learning and artificial intelligence: cases from the field and possible ways forward > Guidelines and secondary literature on AI ethics, its dimensions and stakes
What the sources say
- Not checked yetcites Floridi and Cowls, 2019
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: fails — The paper’s conclusions may not depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim. The paper’s conclusions may not depend on it. It may state two separate claims as one.
In the paper
“Attributing responsibilities in complicated projects where many parties and developers may be involved, an issue known as the problem of many hands (Nissenbaum, 1996), may indeed be very difficult.”
§ Ethical principles in machine learning and artificial intelligence: cases from the field and possible ways forward > Points of friction between ethical dimensions.
What the sources say
- Not checked yetcites Nissenbaum, 1996 source
hedged- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: fails — The paper’s conclusions may not depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim. The paper’s conclusions may not depend on it.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Larger interpretability could be in principle achieved by using simpler algorithms, although this may come at the expenses of accuracy.”
§ Ethical principles in machine learning and artificial intelligence: cases from the field and possible ways forward > Points of friction between ethical dimensions.
hedgedconditional- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: fails — The paper’s conclusions may not depend on it.
- Based: fails — The paper may give no reason why it holds.
Probably not a substantial claim. The paper’s conclusions may not depend on it. The paper may give no reason why it holds. It may state two separate claims as one.
In the paper
“the increasing use of AI in many fields raises new important questions about its suitability (Yu et al., 2018).”
§ Ethical principles in machine learning and artificial intelligence: cases from the field and possible ways forward > Introduction
What the sources say
- Not checked yetcites Yu et al., 2018
plaindirect- Asserted: passes — The authors put it forward as true.
- Checkable: fails — It may be too vague to check against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim. It may be too vague to check against evidence.
- Rests on an assumption: Decision-making on numerous aspects of daily lives is being outsourced to ML algorithms and AI, motivated by speed and efficiency in the decision process.
In the paper
“The France's Digital Republic Act gives the right to an explanation as regards decisions on an individual made through the use of administrative algorithms (Edwards and Veale, 2018).”
§ Ethical principles in machine learning and artificial intelligence: cases from the field and possible ways forward > Guidelines and secondary literature on AI ethics, its dimensions and stakes
What the sources say
- Not checked yetcites Edwards and Veale, 2018 source
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: fails — The paper’s conclusions may not depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim. The paper’s conclusions may not depend on it.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Secondly, a formal profession with a set of clearly defined and governed good-behaviour practices exists in medicine.This is not the case for AI”
§ Ethical principles in machine learning and artificial intelligence: cases from the field and possible ways forward > Guidelines and secondary literature on AI ethics, its dimensions and stakes
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
It may state two separate claims as one.
In the paper
“In this context, China has also been allocating efforts on privacy and data protection (Roberts, 2019).”
§ Ethical principles in machine learning and artificial intelligence: cases from the field and possible ways forward > Guidelines and secondary literature on AI ethics, its dimensions and stakes
What the sources say
- Not checked yetcites Roberts, 2019
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: fails — The paper’s conclusions may not depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim. The paper’s conclusions may not depend on it.
In the paper
“In all these fields, an increasing amount of functions are being ceded to algorithms to the detriment of human control, raising concern for loss of fairness and equitability (Sareen et al., 2020).”
§ Ethical principles in machine learning and artificial intelligence: cases from the field and possible ways forward > Introduction
What the sources say
- Not checked yetcites Sareen et al., 2020 source
plaindirect- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
It may state two separate claims as one.
- Rests on an assumption: Decision-making on numerous aspects of daily lives is being outsourced to ML algorithms and AI, motivated by speed and efficiency in the decision process.
In the paper
“The production of dedicated documents has been skyrocketing from 2016 (Jobin et al., 2019).”
§ Ethical principles in machine learning and artificial intelligence: cases from the field and possible ways forward > Guidelines and secondary literature on AI ethics, its dimensions and stakes
What the sources say
- Not checked yetcites Jobin et al., 2019
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: fails — The paper’s conclusions may not depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim. The paper’s conclusions may not depend on it.
In the paper
“Other potentially relevant dimensions, such as accountability and responsibility, were rarely defined in the studies reviewed by these authors.”
§ Ethical principles in machine learning and artificial intelligence: cases from the field and possible ways forward > Guidelines and secondary literature on AI ethics, its dimensions and stakes
What the sources say
- Not checked yetcites Jobin et al. (2019)
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: fails — The paper’s conclusions may not depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim. The paper’s conclusions may not depend on it.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Firstly, the medical professional rests on common aims and fiduciary duties, which AI developers lack.”
§ Ethical principles in machine learning and artificial intelligence: cases from the field and possible ways forward > Guidelines and secondary literature on AI ethics, its dimensions and stakes
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: unclear — It is unclear why the paper says it holds.
It may state two separate claims as one.
In the paper
“According to Floridi and Cowls (2019) a prominent role is also played by the autonomy dimension; the possibility of refraining from ceding decision power to AI for overriding reasons (e.g., the gain of efficacy is not deemed fit to justify the loss of control over decision-making).”
§ Ethical principles in machine learning and artificial intelligence: cases from the field and possible ways forward > Points of friction between ethical dimensions.
What the sources say
- Not checked yetcites Floridi and Cowls (2019)
plaincorrelational- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: fails — The paper’s conclusions may not depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim. The paper’s conclusions may not depend on it.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“While mistakes about monetary values may be easy to spot, the situation may become more complex and less intelligible when incommensurable dimensions come to play.”
§ Ethical principles in machine learning and artificial intelligence: cases from the field and possible ways forward > Introduction
hedgedconditional- Asserted: passes — The authors put it forward as true.
- Checkable: fails — It may be too vague to check against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: fails — The paper may give no reason why it holds.
Probably not a substantial claim. It may be too vague to check against evidence. The paper may give no reason why it holds. It may state two separate claims as one.
In the paper
“The latter is a new dimension specifically acknowledged in the case of AI, while the others were already identified in the controversial domain of bioethics.”
§ Ethical principles in machine learning and artificial intelligence: cases from the field and possible ways forward > Guidelines and secondary literature on AI ethics, its dimensions and stakes
What the sources say
- Not checked yetcites Floridi and Cowls, 2019
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: fails — The paper’s conclusions may not depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim. The paper’s conclusions may not depend on it. It may state two separate claims as one.
In the paper
“Mittelstadt (2019) critically analysed the current debate and actions in the field of AI ethics and noted that the dimensions addressed in AI ethics are converging towards those of medical ethics.”
§ Ethical principles in machine learning and artificial intelligence: cases from the field and possible ways forward > Guidelines and secondary literature on AI ethics, its dimensions and stakes
What the sources say
- Not checked yetcites Mittelstadt (2019) source
plaincorrelational- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: fails — The paper’s conclusions may not depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim. The paper’s conclusions may not depend on it.
In the paper
“Seven of the most prominent value statements from the AI/ML fields were examined in Greene et al. (2019)”
§ Ethical principles in machine learning and artificial intelligence: cases from the field and possible ways forward > Guidelines and secondary literature on AI ethics, its dimensions and stakes
What the sources say
- Not checked yetcites Greene et al. (2019)
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: fails — The paper’s conclusions may not depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim. The paper’s conclusions may not depend on it.
In the paper
“de Laat (2018) argues that transparency and accountability may even be at odds in the case of algorithms.”
§ Ethical principles in machine learning and artificial intelligence: cases from the field and possible ways forward > Points of friction between ethical dimensions.
What the sources say
- Not checked yetcites de Laat (2018) source
hedgedconditional- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim.
In the paper
“Contrasting dimensions in terms of the theoretical framing of the issue also emerged from the review of Jobin et al. (2019), as regards interpretation of ethical principles, reasons for their importance, ownership and responsibility of their implementation.”
§ Ethical principles in machine learning and artificial intelligence: cases from the field and possible ways forward > Points of friction between ethical dimensions.
