Democracy, epistemic agency, and AI: political epistemology in times of artificial intelligence
Open accessThe DOI matches its Crossref record; 48 of the 48 references it cites were found in a registry or an index; 5 of the 21 sources its claims cite were checked for support.
Abstract
Abstract Democratic theories assume that citizens have some form of political knowledge in order to vote for representatives or to directly engage in democratic deliberation and participation. However, apart from widespread attention to the phenomenon of fake news and misinformation, less attention has been paid to how they are supposed to acquire that knowledge in contexts shaped by artificial intelligence and related digital technologies. While this topic can also be approached from an empirical angle, this paper contributes to supporting concerns about AI and democracy by looking at the issue through the lens of political epistemology, in particular using the concept of epistemic agency. It argues that artificial intelligence (AI) endangers democracy since it risks to diminish the epistemic agency of citizens and thereby undermine the relevant kind of political agency in democracy. It shows that next to fake news and manipulation by means of AI analysis of big data, epistemic bubbles and the defaulting of statistical knowledge endanger the epistemic agency of citizens when they form and wish to revise their political beliefs. AI risks to undermine trust in one’s own epistemic capacities and hinder the exercise of those capacities. If we want to protect the knowledge basis of our democracies, we must address these problems in education and technology policy.
Overview
CoChat Summary
TL;DR
The paper argues that AI—especially machine learning in social media—threatens democracy by diminishing citizens’ epistemic agency, their control over forming and revising political beliefs. It analyzes fake news, manipulation, epistemic bubbles, and the defaulting of statistical knowledge, and recommends education and technology regulation to protect democracy.
Key findings
- AI can undermine epistemic agency by eroding trust in one’s own epistemic capacities: fake news and deepfakes make citizens doubt their ability to distinguish truth from falsehood, opening the door to technocracy and totalitarianism. The paper cites a 2019 altered video of Nancy Pelosi that was debunked but shared on Facebook and then Twitter by Donald Trump, receiving millions of views.
- AI-based micro-targeting can directly manipulate political beliefs. The paper cites Cambridge Analytica, where Facebook user data were used by the Trump presidential campaign in 2016 to target voters by personality type, and AI automatically tested thousands of variations of an ad before deciding which one to place.
- Non-intended effects also reduce epistemic agency: recommender and search algorithms create epistemic bubbles/filter bubbles that reduce exposure to opposing views and make belief revision harder; AI also defaults statistical/correlational knowledge over causal scientific evidence, illustrated by an adapted thought experiment about Claire, raised in a white supremacist environment, who may be prevented from revising her belief by AI-provided statistics.
- The paper recommends education about statistics and its limitations, critical reasoning, and epistemic virtues; regulation of AI developers and Big Tech; democratic governance of AI; possible epistemic rights; and global governance, since current institutions are insufficiently supranational.
Methods
- Conceptual/philosophical paper in political epistemology; no empirical data collection or sample. It focuses on AI in the sense of machine learning algorithms used in digital social media and on the Internet.
- Main analysis uses the concept of epistemic agency to build conceptual arguments and justifications about AI, knowledge, and democracy, drawing on examples such as Pizzagate, Cambridge Analytica, the Pelosi deepfake, and claims about China and Myanmar, plus an adapted thought experiment from Bondy.
- The paper derives normative policy recommendations from this conceptual analysis rather than from empirical testing.
Limitations
- The paper states it is limited to offering conceptual arguments and justifications based on epistemic agency; it acknowledges that empirical reality is more complex, that epistemic agency was already under attack before AI, and that AI might also have good effects on democracy, though that is not its topic.
- The paper states its argument relies on “if” and “in so far” qualifications—for example, that epistemic bubbles are created and effective—and notes empirical criticisms that the echo chamber is overstated and that depolarization tendencies have been observed.
- Evident from the text: there is no empirical measurement or testing; examples are illustrative and sometimes anecdotal; the Claire case is a thought experiment; and the causal claims about AI’s effects on epistemic agency are not systematically tested.
Model-written summary. Check it against the paper before relying on it.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
5 Valid · 1 Not valid · 12 Not checked · 69 Own results and interpretations
Every source the paper cites for this claim was checked and supports it.
In the paper
“Bozdag and van den Hoven [10] offer some examples of software designs that try to ‘break’ filter bubbles, while noting that such tools are often limited in terms of the range of democracy models they use.”
§ Democracy, epistemic agency, and AI: political epistemology in times of artificial intelligence > Conclusion with policy recommendations
What the sources say
- Supportedcites Bozdag and van den Hoven [10] Bozdag, E., van den Hoven, J.: Breaking the filter bubble: democracy and design. Eth. Inf. Technol. 17, 249–265 (2015) source
The Abstract says the paper provides examples of software designs intended to break filter bubbles and finds that norms from two democracy models dominate those tools while norms from other models are missing. The Conclusion identifies the dominant models as liberal and deliberative, supporting the claim that the designs draw on a limited range of democracy models. (LLM, conf=0.94)
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.
