Understanding Artificial Agency
Open accessThe DOI matches its Crossref record; 75 of the 75 references it cites were found in a registry or an index.
Abstract
Which artificial intelligence (AI) systems are agents? To answer this question, I propose a multidimensional account of agency. According to this account, a system's agency profile is jointly determined by its level of goal-directedness and autonomy as well as is abilities for directly impacting the surrounding world, long-term planning and acting for reasons. Rooted in extant theories of agency, this account enables fine-grained, nuanced comparative characterizations of artificial agency. I show that this account has multiple important virtues and is more informative than alternatives. More speculatively, it may help to illuminate two important emerging questions in AI ethics: 1. Can agency contribute to the moral status of non-human beings, and how? 2. When and why might AI systems exhibit power-seeking behaviour and does this pose an existential risk to humanity?
Overview
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
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?
In the citing paper
“Formulating an agency indicator is challenging because accounts of agency vary widely [65].”
From Identifying indicators of consciousness in AI systems (2026) · cites this paper as [65]
Citations
Cites
Reinforcement Learning: An Introduction
Sutton, R. S., & Barto, A. G. (1998) · IEEE Transactions on Neural Networks · doi:10.1109/tnn.1998.712192 · Cited by 27,225
Artificial intelligence: A modern approach
Nilsson, N. J. (1996) · Artificial Intelligence · doi:10.1016/0004-3702(96)00007-0 · Cited by 10,583
Nagel, T. (1974) · The Philosophical Review · doi:10.2307/2183914 · Cited by 7,440
Stanford Encyclopedia of Philosophy
(2015) · Choice Reviews Online · doi:10.5860/choice.187763 · Cited by 5,309
Davidson, D. (1963) · The Journal of Philosophy · doi:10.2307/2023177 · Cited by 2,938 · cited 1×
“As Butlin himself grants, philosophers conceive of intentional action often in a more demanding way (e.g. Davidson 1963; Korsgaard 2008; 2018).”
BackgroundFrontiers of Justice: Disability, Nationality, Species Membership
Nussbaum, M. C. (2007) · Harvard University Press eBooks · doi:10.2307/j.ctv1c7zftw · Cited by 1,517 · cited 1×
“According to the dominant view, moral status is determined by sentience, i.e. the capacity to undergo pleasant and unpleasant conscious experiences such as joy, pain and fear (Kriegel 2019; Nussbaum 2007; Singer 2011).”
BackgroundList, C., & Pettit, P. (2011) · doi:10.1093/acprof:oso/9780199591565.001.0001 · Cited by 1,344
Dennett, D. C. (1991) · The Journal of Philosophy · doi:10.2307/2027085 · Cited by 973
Singer, P. (2011) · Cambridge University Press eBooks · doi:10.1017/cbo9780511975950 · Cited by 882
Superintelligence: Paths, Dangers, Strategies
Mulgan, T. (2015) · The Philosophical Quarterly · doi:10.1093/pq/pqv034 · Cited by 878
Research works printed but not linked: 5
- Bourget, 2019, Phenomenal Intentionality
- Dennett, 1987, The Intentional Stance
- Jaworska, 2021, The Grounds of Moral Status
- Luper, 2021, Death
- Schlosser, 2019, Agency
Cited by
Floridi, L. (2025) · Philosophy & Technology · doi:10.1007/s13347-025-00858-9 · Cited by 154
“For example, (Dung, 2024) proposes a five-dimensional framework for understanding Artificial Agency encompassing goal-directedness, autonomy, efficacy, planning, and intentionality.”
Identifying indicators of consciousness in AI systems
Butlin, P., Long, R., Bayne, T., Bengio, Y., Birch, J., Chalmers, D., Constant, A., Deane, G., Elmoznino, E., Fleming, S. M., et al. (2026) · Trends in Cognitive Sciences · doi:10.1016/j.tics.2025.10.011 · Cited by 56
The argument for near-term human disempowerment through AI
Dung, L. (2025) · AI & SOCIETY · doi:10.1007/s00146-024-01930-2 · Cited by 27
Self-citationInherent and emergent liability issues in LLM-based agentic systems: a principal-agent perspective
Gabison, G. A., & Xian, R. P. (2025) · Proceedings of the 1st Workshop for Research on Agent Language Models (REALM 2025) · doi:10.18653/v1/2025.realm-1.9 · Cited by 19
“PAT requires a clarification of the agency relationship, which remains a hotly debated interdisciplinary topic for LLM-based systems (Shavit et al., 2023; Dai, 2024; Barandiaran and Almen-dros, 2024; Dung, 2024; Perrier and Bennett, 2025; Mattingly and Cibralic, 2025; Butlin, 2025; Das, 2025).”