What the sources say
- Not checked yetcites Jobin et al. (2019)
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: fails — The paper’s conclusions may not depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim. The paper’s conclusions may not depend on it.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“However, these correlations may be discriminatory, if not illegal.”
§ Ethical principles in machine learning and artificial intelligence: cases from the field and possible ways forward > Introduction
hedgedcorrelational- Asserted: passes — The authors put it forward as true.
- Checkable: fails — It may be too vague to check against evidence.
- Load-bearing: fails — The paper’s conclusions may not depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim. It may be too vague to check against evidence. The paper’s conclusions may not depend on it. It may state two separate claims as one.
In the paper
“according to O'Neil (2016), a main shortcoming of ML approaches is the fact these resort to proxies for driving trends, such as person's ZIP code or language in relation with the capacity of an individual to pay back a loan or handle a job, respectively.”
§ Ethical principles in machine learning and artificial intelligence: cases from the field and possible ways forward > Introduction
What the sources say
- Not checked yetcites O'Neil (2016)
plaindirect- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: fails — The paper’s conclusions may not depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim. The paper’s conclusions may not depend on it.
In the paper
“In other words, machines autonomy could be reduced in favour of human autonomy according to this metaautonomy dimension.”
§ Ethical principles in machine learning and artificial intelligence: cases from the field and possible ways forward > Points of friction between ethical dimensions.
What the sources say
- Not checked yetcites Floridi and Cowls (2019)
hedgeddirect- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: fails — The paper’s conclusions may not depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim. The paper’s conclusions may not depend on it.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Conflicts may also emerge between the requirements to overcome potential algorithm deficits in accuracy associated with large data bases and the individual rights to privacy and autonomy of decision.”
§ Ethical principles in machine learning and artificial intelligence: cases from the field and possible ways forward > Points of friction between ethical dimensions.
hedgedconditional- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: fails — The paper’s conclusions may not depend on it.
- Based: fails — The paper may give no reason why it holds.
Probably not a substantial claim. The paper’s conclusions may not depend on it. The paper may give no reason why it holds.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“This also applies to different ethical principles, resulting in the trade-offs previously discussed, difficulties in setting prioritisation strategies, operationalisation and actual compliance with the guidelines.”
§ Ethical principles in machine learning and artificial intelligence: cases from the field and possible ways forward > Points of friction between ethical dimensions.
plaindirect- Asserted: passes — The authors put it forward as true.
- Checkable: fails — It may be too vague to check against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim. It may be too vague to check against evidence. It may state two separate claims as one.
The paper states this without a source or data behind it here, so there is nothing to check it against. The quote shows it is in the paper.
In the paper
“Sensitive governmental areas, such as national security and defence, and the private sector (the largest user and producer of ML algorithms by far) are excluded from this document.”
§ Ethical principles in machine learning and artificial intelligence: cases from the field and possible ways forward > Guidelines and secondary literature on AI ethics, its dimensions and stakes
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: fails — The paper’s conclusions may not depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim. The paper’s conclusions may not depend on it.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“The algorithm may also face a dilemma between low probability of a serious harm and higher probability of a mild harm.”
§ Ethical principles in machine learning and artificial intelligence: cases from the field and possible ways forward > Machine-learning algorithms in the field of autonomous vehicles
hedged- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: fails — The paper’s conclusions may not depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim. The paper’s conclusions may not depend on it.
In the paper
“Additionally, a modelling of the modelling process would assist in the model transparency and in addressing questions such as: Are the results from a particular model more sensitive to changes in the model and the methods used to estimate its parameters, or to changes in the data?(Majone, 1989).”
§ Ethical principles in machine learning and artificial intelligence: cases from the field and possible ways forward > Discussion and conclusions
What the sources say
- Not checked yetcites Majone, 1989
hedgedconditional- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: fails — The paper’s conclusions may not depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim. The paper’s conclusions may not depend on it. It may state two separate claims as one.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Accuracy in the context of self-autonomous vehicles rests on their capacity to correctly simulate the course of the events.”
§ Ethical principles in machine learning and artificial intelligence: cases from the field and possible ways forward > Machine-learning algorithms in the field of autonomous vehicles
plaindirect- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: fails — The paper may give no reason why it holds.
Probably not a substantial claim. The paper may give no reason why it holds.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“The case of autonomous vehicles, also known as self-driving vehicles, poses different challenges as a continuity of decisions is to be enacted while the vehicle is moving.It is not a one-off decision as in the case of the assessment of recidivism risk.”
§ Ethical principles in machine learning and artificial intelligence: cases from the field and possible ways forward > Machine-learning algorithms in the field of autonomous vehicles
plaindirect- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: fails — The paper’s conclusions may not depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim. The paper’s conclusions may not depend on it.
In the paper
“While the total number of variables amounts to about 140, only the core variables were disclosed (Northpointe, 2012).The race of the subject was not one of those.”
§ Ethical principles in machine learning and artificial intelligence: cases from the field and possible ways forward > Points of friction between ethical dimensions.
What the sources say
- Not checked yetcites Northpointe, 2012
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
It may state two separate claims as one.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Eventually, the set of decision rules underpinning the AI algorithm derives from human-made assumptions, such as, where to define the boundary between action and no action, between different possible choices.”
§ Ethical principles in machine learning and artificial intelligence: cases from the field and possible ways forward > Discussion and conclusions
plaindirect- Asserted: passes — The authors put it forward as true.
- Checkable: fails — It may be too vague to check against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim. It may be too vague to check against evidence.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“It also poses further tension between the accuracy a vehicle manufacturer seeks and the capability to keep up the agreed fairness standards upstream from the algorithm development process.”
§ Ethical principles in machine learning and artificial intelligence: cases from the field and possible ways forward > Machine-learning algorithms in the field of autonomous vehicles
plaindirect- Asserted: passes — The authors put it forward as true.
- Checkable: fails — It may be too vague to check against evidence.
- Load-bearing: fails — The paper’s conclusions may not depend on it.
- Based: fails — The paper may give no reason why it holds.
Probably not a substantial claim. It may be too vague to check against evidence. The paper’s conclusions may not depend on it. The paper may give no reason why it holds.
The paper states this without a source or data behind it here, so there is nothing to check it against. The quote shows it is in the paper.
In the paper
“This analysis was debunked by the company, which, however, refused to disclose the full details of its proprietary code.”
§ Ethical principles in machine learning and artificial intelligence: cases from the field and possible ways forward > Points of friction between ethical dimensions.
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: fails — The paper’s conclusions may not depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim. The paper’s conclusions may not depend on it. It may state two separate claims as one.
In the paper
“This corresponds to a risk management situation by all means, with a number of nuances and inherent complexity (Goodall, 2016).”
§ Ethical principles in machine learning and artificial intelligence: cases from the field and possible ways forward > Machine-learning algorithms in the field of autonomous vehicles
What the sources say
- Not checked yetcites Goodall, 2016 source
plain- Asserted: passes — The authors put it forward as true.
- Checkable: fails — It may be too vague to check against evidence.
- Load-bearing: fails — The paper’s conclusions may not depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim. It may be too vague to check against evidence. The paper’s conclusions may not depend on it.
In the paper
“Even the data on which an algorithm is trained on are not an objective truth, they are dependent upon the context in which they have been produced (Neff et al., 2017).”
§ Ethical principles in machine learning and artificial intelligence: cases from the field and possible ways forward > Discussion and conclusions
What the sources say
- Not checked yetcites Neff et al., 2017 source
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
It may state two separate claims as one.
The paper states this without a source or data behind it here, so there is nothing to check it against. The quote shows it is in the paper.
In the paper
“As to address these dimensions, value statements and guidelines have been elaborated by political and multi-stakeholder organisations.”
§ Ethical principles in machine learning and artificial intelligence: cases from the field and possible ways forward > Discussion and conclusions
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: fails — The paper’s conclusions may not depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim. The paper’s conclusions may not depend on it.
None of the sources cited for this claim have been checked yet.
