Algorithmic Accountability and Public Reason
Open accessThe DOI matches its Crossref record; 61 of the 61 references it cites were found in a registry or an index; 4 of the 35 sources its claims cite were checked for support.
Abstract
The ever-increasing application of algorithms to decision-making in a range of social contexts has prompted demands for algorithmic accountability. Accountable decision-makers must provide their decision-subjects with justifications for their automated system's outputs, but what kinds of broader principles should we expect such justifications to appeal to? Drawing from political philosophy, I present an account of algorithmic accountability in terms of the democratic ideal of 'public reason'. I argue that situating demands for algorithmic accountability within this justificatory framework enables us to better articulate their purpose and assess the adequacy of efforts toward them.
Overview
CoChat Summary
TL;DR
Binns argues that algorithmic accountability should be understood through the democratic ideal of public reason, requiring decision-makers to justify automated decisions using epistemic and normative standards acceptable to all reasonable people. This framework helps address conflicts arising from reasonable pluralism and assess whether accountability efforts are adequate.
Key findings
- Algorithmic decision-making necessarily embeds contestable epistemic and normative assumptions, so accountability should draw out those assumptions rather than merely provide technical explanations.
- Because decision-makers and decision-subjects may reasonably disagree about epistemic and moral standards, algorithmic accountability faces a dilemma: giving absolute primacy to either side would make accountability too one-sided.
- Public reason offers a partial resolution by requiring justifications grounded in common principles acceptable to all reasonable people; it can reassert universal principles against biases inherited by code, force articulation of goals and constraints, help navigate public/private boundaries, clarify epistemic standards, and constrain decision-subjects’ grievances.
- Opacity is a challenge, but the paper argues it is not hopeless: many systems use interpretable models, explanation methods exist, and public reason can help decide whether opacity is acceptable or whether a system should be built at all.
Methods
- Conceptual and normative analysis in philosophy and technology; no empirical data, sample, or experimental design.
- Draws on political philosophy accounts of accountability and public reason, especially Rawls, Quong, and related debates, to analyze algorithmic decision-making and accountability.
- Argument proceeds by identifying the dilemma of reasonable pluralism in algorithmic accountability, proposing public reason as a justificatory framework, and considering objections and limitations.
Limitations
- The paper states two potential limitations: public reason may be redundant if substantive laws already regulate algorithmic decision-making, and opacity of machine learning models may threaten decision-makers’ ability to account for systems.
- The paper acknowledges that it does not presuppose any particular form of public reason and that disagreements about its scope, content, and basis remain.
- Evident from the text: the argument is theoretical and programmatic, with no empirical testing or case study, and the precise content of common principles is left to reflective equilibrium.
Model-written summary. Check it against the paper before relying on it.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
4 Valid · 3 Not valid · 17 Not checked · 83 Own results and interpretations
The sources with a verdict support this claim; the rest weren’t checked yet, have no full text we could read, or their verdicts aren’t rated yet.
In the paper
“If an algorithm is trained on data that are biased or reflect unjust structural inequalities of gender, race or other sensitive attributes, it may ‘learn’ to discriminate using those attributes (or proxies for them).”
§ Algorithmic Accountability and Public Reason > The Rise of Algorithmic Decision-Making > Algorithmic Decision-Making Necessarily Embodies Contestable Epistemic and Normative Assumptions
What the sources say
- Supportedcites Kamiran & Calders, 2012 source
Section 2.1 (p. 3) says a model trained directly on historically biased hiring data will learn discriminatory behavior. Section 6.1 (p. 23) adds that correlated attributes can produce indirect discrimination even when the sensitive attribute is removed. (LLM, conf=0.93)
- Not checked — no full text of the sourcecites Bozdag, 2013 source
- Not checked — no full text of the sourcecites Diakopoulos, 2015 source
hedgedstates a possible cause- 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.
- Supported
The cited sources support only part of this claim — read each source’s verdict below.