- 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
“It argues that artificial intelligence (AI) endangers democracy since it risks to diminish the epistemic agency of citizens and thereby undermine the relevant kind of political agency in democracy.”
§ Democracy, epistemic agency, and AI: political epistemology in times of artificial intelligence > Abstract
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.
A statement the paper makes itself — it cites no source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Political democratic agency seems to rely on epistemic agency, in the sense that as a citizen in a democracy I need to have some control over the formation of my political knowledge.”
§ Democracy, epistemic agency, and AI: political epistemology in times of artificial intelligence > Introduction: AI, democracy, and epistemic agency
hedgedstates a 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.
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
“and by (often non-intended) phenomena such as epistemic bubbles and the defaulting of statistical knowledge over other types of scientific knowledge that rely on causal evidence.”
§ Democracy, epistemic agency, and AI: political epistemology in times of artificial intelligence > Introduction: AI, democracy, and epistemic agency
plainstates a 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: unclear — It is unclear why the paper says 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
“In addition, if my political beliefs are manipulated, neither voting nor deliberative democracy seem to get off the ground: they all assume that citizens, whether as voters or as participants in deliberation, have control over their own political knowledge, for example over their political beliefs, which they then express by voting, argue for in a deliberative procedure, or defend in an agonistic”
§ Democracy, epistemic agency, and AI: political epistemology in times of artificial intelligence > Introduction: AI, democracy, and epistemic agency
hedgedstates a 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.
28 more claims
Create a free account to see all 87 claims and every source verdict.
- It may state two separate claims as one. 40
- Probably not a substantial claim. 22
- The quote only partly states this claim. 3
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
§ Democracy, epistemic agency, and AI: political epistemology in times of artificial intelligence > Introduction: AI, democracy, and epistemic agency
“Epistemic agency concerns the question regarding control over one’s beliefs (Scholosser [49])”
§ Democracy, epistemic agency, and AI: political epistemology in times of artificial intelligence > References
“49. Schlosser, Markus. 2019. Agency.”
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
§ Democracy, epistemic agency, and AI: political epistemology in times of artificial intelligence > Introduction: AI, democracy, and epistemic agency
“the social-epistemic effects of AI (for example Burell [12])”
§ Democracy, epistemic agency, and AI: political epistemology in times of artificial intelligence > References
“12. Burrell J How the machine ‘thinks’”
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
§ Democracy, epistemic agency, and AI: political epistemology in times of artificial intelligence > Why fake news and misinformation are extra problematic for democracy
“(see for example Zimdars and McLeod [56])”
§ Democracy, epistemic agency, and AI: political epistemology in times of artificial intelligence > References
“56. Zindars M McLeod K Fake news: understanding media and misinformation in the digital age”
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
§ Democracy, epistemic agency, and AI: political epistemology in times of artificial intelligence > Introduction: AI, democracy, and epistemic agency
“Sudmann [52]”
§ Democracy, epistemic agency, and AI: political epistemology in times of artificial intelligence > References
“52. Suddmann A The democratization of artificial intelligence”
Citations
Cites
Nudge: improving decisions about health, wealth, and happiness
(2008) · Choice Reviews Online · doi:10.5860/choice.46-0977 · Cited by 11,572
Epistemic Injustice: Power and the Ethics of Knowing
Fricker, M. (2007) · doi:10.1093/acprof:oso/9780198237907.001.0001 · Cited by 9,198
Strong Democracy: Participatory Politics for a New Age
Bostaph, S., & Barber, B. R. (1985) · Southern Economic Journal · doi:10.2307/1059650 · Cited by 4,000
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×
“It also complements existing literature in ethics of AI that tends to focus on opacity/transparency when it comes to the social-epistemic effects of AI (for example Burell [12]); to this discussion it brings an interesting new topic: epistemic agency.”