In the paper
“Tools of post-normal-science inspiration for knowledge and modelling quality assessment could be adapted to the analysis of algorithms, such as the NUSAP (Numeral Unit Spread Assessment Pedigree) notation system for the management and communication of uncertainty (Funtowicz and Ravetz, 1990;Van Der Sluijs, 2005) and sensitivity auditing (Saltelli and Funtowicz, 2014), respectively.”
§ Ethical principles in machine learning and artificial intelligence: cases from the field and possible ways forward > Discussion and conclusions
What the sources say
hedgedconditional- Asserted: passes — The authors put it forward as true.
- Checkable: fails — It may be too vague to check against evidence.
- Load-bearing: fails — The paper’s conclusions may not depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim. It may be too vague to check against evidence. The paper’s conclusions may not depend on it.
The paper states this without a source or data behind it here, so there is nothing to check it against. The quote shows it is in the paper.
In the paper
“The COMPAS algorithm, developed by the private company Northpointe, attributes a 2-year recidivism-risk score to arrested people.It also evaluates the risk of violent recidivism as a score.”
§ Ethical principles in machine learning and artificial intelligence: cases from the field and possible ways forward > Points of friction between ethical dimensions.
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
It may state two separate claims as one.
In the paper
“This country faces the issue of the world's highest incarcerated population, both in absolute and per-capita terms (Brief, 2020).”
§ Ethical principles in machine learning and artificial intelligence: cases from the field and possible ways forward > Points of friction between ethical dimensions.
What the sources say
- Not checked yetcites Brief, 2020
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: fails — The paper’s conclusions may not depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim. The paper’s conclusions may not depend on it.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Coding algorithms that assure fairness in autonomous vehicles can be a very challenging issue.”
§ Ethical principles in machine learning and artificial intelligence: cases from the field and possible ways forward > Machine-learning algorithms in the field of autonomous vehicles
plain- Asserted: passes — The authors put it forward as true.
- Checkable: fails — It may be too vague to check against evidence.
- Load-bearing: fails — The paper’s conclusions may not depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim. It may be too vague to check against evidence. The paper’s conclusions may not depend on it.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Hence, no corrective action in the algorithm code may be possible at this stage, with no room for improvement in accuracy.”
§ Ethical principles in machine learning and artificial intelligence: cases from the field and possible ways forward > Machine-learning algorithms in the field of autonomous vehicles
hedged- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
It may state two separate claims as one.
None of the sources cited for this claim have been checked yet.
In the paper
“Tools for technically scrutinising the potential behaviour of an algorithm and its uncertainty already exist and could be included in the workflow of algorithm development.”
§ Ethical principles in machine learning and artificial intelligence: cases from the field and possible ways forward > Discussion and conclusions
hedged- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
It may state two separate claims as one.
In the paper
“For instance, global sensitivity analysis (Saltelli, 2008) may help in exploring how the uncertainty in the input parameters and modelling assumptions would affect the output.”
§ Ethical principles in machine learning and artificial intelligence: cases from the field and possible ways forward > Discussion and conclusions
What the sources say
- Not checked yetcites Saltelli, 2008
hedgedconditional- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: fails — The paper’s conclusions may not depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim. The paper’s conclusions may not depend on it.
None of the sources cited for this claim have been checked yet.
In the paper
“Rudin (2019) found that the algorithm seemed to behave differently from the intentions of their creators (Northpointe, 2012) with a non-linear dependence on age and a weak correlation with one's criminal history.”
§ Ethical principles in machine learning and artificial intelligence: cases from the field and possible ways forward > Points of friction between ethical dimensions.
What the sources say
- Not checked yetcites Rudin (2019)
- Not checked yetcites Northpointe, 2012
hedgedcorrelational- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
It may state two separate claims as one.
In the paper
“An exercise to appreciate the value-ladenness of these decisions is the moral-machine experiment (Massachussets Institute of Technology 2019)-a serious game where users are requested to fulfil the function of an autonomous-vehicle decision-making algorithm in a situation of danger.This experiment entails performing choices that would prioritise the safety of some categories of users over others.”
§ Ethical principles in machine learning and artificial intelligence: cases from the field and possible ways forward > Machine-learning algorithms in the field of autonomous vehicles
What the sources say
- Not checked yetcites Massachussets Institute of Technology 2019
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: fails — The paper’s conclusions may not depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim. The paper’s conclusions may not depend on it.
In the paper
“While this is based on physics and can be informed by the numerous sensors these vehicles are equipped with, unforeseen events can still play a prominent role, and profoundly affect the vehicles behaviour and reactions (Yurtsever et al., 2020).”
§ Ethical principles in machine learning and artificial intelligence: cases from the field and possible ways forward > Machine-learning algorithms in the field of autonomous vehicles
What the sources say
- Not checked yetcites Yurtsever et al., 2020
plaindirect- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
It may state two separate claims as one.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“However, the field of AI ethics is just at its infancy and it is still to be conceptualised how AI developments that encompass ethical dimensions could be attained.”
§ Ethical principles in machine learning and artificial intelligence: cases from the field and possible ways forward > Discussion and conclusions
plain- Asserted: passes — The authors put it forward as true.
- Checkable: fails — It may be too vague to check against evidence.
- Load-bearing: fails — The paper’s conclusions may not depend on it.
- Based: fails — The paper may give no reason why it holds.
Probably not a substantial claim. It may be too vague to check against evidence. The paper’s conclusions may not depend on it. The paper may give no reason why it holds. It may state two separate claims as one.
In the paper
“According to the authors of the report, the recidivism-risk was systematically overestimated for black people: the decile distribution of white defendants was skewed towards the lower end.Conversely, the decile distribution of black defendants was only slightly decreasing towards the higher end.The risk of violent recidivism within 2 years followed a similar trend.”
§ Ethical principles in machine learning and artificial intelligence: cases from the field and possible ways forward > Points of friction between ethical dimensions.
What the sources say
- Not checked yetcites Angwin and Larson, 2016
plaincorrelational- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim. It may state two separate claims as one.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Minimising the overall physical harm may be achieved by implementing an algorithm that, in the circumstance of an unavoidable collision, would target the vehicles with the highest safety standards.”
§ Ethical principles in machine learning and artificial intelligence: cases from the field and possible ways forward > Machine-learning algorithms in the field of autonomous vehicles
hedgedconditional- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: fails — The paper’s conclusions may not depend on it.
- Based: fails — The paper may give no reason why it holds.
Probably not a substantial claim. The paper’s conclusions may not depend on it. The paper may give no reason why it holds.
In the paper
“For instance, The Alan Turing Institute released a guide for responsible design and implementation of AI (Leslie, 2019) that covers the whole life-cycle of design, use, and monitoring.”
§ Ethical principles in machine learning and artificial intelligence: cases from the field and possible ways forward > Discussion and conclusions
What the sources say
- Not checked yetcites Leslie, 2019
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: fails — The paper’s conclusions may not depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim. The paper’s conclusions may not depend on it.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“One should also not forget that these algorithms are learning by direct experience and they may still end up conflicting with the initial set of ethical rules around which they have been conceived.”
§ Ethical principles in machine learning and artificial intelligence: cases from the field and possible ways forward > Machine-learning algorithms in the field of autonomous vehicles
hedgedconditional- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
It may state two separate claims as one.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Odds may emerge between the interest of the vehicle owner and passengers, on one side, and the collective interest of minimising the overall harm, on the other.”
§ Ethical principles in machine learning and artificial intelligence: cases from the field and possible ways forward > Machine-learning algorithms in the field of autonomous vehicles
hedgedconditional- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: fails — The paper’s conclusions may not depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim. The paper’s conclusions may not depend on it.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Such conflicts may exacerbate tensions, further complicating agreeing on standards and practices.”
§ Ethical principles in machine learning and artificial intelligence: cases from the field and possible ways forward > Points of friction between ethical dimensions.
hedgedconditional- Asserted: passes — The authors put it forward as true.
- Checkable: fails — It may be too vague to check against evidence.
- Load-bearing: fails — The paper’s conclusions may not depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim. It may be too vague to check against evidence. The paper’s conclusions may not depend on it.