In the paper
“In the case of knowledge-based systems, the knowledge that is fed into the system, and assumptions that are involved in modelling it, may reflect biases of the system designers and data collection process”
§ Algorithmic Accountability and Public Reason > The Rise of Algorithmic Decision-Making > Algorithmic Decision-Making Necessarily Embodies Contestable Epistemic and Normative Assumptions
What the sources say
- Not checked — no full text of the sourcecites Wiener, 1960 source
- Not checked — no full text of the sourcecites Weizenbaum, 1972 source
- Partly supportedcites Dutton & Kraemer, 1980 sourceExperimental — accuracy not yet measured
Our copy supports the core mechanism the claim attributes: 'Embedded within every fiscal impact budgeting system are explicit and implicit assumptions and theories which can systematically influence the model outputs in ways which advantage certain groups' and 'models are embedded with various normative assumptions and biases that exist less by design than by the numerous technical choices made by the model designers' (FIBS in Theory and Practice), reinforced in Reforming FIBS Use ('FIBS, like many other computer applications, tend to automate the existing biases of the budgetary process'; departmental involvement lets departments 'inject their traditional biases into the final model outputs'). Two parts of the attribution are not in our copy: the source's subject is fiscal impact budgeting/forecasting models in local government (and computer models generally, 'HBS and other computer models'), not knowledge-based systems whose bias comes from the knowledge fed into them; and it nowhere attributes bias to data collection processes — it notes only reliance on historical expenditure/revenue data and selective use of computer-based information. Searching 'knowledge base', 'expert system', 'data collection' returned no passage on those points. (LLM, conf=0.72)
hedgedstates a possible cause- 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.
The cited sources support only part of this claim — read each source’s verdict below.
In the paper
“Divergence between epistemic standards seems entirely legitimate, for instance, given widespread debates about the soundness and robustness of machine learning models”
§ Algorithmic Accountability and Public Reason > The Rise of Algorithmic Decision-Making > The Dilemma of Reasonable Pluralism
What the sources say
- Partly supportedcites Szegedy et al., 2013 sourceExperimental — accuracy not yet measured
Section 4, p. 4, shows that small, imperceptible changes can cause deep neural networks to misclassify images, supporting the sentence’s reference to concerns about model robustness. The paper does not discuss divergence between epistemic standards or establish that debates about model soundness and robustness are widespread; the legitimacy of such divergence is the citing authors’ own judgment. (LLM, conf=0.78)
- Not checked — no full text of the sourcecites Nguyen, Yosinski, & Clune, 2015 source
hedged- Asserted: passes — The authors put it forward as true.
- Checkable: unclear — It is unclear whether it could be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Partly supported
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 suggests that public reason could act as a constraint on algorithmic decision-making power by ensuring that decision-makers must be able to account for their system’s outputs according to epistemic and normative standards which are acceptable to all reasonable people.”
§ Algorithmic Accountability and Public Reason > Algorithmic Accountability as Public Reason > Public Reason as a Constraint in Algorithmic Accountability
hedgedstates a possible cause- Asserted: passes — The authors put it forward as true.
- Checkable: unclear — It is unclear whether it could be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Rests on an assumption not shown here.
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
“Giving absolute primacy to either the decision-maker or the decision-subject would render algorithmic accountability too one-sided, allowing one party to hold the other to standards that they could not reasonably accept.”
§ Algorithmic Accountability and Public Reason > The Rise of Algorithmic Decision-Making > The Dilemma of Reasonable Pluralism
hedgedstates a cause- Asserted: passes — The authors put it forward as true.
- Checkable: unclear — It is unclear whether it could be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
42 more claims
Create a free account to see all 107 claims and to verify the 4 references still open.
- Probably not a substantial claim. 55
- It may state two separate claims as one. 14
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
“However, the literature we have reviewed provides a mixed view of the relationships between transparency, understanding, fairness and trust (Binns 2018;Cheng et al. 2019;Kizilcec 2016).”
From Tensions in transparent urban AI: designing a smart electric vehicle charge point (2023) · cites this paper as Binns 2018
Experimental — accuracy not yet measuredThe source backs part of the claim, not all of it — or backs it for a different population, condition or comparison.
In the citing paper
“In public-law terms, the concern is not only whether AI improves efficiency but also whether algorithmic assistance remains compatible with accountable justification, discretionary control, and rule-of-law legitimacy (Danaher, 2016;Binns, 2018;Busuioc, 2021).”
From From Human Oversight to Cognitive Sovereignty: A Process-Based Governance Standard for AI-Assisted Legal Reasoning (2026) · cites this paper as Binns, 2018
Experimental — accuracy not yet measuredThe source backs part of the claim, not all of it — or backs it for a different population, condition or comparison.
In the citing paper
“Public-law scholarship shows that algorithmic decision-making raises problems of accountability, public reason, judicial review and rule-of-law legitimacy (Binns, 2018;Cobbe, 2019;Busuioc, 2021;Zalnieriute et al., 2019).”