Inherent and emergent liability issues in LLM-based agentic systems: a principal-agent perspective
Gabison, G. A., & Xian, R. P. (2025) · arXiv · doi:10.48550/arxiv.2504.03255 · Cited by 19
In ChatGPT they trust: a study of students' perceptions and misuse of ChatGPT in higher education
Yeung, L. K. C., Chow, D. P. L., Wong, P. H., & Lau, S. S. S. (2026) · AI and Ethics · doi:10.1007/s43681-025-00855-w · Cited by 16
“When an entity generates responses that appear reasoned or purposeful, it is more likely to be viewed as intentional [18].”
InfluentialIs superintelligence necessarily moral?
Dung, L. (2024) · Analysis · doi:10.1093/analys/anae033 · Cited by 16
Self-citationHome as mind: AI extenders and affective ecologies in dementia care
Krueger, J. W. (2025) · Synthese · doi:10.1007/s11229-025-04923-w · Cited by 9
Henderson, M. D. (2025) · Leadership & Organization Development Journal · doi:10.1108/lodj-05-2025-0319 · Cited by 15
Evaluating approaches for reducing catastrophic risks from AI
Dung, L. (2024) · AI and Ethics · doi:10.1007/s43681-024-00475-w · Cited by 13
Self-citationMoret, A. (2025) · Philosophical Studies · doi:10.1007/s11098-025-02343-7 · Cited by 25
“More specifically a growing number of philosophers and AI researchers have argued that advanced AI systems pose a catastrophic risk to humanity (Bostrom, 2014; Russell, 2019; Ord, 2020; Vold and Harris, 2021; Cohen, et al., 2022; Carlsmith, 2022; Dung, 2024a).”
InfluentialEthically charged decisions and the future of ‘AI Ethics’
Buccella, A. (2025) · AI & Society · doi:10.1007/s00146-025-02573-7 · Cited by 5
Agentic AI as a service innovation: A mixed-methods study of satisfaction, trust, and continued use
Al-Sharafi, M. A., Alzaeemi, S. A., Dwivedi, Y. K., Abdulkareem, Z. O., Ali, A., & Chae, I. (2026) · Journal of Innovation & Knowledge · doi:10.1016/j.jik.2026.101039 · Cited by 8
Butlin, P. (2024) · Inquiry · doi:10.1080/0020174x.2024.2439995 · Cited by 13
“They are also very di ff erent from deep reinforcement learning agents such as AlphaGo and its descendants (Silver 2016; Schrittwieser 2020), which are the clearest cases of agency in AI (Butlin 2022, 2024; Dung 2024).”
Agentic profiles for effective AI governance
Kasirzadeh, A., & Gabriel, I. (2026) · Nature · doi:10.1038/s41586-026-10805-z · Cited by 93
The Philosophy of Language Models
Millière, R., & Buckner, C. J. (2026) · Philosophy Compass · doi:10.1111/phc3.70095 · Cited by 10
“Similarly, Dung (2024) proposes that agency is not a binary property but a multidimensional construct.”
Becoming oneself online: Social media platforms as environments for self-transformation
Marin, L. (2025) · Humanities and Social Sciences Communications · doi:10.1057/s41599-025-04495-7 · Cited by 5
“In a classical view of agency (Dung 2024, pp. 3-8), there are several conditions for someone to have agency: X has agency in regard to A if X autonomously decided to do A, if they achieved A ef fi caciously (they were able to trigger the change and were not reliant on another agent to initiate it),…”
Dung, L., & Newen, A. (2025) · Philosophical Studies · doi:10.1007/s11098-025-02363-3 · Cited by 3
Self-citationMisalignment or misuse? The AGI alignment tradeoff
Hellrigel-Holderbaum, M., & Dung, L. (2025) · Philosophical Studies · doi:10.1007/s11098-025-02403-y · Cited by 11
Self-citationInterrogating artificial agency
Wong, H. Y. (2025) · Frontiers in Psychology · doi:10.3389/fpsyg.2024.1449320 · Cited by 13
AI Alignment Versus AI Ethical Treatment: 10 Challenges
Bradley, A., & Saad, B. (2025) · Analytic Philosophy · doi:10.1111/phib.12380 · Cited by 7
Artificial Externality: A Three-Layer Model of Reality from Substrate to Smart Contract
Suzuki, K. (2026) · Philosophy & Technology · doi:10.1007/s13347-026-01158-6 · Cited by 3
Respess, S. (2026) · Philosophy & Technology · doi:10.1007/s13347-025-01028-7 · Cited by 1
“Though there have been recent strides to implement a relational view of agency in AI (e.g. Mhlambi & Tiribelli, 2023 ), much of the field is still dominated by models of independent goal-directedness and planning (Dung, 2025).”