In the paper
“An overarching meta-framework for the governance of AI in experimental technologies (i.e., robot use) has also been proposed (Rego de Almeida et al., 2020).”
§ Ethical principles in machine learning and artificial intelligence: cases from the field and possible ways forward > Discussion and conclusions
What the sources say
- Not checked yetcites Rego de Almeida et al., 2020
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: fails — The paper’s conclusions may not depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim. The paper’s conclusions may not depend on it.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“If not, transparency today may not be helpful in understanding what the system does tomorrow.”
§ Ethical principles in machine learning and artificial intelligence: cases from the field and possible ways forward > Discussion and conclusions
hedgedconditional- Asserted: passes — The authors put it forward as true.
- Checkable: fails — It may be too vague to check against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim. It may be too vague to check against evidence.
In the paper
“Trying to revert the current state of affairs may expose the first movers in the AI field to a competitive disadvantage (Morley et al., 2019).”
§ Ethical principles in machine learning and artificial intelligence: cases from the field and possible ways forward > Discussion and conclusions
What the sources say
- Not checked yetcites Morley et al., 2019
hedgedconditional- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: fails — The paper’s conclusions may not depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim. The paper’s conclusions may not depend on it.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“As are the attempt to make the process more inclusive, with a higher participation from all the stakeholders.”
§ Ethical principles in machine learning and artificial intelligence: cases from the field and possible ways forward > Discussion and conclusions
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: fails — The paper may give no reason why it holds.
Probably not a substantial claim. The paper may give no reason why it holds.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“This initiative stems from the attempt to include all the forms of governance put forth and would rest on an integrated set of feedback and interactions across dimensions and actors.”
§ Ethical principles in machine learning and artificial intelligence: cases from the field and possible ways forward > Discussion and conclusions
hedgeddirect- Asserted: passes — The authors put it forward as true.
- Checkable: fails — It may be too vague to check against evidence.
- Load-bearing: fails — The paper’s conclusions may not depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim. It may be too vague to check against evidence. The paper’s conclusions may not depend on it. It may state two separate claims as one.
In the paper
“An interesting proposal comes from Berk (2019), who asked for the intervention of super partes authorities to define standards of transparency, accuracy and fairness for algorithm developers in line with the role of the Food and Drug administration in the US and other regulation bodies.”
§ Ethical principles in machine learning and artificial intelligence: cases from the field and possible ways forward > Discussion and conclusions
What the sources say
- Not checked yetcites Berk, 2019
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: fails — The paper’s conclusions may not depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim. The paper’s conclusions may not depend on it.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“If yes, the algorithm could become defective.”
§ Ethical principles in machine learning and artificial intelligence: cases from the field and possible ways forward > Discussion and conclusions
hedgedconditional- Asserted: passes — The authors put it forward as true.
- Checkable: fails — It may be too vague to check against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: fails — The paper may give no reason why it holds.
Probably not a substantial claim. It may be too vague to check against evidence. The paper may give no reason why it holds.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Attempts to expose to public scrutiny the algorithms developed are yet scant.”
§ Ethical principles in machine learning and artificial intelligence: cases from the field and possible ways forward > Discussion and conclusions
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: fails — The paper may give no reason why it holds.
Probably not a substantial claim. The paper may give no reason why it holds.
In the paper
“The post-normal-science concept of extended peer communities could assist also in this endeavour (Funtowicz and Ravetz, 1997).”
§ Ethical principles in machine learning and artificial intelligence: cases from the field and possible ways forward > Discussion and conclusions
What the sources say
- Not checked yetcites Funtowicz and Ravetz, 1997
hedgedconditional- Asserted: passes — The authors put it forward as true.
- Checkable: fails — It may be too vague to check against evidence.
- Load-bearing: fails — The paper’s conclusions may not depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim. It may be too vague to check against evidence. The paper’s conclusions may not depend on it.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“The development pace of new algorithms would be necessarily reduced so as to comply with the standards defined and the required clearance processes.”
§ Ethical principles in machine learning and artificial intelligence: cases from the field and possible ways forward > Discussion and conclusions
hedgeddirect- Asserted: passes — The authors put it forward as true.
- Checkable: fails — It may be too vague to check against evidence.
- Load-bearing: fails — The paper’s conclusions may not depend on it.
- Based: fails — The paper may give no reason why it holds.
Probably not a substantial claim. It may be too vague to check against evidence. The paper’s conclusions may not depend on it. The paper may give no reason why it holds.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“The development of decision-making algorithms remains quite obscure in spite of the concerns raised and the intentions manifested to address them.”
§ Ethical principles in machine learning and artificial intelligence: cases from the field and possible ways forward > Discussion and conclusions
plain- Asserted: passes — The authors put it forward as true.
- Checkable: fails — It may be too vague to check against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: fails — The paper may give no reason why it holds.
Probably not a substantial claim. It may be too vague to check against evidence. The paper may give no reason why it holds.
In the paper
“Example-based explanations (Molnar, 2020) may also contribute to an effective engagement of all the parties by helping in bridging technical divides across developers, experts in other fields, and lay-people.”
§ Ethical principles in machine learning and artificial intelligence: cases from the field and possible ways forward > Discussion and conclusions
What the sources say
- Not checked yetcites Molnar, 2020
hedgedconditional- Asserted: passes — The authors put it forward as true.
- Checkable: fails — It may be too vague to check against evidence.
- Load-bearing: fails — The paper’s conclusions may not depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim. It may be too vague to check against evidence. The paper’s conclusions may not depend on it.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“A shared regulation could help in tackling the potential competitive disadvantage a first mover may suffer.”
§ Ethical principles in machine learning and artificial intelligence: cases from the field and possible ways forward > Discussion and conclusions
hedgedconditional- Asserted: passes — The authors put it forward as true.
- Checkable: fails — It may be too vague to check against evidence.
- Load-bearing: fails — The paper’s conclusions may not depend on it.
- Based: fails — The paper may give no reason why it holds.
Probably not a substantial claim. It may be too vague to check against evidence. The paper’s conclusions may not depend on it. The paper may give no reason why it holds.
The paper states this without a source or data behind it here, so there is nothing to check it against. The quote shows it is in the paper.
In the paper
“As noted by a perceptive reviewer, ML systems that keep learning are dangerous and hard to understand because they can quickly change.”
§ Ethical principles in machine learning and artificial intelligence: cases from the field and possible ways forward > Discussion and conclusions
plaindirect- Asserted: fails — The authors may not put it forward as true.
- Checkable: fails — It may be too vague to check against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: fails — The paper may give no reason why it holds.
Probably not a substantial claim. The authors may not put it forward as true. It may be too vague to check against evidence. The paper may give no reason why it holds. It may state two separate claims as one.
In the paper
“Greene et al. (2019) found seven common core elements across these documents: (i) design's moral background (universal concerns, objectively measured); (ii) expert oversight; (iii) valuesdriven determinism; (iv) design as locus of ethical scrutiny; (v) better building; (vi) stakeholder-driven legitimacy; and, (vii) machine translation.”
§ Ethical principles in machine learning and artificial intelligence: cases from the field and possible ways forward > Discussion and conclusions
What the sources say
- Not checked yetcites Greene et al., 2019
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim. It may state two separate claims as one.
In the paper
“Hence, the development process of the algorithm cannot be perfect in this setting, one has to be open to negotiation and unavoidably work with imperfections and clumsiness (Ravetz, 1987).”
§ Ethical principles in machine learning and artificial intelligence: cases from the field and possible ways forward > Discussion and conclusions
What the sources say
- Not checked yetcites Ravetz, 1987 source
plain- Asserted: passes — The authors put it forward as true.
- Checkable: fails — It may be too vague to check against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim. It may be too vague to check against evidence.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“One should also not forget that points of friction across ethical dimensions may emerge, e.g., between transparency and accountability, or accuracy and fairness as highlighted in the case studies.”
§ Ethical principles in machine learning and artificial intelligence: cases from the field and possible ways forward > Discussion and conclusions
hedgedconditional- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“This would prevent the algorithmlearning process from conflicting with the standards agreed.”