From From Human Oversight to Cognitive Sovereignty: A Process-Based Governance Standard for AI-Assisted Legal Reasoning (2026) · cites this paper as Binns, 2018
Experimental — accuracy not yet measuredThe cited source says what the claim attributes to it.
In the citing paper
“Binns (2018) frames algorithmic accountability as a matter of public reason: affected persons and institutions require intelligible justifications, not merely correct outputs.”
From From Human Oversight to Cognitive Sovereignty: A Process-Based Governance Standard for AI-Assisted Legal Reasoning (2026) · cites this paper as Binns (2018)
The cited source says what the claim attributes to it.
In the citing paper
“It also reflects the public-reason requirement that exercises of authority be justified through reasons that can be inspected, contested and institutionally attributed (Binns, 2018).”
From From Human Oversight to Cognitive Sovereignty: A Process-Based Governance Standard for AI-Assisted Legal Reasoning (2026) · cites this paper as Binns, 2018
5 more claims about this paper
Create a free account to see every claim other papers make about this one.
Citations
Cites
Artificial intelligence: a modern approach
(1995) · Choice Reviews Online · doi:10.5860/choice.33-1577 · Cited by 21,974
Causality: Models, Reasoning, and Inference
Pearl, J. (2009) · doi:10.1017/cbo9780511803161 · Cited by 11,653
Intriguing properties of neural networks
Szegedy, C., Zaremba, W., Sutskever, I., Bruna, J., Erhan, D., Goodfellow, I., & Fergus, R. (2013) · arXiv · doi:10.48550/arxiv.1312.6199 · Cited by 8,143
Empirical Model Building and Response Surfaces
Freeny, A. (1988) · Technometrics · doi:10.1080/00401706.1988.10488371 · Cited by 5,235
Deep neural networks are easily fooled: High confidence predictions for unrecognizable images
Nguyen, A., Yosinski, J., & Clune, J. (2015) · 2015 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr.2015.7298640 · Cited by 3,258
How the machine “thinks”: Understanding opacity in machine learning algorithms
Burrell, J. (2016) · Big Data & Society · doi:10.1177/2053951715622512 · Cited by 3,196 · cited 1×
“This may lead to ‘algocracy’, in which ‘the legitimacy of public decision-making processes’ is thwarted by ‘the opacity of certain algocratic governance systems’ (Danaher, 2016).”
BackgroundEmpirical Model-Building and Response Surfaces.
Kshirsagar, A. M., Box, G. E. P., & Draper, N. R. (1988) · Journal of the American Statistical Association · doi:10.2307/2288890 · Cited by 1,894
Artificial Intelligence: A Modern Approach
Ligeza, A. (1995) · Neurocomputing · doi:10.1016/0925-2312(95)90020-9 · Cited by 1,857
Ananny, M., & Crawford, K. (2018) · New Media & Society · doi:10.1177/1461444816676645 · Cited by 1,836 · cited 2×
“The lack of interpretability in such algorithmic decision-making systems therefore threatens the ability of decision-makers to account for their systems (Neyland, 2007; Anderson, 2011; O’Reilly & Goldstein, 2013; Burrell, 2016; Ananny & Crawford, 2017) .”
BackgroundThe Unreasonable Effectiveness of Data
Halevy, A., Norvig, P., & Pereira, F. (2009) · IEEE Intelligent Systems · doi:10.1109/mis.2009.36 · Cited by 1,767
Research works printed but not linked: 5
- Barocas, S., & Selbst, A. D.. (2016). “Big Data’s Disparate Impact.” https://papers.ssrn.com/sol3/papers.cfm?abstract_id=2477899 .
- Deville, J. (2013). “Leaky Data: How Wonga Makes Lending Decisions.” Charisma: Consumer Market Studies . http://www.academia.edu/download/34144234/Deville_-_2014_-_Leaky_data.doc .
- Rawls, J. (1996). Political liberalism (with a new introduction and the “reply to Habermas”). New York, Columbia University Press, 1(5), 11–11.
- Sandvig, C. (2015). “Seeing the Sort: The Aesthetic and Industrial Defense of ‘the Algorithm.” Journal of the New Media Caucus| ISSN, 017X.