BackgroundDoyle, T. (2017) · The Information Society · doi:10.1080/01972243.2017.1354593 · Cited by 3,071
Moral consciousness and communicative action
(1991) · Choice Reviews Online · doi:10.5860/choice.28-3821 · Cited by 1,877
The filter bubble: how the new personalized web is changing what we read and how we think
(2012) · Choice Reviews Online · doi:10.5860/choice.50-0926 · Cited by 1,319
Conceptualizing Epistemic Oppression
Dotson, K. (2014) · Social Epistemology · doi:10.1080/02691728.2013.782585 · Cited by 1,232
ECHO CHAMBERS AND EPISTEMIC BUBBLES
Nguyen, C. T. (2020) · Episteme · doi:10.1017/epi.2018.32 · Cited by 1,115
The echo chamber is overstated: the moderating effect of political interest and diverse media
Dubois, E., & Blank, G. (2018) · Information, Communication & Society · doi:10.1080/1369118x.2018.1428656 · Cited by 1,055
Research works printed but not linked: 6
- Anders, G.: On promethean shame. In: Müller, C.J. (ed.) Prometheanism, pp. 29–95. Rowman & Littlefield, London (2016)
- Arendt, H. 1951/2017. The Origins of Totalitarianism. London: Penguin
- Bergamini, D.: Need for democratic governance of artificial intelligence. Council of Europe, Strasbourg (2020)
- Berlin, I.: Two concepts of liberty. In: Berlin, I. (ed.) The proper study of mankind, pp. 191–242. Chatto & Windus, London (1997)
- O’Neil, C.: Weapons of math destruction. How big data increases inequality and threatens democracy. The Crown Publishing Group, New York (2016)
- Palumbo, A.: Epistemic turn or democratic U-turn? on the tension between philosophical reasoning and political action in deliberative democracy. Teor. Polit. 2, 269–291 (2012)
Cited by
Filter Bubbles and the Unfeeling: How AI for Social Media Can Foster Extremism and Polarization
Rodilosso, E. (2024) · Philosophy & Technology · doi:10.1007/s13347-024-00758-4 · Cited by 138
“In this regard, Mark Coeckelbergh and Elisabeth Anderson point out how filter bubbles undermine subjects’ epistemic agency and can distort and take our opinions to extremes (Anderson, 2021; Coeckelbergh, 2023).”
Wu, J., Lee, Y., Chai, C. S., & Tsai, C. (2025) · Educational Researcher · doi:10.3102/0013189x251333628 · Cited by 113
“The danger of losing epistemic agency is not limited to education; it has also been observed when AI influences political knowledge and beliefs (Coeckelbergh, 2022).”
Coeckelbergh, M., & Sætra, H. S. (2023) · Technology in Society · doi:10.1016/j.techsoc.2023.102406 · Cited by 57
Self-citationReconstructing AI Ethics Principles: Rawlsian Ethics of Artificial Intelligence
Westerstrand, S. (2024) · Science and Engineering Ethics · doi:10.1007/s11948-024-00507-y · Cited by 46
“The development of AI technologies has led to concentration of power over many aspects of our lives to tech giants (Coeckelbergh, 2022a; Nemitz, 2018, p. 2–3), which means that the interests of the shareholders of these companies have an increasing influence on the distribution of inequalities, as…”
BackgroundInfluentialLLMs, Truth, and Democracy: An Overview of Risks
Coeckelbergh, M. (2025) · Science and Engineering Ethics · doi:10.1007/s11948-025-00529-0 · Cited by 70
Self-citation
Metadata
- Authors
- Coeckelbergh, M.
- Affiliations
- University of Vienna (AT)
- Corresponding author
- Coeckelbergh, Mark
- ORCID
- Coeckelbergh, Mark
- Venue
- AI and Ethics
- Published
- 2022-11-22
- Volume
- 3
- Issue
- 4
- Pages
- 1341-1350
- ISSN
- 2730-5953 (ISSN-L)
- 2730-5961
- PMC
- PMC9685050
- OpenAlex
- W4309712795
- Semantic Scholar
- dbafd147f87abee8899662b88ed7c1f759e0827b
- Cited by
- 115
- References
- 46
- Citations by year
- 2026: 48 · 2025: 40 · 2024: 18 · 2023: 9
- FWCI
- 19.40
- Type
- journal article
- Language
- en
- Open access
- hybrid
- doi.org
- Licence
- cc-by
- creativecommons.org/licenses/by/4.0 (tdm)
- creativecommons.org/licenses/by/4.0 (vor)
- Locations
- AI and Ethics (publishedVersion, cc-by)
- PubMed Central (publishedVersion, cc-by)
- PubMed (publishedVersion)
- Funders
- University of Vienna
- Topics
- Physical Sciences › Computer Science › Artificial Intelligence › Ethics and Social Impacts of AI
- Social Sciences › Sociology and Political Science › Misinformation and Its Impacts
- Social Sciences › Political Science and International Relations › Political Theory and Democracy
- Keywords
- artificial intelligence, democracy, epistemic agency, political epistemology, political knowledge, fake news, misinformation, epistemic bubbles, political agency, technology policy
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 2026-10-03, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).
A second model labelled each claim on 2026-10-03 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).
Cited sources are checked with a language model’s help (gpt-6-sol).
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: Democracy, epistemic agency, and AI: political epistemology in times of artificial intelligence. https://app.cochat.ai/papers/10.1007/s43681-022-00239-4
- The paper
- Coeckelbergh, M. (2023). Democracy, epistemic agency, and AI: political epistemology in times of artificial intelligence. AI and Ethics. https://doi.org/10.1007/s43681-022-00239-4
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.