§ Ethical principles in machine learning and artificial intelligence: cases from the field and possible ways forward > Discussion and conclusions
hedgeddirect- Asserted: passes — The authors put it forward as true.
- Checkable: fails — It may be too vague to check against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: fails — The paper may give no reason why it holds.
Probably not a substantial claim. It may be too vague to check against evidence. The paper may give no reason why it holds.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Identifying a relevant pool of social actors may require an important effort in terms of stakeholders' mapping so as to assure a complete, but also effective, governance in terms of number of participants and simplicity of working procedures.”
§ Ethical principles in machine learning and artificial intelligence: cases from the field and possible ways forward > Discussion and conclusions
hedgedconditional- Asserted: passes — The authors put it forward as true.
- Checkable: fails — It may be too vague to check against evidence.
- Load-bearing: fails — The paper’s conclusions may not depend on it.
- Based: fails — The paper may give no reason why it holds.
Probably not a substantial claim. It may be too vague to check against evidence. The paper’s conclusions may not depend on it. The paper may give no reason why it holds.
In the paper
“In this setting, seeking algorithm transparency would not be harmful for their developers as scrutiny would be delegated to entrusted intermediate parties, to take place behind closed doors (de Laat, 2018).”
§ Ethical principles in machine learning and artificial intelligence: cases from the field and possible ways forward > Discussion and conclusions
What the sources say
- Not checked yetcites de Laat, 2018 source
hedged- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: fails — The paper’s conclusions may not depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim. The paper’s conclusions may not depend on it.
The paper relies on this without support.
Decision-making on numerous aspects of daily lives is being outsourced to ML algorithms and AI, motivated by speed and efficiency in the decision process.
“Decision-making on numerous aspects of our daily lives is being outsourced to machinelearning (ML) algorithms and artificial intelligence (AI), motivated by speed and efficiency in the decision process.”
The paper relies on this without support.
ML approaches are typically developed as black boxes, so their code scripts are rarely scrutinised and interpretability is usually sacrificed in favour of usability and effectiveness.
“ML approaches-one of the typologies of algorithms underpinning artificial intelligence-are typically developed as black boxes.The implication is that ML code scripts are rarely scrutinised; interpretability is usually sacrificed in favour of usability and effectiveness.”
The paper relies on this without support.
Room for improvement in practices associated with programme development has been flagged along dimensions including fairness, accuracy, accountability, and transparency.
“Room for improvement in practices associated with programme development have also been flagged along other dimensions, including inter alia fairness, accuracy, accountability, and transparency.”
The paper relies on this without support.
Developers of algorithms may not be capable of explaining in plain language how a given tool works and what functional elements it is based on.
“Moreover, developers of algorithm may not be capable of explaining in plain language how a given tool works and what functional elements it is based on.”
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?
The source backs part of the claim, not all of it — or backs it for a different population, condition or comparison.
In the citing paper
“Predictive systems which are powered by algorithms jeopardize students and teachers’ autonomy and their ability to govern their own life [46, 47].”
From Artificial intelligence in education: Addressing ethical challenges in K-12 settings (2022) · cites this paper as [46]
Experimental — accuracy not yet measuredIn the citing paper
“Given the nascent stage of AI ethics, there is a need to devise frameworks for AI advancements to incorporate ethical considerations (Lo Piano, 2020).”
From Navigating the perils of artificial intelligence: a focused review on ChatGPT and responsible research and innovation (2024) · cites this paper as Lo Piano, 2020
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.
Rated minor: it would not touch a result — a value no result rests on, a p-value that stays on the same side of the significance level, or a conflict that is clerical, resolved elsewhere in the paper, or outside its own findings.
In the paper
§ Ethical principles in machine learning and artificial intelligence: cases from the field and possible ways forward > Introduction
“This simulation may limit itself to some simple predictable features, thus limiting human complexity (Cowls, 2019).”
§ Ethical principles in machine learning and artificial intelligence: cases from the field and possible ways forward > References
“13. Cowls J (2020) Deciding how to decide: six key questions for reducing AI's democratic deficit.”
Rated minor: it would not touch a result — a value no result rests on, a p-value that stays on the same side of the significance level, or a conflict that is clerical, resolved elsewhere in the paper, or outside its own findings.
In the paper
§ Ethical principles in machine learning and artificial intelligence: cases from the field and possible ways forward > Guidelines and secondary literature on AI ethics, its dimensions and stakes
“Jobin et al. (2019) reviewed 84 documents”
§ Ethical principles in machine learning and artificial intelligence: cases from the field and possible ways forward > References
“31. Jobi A, Ienca M, Vayena E (2019) Artificial intelligence: the global landscape of ethics guidelines.”
Rated minor: it would not touch a result — a value no result rests on, a p-value that stays on the same side of the significance level, or a conflict that is clerical, resolved elsewhere in the paper, or outside its own findings.
In the paper
§ Ethical principles in machine learning and artificial intelligence: cases from the field and possible ways forward > Guidelines and secondary literature on AI ethics, its dimensions and stakes
“further action has been recently demanded from the European Parliament (De Sutter, 2019).”
§ Ethical principles in machine learning and artificial intelligence: cases from the field and possible ways forward > References
“16. De Sutter P (2020) Automated decision-making processes: ensuring consumer protection, and free movement of goods and services.”
Citations
Cites
Artificial intelligence: a modern approach
Kumar, D. A., Balasubramanian, S., Byeon, D. H., & Manerkar, P. G. V. (1995) · Choice Reviews Online · doi:10.5860/choice.33-1577 · Cited by 22,191
Rudin, C. (2019) · Nature Machine Intelligence · doi:10.1038/s42256-019-0048-x · Cited by 10,561 · cited 5×
“Rudin (2019) argued that the community of algorithm developers should go beyond explaining black-box models by developing interpretable models in the fi rst place.”
BackgroundMethodResultInfluentialGlobal Sensitivity Analysis. The Primer
Saltelli, A., Ratto, M., Andres, T., Campolongo, F., Cariboni, J., Gatelli, D., Saisana, M., & Tarantola, S. (2007) · doi:10.1002/9780470725184 · Cited by 6,430
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 4×
“Jobin et al. (2019) reviewed 84 documents, which were produced by several actors of the fi eld, almost half of which from private companies or governmental agencies.”
InfluentialSome Studies in Machine Learning Using the Game of Checkers
Samuel, A. L. (1959) · IBM Journal of Research and Development · doi:10.1147/rd.33.0210 · Cited by 4,443 · cited 2×
“The strength of ML resides in its capacity to learn from data without need to be explicitly programmed (Samuel, 1959); ML algorithms are autonomous and self-sufficient when performing their learning function.”
BackgroundThe black swan: the impact of the highly improbable
(2007) · Choice Reviews Online · doi:10.5860/choice.45-1430 · Cited by 3,483
Weapons of Math Destruction: How Big Data Increases Inequality and Threatens Democracy
Doyle, T. (2017) · The Information Society · doi:10.1080/01972243.2017.1354593 · Cited by 3,111
Big Data and Machine Learning in Health Care
Beam, A. L., & Kohane, I. S. (2018) · JAMA · doi:10.1001/jama.2017.18391 · Cited by 2,506 · cited 3×
“However, attaining this sort of algorithmic fairness would imply inequality of treatment across genders (Berk et al., 2018). Fairness could be further hampered by the combined use of this algorithm with others driving decisions on neighbourhood police patrolling. The fact these algorithms may be prone to drive further…”
BackgroundThe Black Swan: The Impact of the Highly Improbable
(2008) · Library Review · doi:10.1108/00242530810854080 · Cited by 2,465
(2016) · Cited by 2,334
Raji, I. D., Smart, A., White, R. N., Mitchell, M., Gebru, T., Hutchinson, B., Smith-Loud, J., Theron, D., & Barnes, P. (2020) · Proceedings of the 2020 Conference on Fairness, Accountability, and Transparency · doi:10.1145/3351095.3372873 · Cited by 2,245 · cited 1×
“Raji et al. (2020) suggest that a process of algorithmic auditing within the software-development company could help in tackling some of the ethical issues raised.”