- Sandvig, C., Hamilton, K., Karahalios, K., and Langbort, C. (2014). “Auditing Algorithms: Research Methods for Detecting Discrimination on Internet Platforms.” Data and Discrimination: Converting Critical Concerns into Productive Inquiry. https://pdfs.semanticscholar.org/b722/7cbd34766655dea10d0437ab10df3a127396.pdf
Cited by
Morley, J., Floridi, L., Kinsey, L., & Elhalal, A. (2020) · Science and Engineering Ethics · doi:10.1007/s11948-019-00165-5 · Cited by 882
“In this sense, the tools and methodologies represent a pragmatic version of Habermas’s discourse ethics 22 (Mingers and Walsham 2010).”
BackgroundMittelstadt, B., Russell, C., & Wachter, S. (2019) · Proceedings of the Conference on Fairness, Accountability, and Transparency · doi:10.1145/3287560.3287574 · Cited by 794
“Justification has recently received increasing attention by scholars discussing algorithmic accountability (e.g. Binns (2017); Biran and McKeown (2014); Hildebrandt and Koops (2010); Selbst and Barocas (2018); Weller (2017)).”
BackgroundThe Ethics of Artificial Intelligence: Principles, Challenges, and Opportunities
Floridi, L. (2023) · doi:10.1093/oso/9780198883098.001.0001 · Cited by 507
Trust in AI: progress, challenges, and future directions
Afroogh, S., Akbari, A., Malone, E., Kargar, M., & Alambeigi, H. (2024) · Humanities and Social Sciences Communications · doi:10.1057/s41599-024-04044-8 · Cited by 507
“Within this project, the so-called “ right to explanation ” has been central (Ahn et al., 2021; Alam and Mueller, 2021; Ausloos et al., 2020; Binns, 2018; Bovens et al., 2014; Buçinca et al., 2021; Doshi-Velez et al., 2016; Feldman et al., 2019; Smith-Renner et al., 2020; Sperrle et al., 2020;…”
Accountability in artificial intelligence: what it is and how it works
Novelli, C., Taddeo, M., & Floridi, L. (2024) · AI & SOCIETY · doi:10.1007/s00146-023-01635-y · Cited by 495
“: because the agent may have to provide pieces of evidence, explanations and justifications to objections on both technical and social sources of conduct (Binns 2018).”
Metadata
- Authors
- Binns, R.
- Affiliations
- University of Oxford (GB)
- Corresponding author
- Binns, Reuben
- ORCID
- Binns, Reuben
- Venue
- Philosophy & Technology
- Published
- 2017-05-24
- Volume
- 31
- Issue
- 4
- Pages
- 543-556
- ISSN
- 2210-5433 (ISSN-L)
- 2210-5441
- PMC
- PMC6390894
- OpenAlex
- W2619904157
- Semantic Scholar
- 450714af44319cc18f8d53f05b90b91ac97948cd
- Cited by
- 457
- References
- 58
- Citations by year
- 2026: 96 · 2025: 94 · 2024: 28 · 2023: 115 · 2022: 37 · 2021: 37 · 2020: 24 · 2019: 20 · 2018: 6
- FWCI
- 31.35
- Type
- journal article
- Language
- en
- Open access
- hybrid
- doi.org
- Licence
- cc-by
- creativecommons.org/licenses/by/4.0 (tdm)
- Locations
- Philosophy & Technology (publishedVersion, cc-by)
- Europe PMC (PubMed Central) (publishedVersion, other-oa)
- PubMed Central (publishedVersion, cc-by)
- PubMed (publishedVersion)
- Funders
- Engineering and Physical Sciences Research Council (EP/J017728/2)
- Topics
- Physical Sciences › Computer Science › Artificial Intelligence › Ethics and Social Impacts of AI
- Social Sciences › Political Science and International Relations › Political Philosophy and Ethics
- Physical Sciences › Computer Science › Artificial Intelligence › Explainable Artificial Intelligence (XAI)
- Keywords
- algorithmic accountability, public reason, automated decision-making, justification, democratic theory, political philosophy, algorithmic governance, transparency
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.
The claim verdicts this report counts are experimental: their accuracy is not yet measured.
A language model drew the claims from the paper on 2026-10-10, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).
A second model labelled each claim on 2026-10-10 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).
Cited sources are checked with a language model’s help (gpt-6-sol and qwen3.8-omni-flash).
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
How to cite
Cite CoChat for the checks on this page, and the paper for its own findings.
- This page
- CoChat. (2026). Citation check: Algorithmic Accountability and Public Reason. https://app.cochat.ai/papers/10.1007/s13347-017-0263-5
- The paper
- Binns, R. (2018). Algorithmic Accountability and Public Reason. Philosophy & Technology. https://doi.org/10.1007/s13347-017-0263-5
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.