BackgroundPrinciples alone cannot guarantee ethical AI
Mittelstadt, B. (2019) · Nature Machine Intelligence · doi:10.1038/s42256-019-0114-4 · Cited by 1,886 · cited 1×
“Seven of the most prominent value statements from the AI/ML fi elds were examined in Greene et al. (2019 Mittelstadt (2019) critically analysed the current debate and actions in the fi eld of AI ethics and noted that the dimensions addressed in AI ethics are converging towards those of medical ethics.”
BackgroundAnanny, M., & Crawford, K. (2018) · New Media & Society · doi:10.1177/1461444816676645 · Cited by 1,836 · cited 7×
“Ananny and Crawford (2018) comment that resorting to full algorithmic transparency may not be an adequate means to address their ethical dimensions; opening up the black-box would not suffice to disclose their modus operandi. Moreover, developers of algorithm may not be capable of explaining in plain language how a gi…”
BackgroundInfluentialThe Human Use of Human Beings: Cybernetics and Society
Miles, S. B., & Wiener, N. (1951) · Land Economics · doi:10.2307/3159747 · Cited by 1,664
A Unified Framework of Five Principles for AI in Society
Floridi, L., & Cowls, J. (2019) · Harvard Data Science Review · doi:10.1162/99608f92.8cd550d1 · Cited by 1,631 · cited 4×
“According to Floridi and Cowls (2019) a prominent role is also played by the autonomy dimension; the possibility of refraining from ceding decision power to AI for overriding reasons (e.g., the gain of ef fi cacy is not deemed fi t to justify the loss of control over decision-making).”
BackgroundInfluentialUncertainty and Quality in Science for Policy
Funtowicz, S. O., & Ravetz, J. R. (1990) · doi:10.1007/978-94-009-0621-1 · Cited by 1,575 · cited 2×
“Tools of post-normal-science inspiration for knowledge and modelling quality assessment could be adapted to the analysis of algorithms, such as the NUSAP (Numeral Unit Spread Assessment Pedigree) notation system for the management and communication of uncertainty (Funtowicz and Ravetz, 1990; Van Der Sluijs, 2005) and…”
MethodEvidence, Argument, and Persuasion in the Policy Process.
MacRae, D., & Majone, G. (1990) · Contemporary Sociology A Journal of Reviews · doi:10.2307/2073498 · Cited by 1,210
A philosophical essay on probabilities
Laplace, P. S., & Dale, A. I. (1952) · Journal of the Franklin Institute · doi:10.1016/0016-0032(52)90656-x · Cited by 1,140
Interpretable Machine Learning
Molnar, C. (2020) · Cited by 1,104
The accuracy, fairness, and limits of predicting recidivism
Dressel, J., & Farid, H. (2018) · Science Advances · doi:10.1126/sciadv.aao5580 · Cited by 1,071 · cited 1×
“Dressel and Farid (2018) achieved this result by using a linear predictor-logistic regressor that made use of only two variables (age and total number of previous convic-tions of the subject).”
MethodLife itself: a comprehensive inquiry into the nature, origin, and fabrication of life
(1992) · Choice Reviews Online · doi:10.5860/choice.29-4505 · Cited by 1,043 · cited 1×
“A decision-making algorithm will always be based on a formal system, which is a representation of a real system (Rosen, 2005).”
BackgroundFairness in Criminal Justice Risk Assessments: The State of the Art
Berk, R., Heidari, H., Jabbari, S., Kearns, M., & Roth, A. (2021) · Sociological Methods & Research · doi:10.1177/0049124118782533 · Cited by 988 · cited 1×
“However, attaining this sort of algorithmic fairness would imply inequality of treatment across genders (Berk et al., 2018).”
BackgroundMorley, J., Floridi, L., Kinsey, L., & Elhalal, A. (2020) · Science and Engineering Ethics · doi:10.1007/s11948-019-00165-5 · Cited by 882
Wallach, W., & Allen, C. (2009) · Oxford University Press · doi:10.1093/acprof:oso/9780195374049.001.0001 · Cited by 711 · cited 2×
“Furthermore, the way decisions enacted could back fi re in complex contexts to which the algorithms had no extrapolative power, is an unpredictable issue one has to deal with (Wallach and Allen, 2008; Yurtsever et al., 2020).”
BackgroundBahrammirzaee, A. (2010) · Neural Computing and Applications · doi:10.1007/s00521-010-0362-z · Cited by 565 · cited 1×
“…elds, such as (precision) agriculture (Sennaar, 2019), air combat and military training (Gallagher, 2016; Wong, 2020), education (Sears, 2018), fi nance (Bahrammirzaee, 2010), health care (Beam and Kohane, 2018), human resources and recruiting (Hmoud and Laszlo, 2019), music composition (Cheng,…”
BackgroundVan Der Sluijs, J. P., Craye, M., Funtowicz, S., Kloprogge, P., Ravetz, J., & Risbey, J. (2005) · Risk Analysis · doi:10.1111/j.1539-6924.2005.00604.x · Cited by 426
Explanations as Mechanisms for Supporting Algorithmic Transparency
Rader, E., Cotter, K., & Cho, J. (2018) · Proceedings of the 2018 CHI Conference on Human Factors in Computing Systems · doi:10.1145/3173574.3173677 · Cited by 388 · cited 2×
“, 2019), autonomous vehicles and traffic management (Ye, 2018), social-media news-feed (Rader et al., 2018), work scheduling and optimisation (O’Neil, 2016), and several others.”
BackgroundGreene, D., Hoffmann, A. L., & Stark, L. (2019) · Proceedings of the Annual Hawaii International Conference on System Sciences · doi:10.24251/hicss.2019.258 · Cited by 386
The global landscape of AI ethics guidelines
Jobin, A., Ienca, M., & Vayena, E. (2019) · Repository for Publications and Research Data (ETH Zurich) · doi:10.3929/ethz-b-000375277 · Cited by 360
Accountability in a computerized society
Nissenbaum, H. (1996) · Science and Engineering Ethics · doi:10.1007/bf02639315 · Cited by 346 · cited 1×
“Attributing responsibilities in complicated projects where many parties and developers may be involved, an issue known as the problem of many hands (Nissenbaum, 1996), may indeed be very difficult.”
BackgroundUnderstanding artificial intelligence ethics and safetyPreprint
Leslie, D. (2019) · arXiv (Cornell University) · doi:10.48550/arxiv.1906.05684 · Cited by 308 · cited 1×
“For instance, The Alan Turing Institute released a guide for responsible design and implementation of AI (Leslie, 2019) that covers the whole life-cycle of design, use, and monitoring.”
MethodMachine learning for vehicular networks: Recent advances and application examples
Ye, H., Liang, L., Ye Li, G., Kim, J., Lu, L., & Wu, M. (2018) · IEEE Vehicular Technology Magazine · doi:10.1109/mvt.2018.2811185 · Cited by 303
Ethical Decision Making during Automated Vehicle Crashes
Goodall, N. J. (2014) · Transportation Research Record: Journal of the Transportation Research Board · doi:10.3141/2424-07 · Cited by 292 · cited 3×
“Transportation Res Rec: J Transportation Res Board 2424:58–65 Goodall NJ (2016) Away from trolley problems and toward risk management.”
BackgroundInfluentialde Laat, P. B. (2018) · Philosophy & Technology · doi:10.1007/s13347-017-0293-z · Cited by 251 · cited 3×
“In this setting, seeking algorithm transparency would not be harmful for their developers as scrutiny would be delegated to entrusted intermediate parties, to take place behind closed doors (de Laat, 2018).”
BackgroundInfluentialArtificial Intelligence Aided Automated Design for Reliability of Power Electronic Systems
Dragičević, T., Wheeler, P. W., & Blaabjerg, F. (2018) · IEEE Transactions on Power Electronics · doi:10.1109/tpel.2018.2883947 · Cited by 219 · cited 1×
“…recruiting (Hmoud and Laszlo, 2019), music composition (Cheng, 2009/09), customer service (Kongthon et al., 2009), reliable engineering and maintenance (Dragicevic et al., 2019), autonomous vehicles and traf fi c management (Ye, 2018), social-media news-feed (Rader et al., 2018), work scheduling…”
BackgroundBuilding Ethics into Artificial Intelligence
Yu, H., Shen, Z., Miao, C., Leung, C., Lesser, V. R., & Yang, Q. (2018) · Proceedings of the Twenty-Seventh International Joint Conference on Artificial Intelligence · doi:10.24963/ijcai.2018/779 · Cited by 210 · cited 1×
“While re fl ections around the ethical implications of machines and automation deployment were already put forth in the ’ 50s and ’ 60s (Samuel, 1959; Wiener, 1988), the increasing use of AI in many fi elds raises new important questions about its suitability (Yu et al., 2018).”
BackgroundUsable Knowledge, Usable Ignorance: Incomplete Science with Policy Implications
Ravetz, J. R. (1987) · Knowledge · doi:10.1177/107554708700900104 · Cited by 185 · cited 3×
“Funtowicz S, Ravetz J (1997) Environmental problems, post-normal science, and extended peer communities.”
BackgroundInfluentialRaji, I. D., Smart, A., White, R. N., Mitchell, M., Gebru, T., Hutchinson, B., Smith-Loud, J., Theron, D., & Barnes, P. (2020) · arXiv · doi:10.48550/arxiv.2001.00973 · Cited by 179 · cited 1×
“Raji et al. (2020) suggest that a process of algorithmic auditing within the software-development company could help in tackling some of the ethical issues raised.”
BackgroundThe Need for Open Source Software in Machine Learning
Sonnenburg, S., Braun, M. L., Ong, C. S., Samy Bengio, A., Bottou, L., Holmes, G., LeCun, Y., Müller, K., Pereira, F. M. B., Rasmussen, C. E., et al. (2007) · Publikationsdatenbank der Fraunhofer-Gesellschaft (Fraunhofer-Gesellschaft) · Cited by 178
Neff, G., Tanweer, A., Fiore-Gartland, B., & Osburn, L. (2017) · Big Data · doi:10.1089/big.2016.0050 · Cited by 173 · cited 1×
“Even the data on which an algorithm is trained on are not an objective truth, they are dependent upon the context in which they have been produced (Neff et al., 2017).”
BackgroundEnslaving the Algorithm: From a “Right to an Explanation” to a “Right to Better Decisions”?
Edwards, L., & Veale, M. (2018) · IEEE Security & Privacy · doi:10.1109/msp.2018.2701152 · Cited by 141 · cited 2×
“The France’s Digital Republic Act gives the right to an explanation as regards decisions on an individual made through the use of administrative algorithms (Edwards and Veale, 2018).”
BackgroundEthics as Methods: Doing Ethics in the Era of Big Data Research—Introduction
Markham, A. N., Tiidenberg, K., & Herman, A. C. (2018) · Social Media + Society · doi:10.1177/2056305118784502 · Cited by 127 · cited 1×
“This aspect would represent a further open issue to be taken into account in their development (Markham et al., 2018).”
BackgroundAway from Trolley Problems and Toward Risk Management
Goodall, N. J. (2016) · Applied Artificial Intelligence · doi:10.1080/08839514.2016.1229922 · Cited by 124 · cited 1×
“This corresponds to a risk management situation by all means, with a number of nuances and inherent complexity (Goodall, 2016).”
BackgroundMorley, J., Floridi, L., Kinsey, L., & Elhalal, A. (2019) · SSRN Electronic Journal · doi:10.2139/ssrn.3830348 · Cited by 120
A short comment on statistical versus mathematical modelling
Saltelli, A. (2019) · Nature Communications · doi:10.1038/s41467-019-11865-8 · Cited by 108 · cited 1×
“Decision-making-based algorithms rest inevitably on assumptions, even silent ones, such as the quality of data the algorithm is trained on (Saltelli and Funtowicz, 2014), or the actual modelling relations adopted (Hoerl, 2019), with all the implied consequences (Saltelli, 2019).”
BackgroundLearning Certifiably Optimal Rule Lists for Categorical Data
Angelino, E., Larus-Stone, N., Alabi, D., Seltzer, M., & Rudin, C. (2017) · arXiv · doi:10.48550/arxiv.1704.01701 · Cited by 108 · cited 3×
“These exercises (Rudin, 2019; Angelino et al., 2018) showed that it is possible to implement interpretable classification algorithms that lead to a similar accuracy as COMPAS.”
MethodResultElli, A. S. T., & Funtowicz, S. O. (2014) · Cited by 106
(2017) · doi:10.1145/3097983.3098047 · Cited by 104 · cited 3×
“These exercises (Rudin, 2019; Angelino et al., 2018) showed that it is possible to implement interpretable classification algorithms that lead to a similar accuracy as COMPAS.”
MethodResult(2019) · RePEc: Research Papers in Economics · Cited by 102
Artificial Intelligence, Governance and Ethics: Global PerspectivesPreprint
Daly, A., Hagendorff, T., Li, H., Mann, M., Marda, V., Wagner, B., Wang, W., & Witteborn, S. (2019) · SSRN Electronic Journal · doi:10.2139/ssrn.3414805 · Cited by 91
Governance by Numbers. The Making of a Legal Model of Allegiance
Goldoni, M. (2018) · International Journal of Constitutional Law · doi:10.1093/icon/moy079 · Cited by 90
Life itself. A comprehensive inquiry into the nature, origin, and fabrication of life
Kampis, G. (1995) · Biosystems · doi:10.1016/0303-2647(95)90013-6 · Cited by 87
Morley, J., Floridi, L., Kinsey, L., & Elhalal, A. (2019) · arXiv.org · doi:10.48550/arxiv.1905.06876 · Cited by 76
Bonnefon, J.-F., Shariff, A., & Rahwan, I. (2019) · Proceedings of the IEEE · doi:10.1109/jproc.2019.2897447 · Cited by 74 · cited 2×
“In practice, the issue would be framed by the algorithm in terms of a statistical trolley dilemma in the words of Bonnefon et al. (2019), whereby the risk of harm for some road users will be increased.”
BackgroundMachine Learning Risk Assessments in Criminal Justice Settings
Berk, R. (2019) · doi:10.1007/978-3-030-02272-3 · Cited by 55 · cited 6×
“JAMA 319:1317 Berk R (2019) Machine learning risk assessments in criminal justice settings.”
BackgroundInfluential“The Computer Said So”: On the Ethics, Effectiveness, and Cultural Techniques of Predictive Policing
Karppi, T. (2018) · Social Media + Society · doi:10.1177/2056305118768296 · Cited by 54 · cited 3×
“The fact these algorithms may be prone to drive further patrolling in poor neighbourhoods may result from a training bias as crimes occurring in public tend to be more frequently reported (Karppi, 2018).”
BackgroundInfluentialEnslaving the Algorithm: From a 'Right to an Explanation' to a 'Right to Better Decisions'?
Edwards, L., & Veale, M. (2018) · Strathprints: The University of Strathclyde institutional repository (University of Strathclyde) · Cited by 53
Ethics of quantification: illumination, obfuscation and performative legitimation
Sareen, S., Saltelli, A., & Rommetveit, K. (2020) · Palgrave Communications · doi:10.1057/s41599-020-0396-5 · Cited by 45 · cited 2×
“In all these fields, an increasing amount of functions are being ceded to algorithms to the detriment of human control, raising concern for loss of fairness and equitability (Sareen et al., 2020).”
BackgroundEthics of quantification or quantification of ethics?
Saltelli, A. (2019) · Futures · doi:10.1016/j.futures.2019.102509 · Cited by 40 · cited 1×
“A fundamental aspect is how and to what extent the values and the perspectives of the involved stakeholders have been taken care of in the design of the decision-making algorithm (Saltelli, 2020).”
BackgroundThe Financial Modelers' ManifestoPreprint
Derman, E., & Wilmott, P. (2009) · SSRN Electronic Journal · doi:10.2139/ssrn.1324878 · Cited by 36 · cited 1×
“An example comes from the fi eld of fi nancial modelling, with a manifesto elaborated in the aftermath of the 2008 fi nancial crisis (Derman and Wilmott, 2009).”
MethodIntegrating Deep Reinforcement Learning with Model-based Path Planners for Automated Driving
Yurtsever, E., Capito, L., Redmill, K., & Ozgune, U. (2020) · IEEE Intelligent Vehicles Symposium · doi:10.1109/iv47402.2020.9304735 · Cited by 36 · cited 2×
“Furthermore, the way decisions enacted could back fi re in complex contexts to which the algorithms had no extrapolative power, is an unpredictable issue one has to deal with (Wallach and Allen, 2008; Yurtsever et al., 2020).”
BackgroundArtificial Intelligence Governance and Ethics : Global Perspectives
Daly, A., Hagendorff, T., Hui, L., Mann, M., Marda, V., Wagner, B., Wang, W., & Witteborn, S. (2019) · Strathprints: The University of Strathclyde institutional repository (University of Strathclyde) · Cited by 33
Explaining code for publication
Thimbleby, H. (2003) · Software Practice and Experience · doi:10.1002/spe.537 · Cited by 30 · cited 1×
“On a larger scale, the use of open-source software in the context of ML applications has already been advocated for over a decade (Thimbleby, 2003) with an indirect call for tools to execute more interpretable and reproducible programming such as Jupyter Notebooks , available from 2015 onwards.”
MethodGovernance by Numbers : The Making of a Legal Model of Allegiance
(2017) · Hart Publishing eBooks · doi:10.5040/9781509907779 · Cited by 25 · cited 1×
“Some authors are pessimistic, such as Supiot (2017) who speaks of governance by numbers , where quanti fi cation is replacing the traditional decision-making system and profoundly affecting the pillar of equality of judgement.”
BackgroundImplementing an online help desk system based on conversational agent
Kongthon, A., Sangkeettrakarn, C., Kongyoung, S., & Haruechaiyasak, C. (2009) · International Conference on Management of Emergent Digital EcoSystems (MEDES) · doi:10.1145/1643823.1643908 · Cited by 24 · cited 2×
“Applications in our daily lives encompass fields, such as (precision) agriculture (Sennaar, 2019), air combat and military training (Gallagher, 2016; Wong, 2020), education (Sears, 2018), finance (Bahrammirzaee, 2010), health care (Beam and Kohane, 2018), human resources and recruiting (Hmoud and Laszlo, 2019), music…”
BackgroundThe Chinese Approach to Artificial Intelligence: An Analysis of Policy and RegulationPreprint
Roberts, H., Cowls, J., Morley, J., Taddeo, M., Wang, V., & Floridi, L. (2019) · SSRN Electronic Journal · doi:10.2139/ssrn.3469784 · Cited by 18 · cited 1×
“In this context, China has also been allocating efforts on privacy and data protection (Roberts, 2019).”
BackgroundFaye, B., Grelet, Y., & Landais, É. (1991) · HAL (Le Centre pour la Communication Scientifique Directe) · Cited by 15
Artificial Intelligence Regulation: A Meta-Framework for Formulation and Governance
de Almeida, P. G. R., dos Santos, C. D., & Farias, J. (2020) · Hawaii International Conference on System Sciences · doi:10.24251/hicss.2020.647 · Cited by 13 · cited 1×
“Anoverarching meta-framework for the governance of AI in experimental technologies (i.e., robot use) has also been proposed (Rego de Almeida et al., 2020).”
BackgroundThe Explanation Game: A Formal Framework for Interpretable Machine LearningPreprint
Watson, D., & Floridi, L. (2020) · SSRN Electronic Journal · doi:10.2139/ssrn.3509737 · Cited by 7
Deciding How to Decide: Six Key Questions for Reducing AI’s Democratic Deficit
Cowls, J. (2020) · Digital ethics lab yearbook · doi:10.1007/978-3-030-29145-7_7 · Cited by 4 · cited 6×
“(Floridi and Cowls, 2019): beneficence, non-maleficence, autonomy, justice and, explicability. The latter is a new dimension specifically acknowledged in the case of AI, while the others were already identified in the controversial domain of bioethics. Jobin et al. (2019) reviewed 84 documents, which were produced by…”
BackgroundInfluential
Research works printed but not linked: 2
- Funtowicz S, Ravetz J (1997) Environmental problems, post-normal science, and extended peer communities. Études et Recherches sur les Systémes Agraires et le Développement. INRA Editions. pp. 169–175
- McCarthy J, Minsky ML, Rochester N, Shannon CE (2006) A proposal for the dartmouth summer research project on artificial intelligence, August 31, 1955. AI Mag 27:12–12
Cited by
Sorted by citations
Artificial intelligence in education: Addressing ethical challenges in K-12 settings
Akgun, S., & Greenhow, C. (2022) · AI and Ethics · doi:10.1007/s43681-021-00096-7 · Cited by 1,473
“Predictive systems which are powered by algorithms jeopardize students and teachers’ autonomy and their ability to govern their own life [46, 47].”
BackgroundInterpretable machine learning: Fundamental principles and 10 grand challenges
Rudin, C., Chen, C., Chen, Z., Huang, H., Semenova, L., & Zhong, C. (2022) · Statistics Surveys · doi:10.1214/21-ss133 · Cited by 958
The promise of machine learning in predicting treatment outcomes in psychiatry
Chekroud, A. M., Bondar, J., Delgadillo, J., Doherty, G., Wasil, A., Fokkema, M., Cohen, Z., Belgrave, D., DeRubeis, R., Iniesta, R., et al. (2021) · World Psychiatry · doi:10.1002/wps.20882 · Cited by 531
Governance of artificial intelligence
Taeihagh, A. (2021) · Policy and Society · doi:10.1080/14494035.2021.1928377 · Cited by 562
“…ethical principles arising from a particular AI application, conflicts between ethical principles, and the justification of their importance have critical implications on the resulting trade-offs from designing AI systems and the public’s compliance with different ethical guidelines (Piano, 2020).”
BackgroundInfluentialSharifi, A., Allam, Z., Bibri, S. E., & Khavarian-Garmsir, A. R. (2023) · Cities · doi:10.1016/j.cities.2023.104659 · Cited by 387
Metadata
- Authors
- Lo Piano, S.
- ORCID
- Lo Piano, Samuele
- Venue
- Humanities and Social Sciences Communications
- Published
- 2020-06-17
- Volume
- 7
- Issue
- 1
- ISSN
- 2055-1045 (ISSN-L)
- 2662-9992
- OpenAlex
- W3029725182
- Semantic Scholar
- a5bd811e5358e477c079278ee2a0066dc81a386f
- Cited by
- 229
- References
- 69
- Citations by year
- 2026: 18 · 2025: 39 · 2024: 44 · 2023: 45 · 2022: 36 · 2021: 40 · 2020: 7
- FWCI
- 43.61
- Type
- journal article
- Language
- en
- Open access
- gold
- doi.org
- Locations
- Humanities and Social Sciences Communications (publishedVersion, cc-by)
- CentAUR (University of Reading) (submittedVersion)
- RePEc: Research Papers in Economics (submittedVersion)
- DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion, cc-by-sa)
- Funders
- Universitetet i Bergen
- Topics
- Physical Sciences › Computer Science › Artificial Intelligence › Ethics and Social Impacts of AI
- Physical Sciences › Computer Science › Artificial Intelligence › Explainable Artificial Intelligence (XAI)
- Social Sciences › Law › Digitalization, Law, and Regulation
- Keywords
- ethical principles, machine learning, artificial intelligence, algorithmic decision-making, black-box algorithms, interpretability, algorithmic fairness, accountability, transparency, AI governance, autonomous vehicles, AI guidelines
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.
A language model drew the claims from the paper on 2026-10-06, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).
A second model labelled each claim on 2026-10-06 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).
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