Three frameworks for AI mentality
Shevlin, H. (2026) · Frontiers in Psychology · doi:10.3389/fpsyg.2026.1715835 · Cited by 1
“For a taxonomy of different dimensions of agency in the AI context, see Dung (2025). occurrent states, and they are often ascribed on the basis of stable behavioural patterns rather than discrete episodes.”
Roberts, T. (2026) · Inquiry · doi:10.1080/0020174x.2026.2658565 · Cited by 1
Artificial minds and AI duplication: the very idea
Kersten, L., & Dung, L. (2026) · Inquiry · doi:10.1080/0020174x.2026.2684768 · Cited by 2
“These have included, for instance, probing discussions of beliefs (Herrmann and Levinstein 2025), desires (Goldstein and Kirk-Giannini 2025), concepts (Butlin 2021), agency (Dung 2024), strategic deception (Park et al. 2023), and mental representation (Harding 2023), to list a few.”
Self-citationArtificial Intelligence & Human Rights Law: A Thematic Synthesis Review
Prifti, K., Hoppenbrouwers, J., de Leeuw, E., Quintavalla, A., & Temperman, J. (2026) · Minds and Machines · doi:10.1007/s11023-026-09793-w · Cited by 1
“While this academic debate remains unsettled (Veluwenkamp & Hindriks, 2024 ), the multidimensional nature of artificial agency is highlighted, reflecting different layers of autonomy and decision-making (Dung, 2025 ).”
William Fontaine’s (Forgotten) Kantian Analysis of Ernest Everett Just’s Mathematical Biology
Vilbig, R. (2025) · International Journal of Philosophy · doi:10.11648/j.ijp.20251301.13
“In the other direction of this trend, however, are recent papers by Birch et al. (2020) [4] and Dung (2024) [16] that have provided a 5-dimensional analysis of animal and human consciousness [4, 16].”
AI Governance in Academia: Guidelines for Generative AI
Peterson, C., & Deschênes, M. (2025) · Proceedings of the ... International Florida Artificial Intelligence Research Society Conference · doi:10.32473/flairs.38.1.138855 · Cited by 1
Against the Manhattan project framing of AI alignment
Friederich, S., & Dung, L. (2025) · Mind & Language · doi:10.1111/mila.12548 · Cited by 7
“…alignment ” in such a way that we can be sure to prevent losing control to AI, given the enormity of what is at stake, the only responsible option is to make a global, coordinated effort to stop crucial steps towards the development of increasingly general AI systems (Dung, 2024a, section 4.1).”
InfluentialSelf-citationMorandín-Ahuerma, F. (2025) · International Journal of Human-Computer Interaction · doi:10.1080/10447318.2025.2565393 · Cited by 1
“The foundation of computational compatibilism—a theory that claims artificial agency can be assessed as free in a functional sense, without requiring awareness, consciousness or subjective will—is these three criteria (Dung, 2025; Fritz et al., 2020; Laukyte, 2017).”
A Two-Step, Multidimensional Account of Deception in Language Models
Dung, L. (2025) · Erkenntnis · doi:10.1007/s10670-025-01017-4 · Cited by 3
“A key assumption of technical AI alignment research is that AI capability, what AI systems can do, and AI alignment – which final goals they pursue – are indepen - dent and thus based on different mechanisms (Bostrom, 2012; Dung, 2024b; Müller & Cannon, 2022).”
InfluentialSelf-citationMultidimensional Properties: Primitivism vs Reductionism
Michels, R. (2025) · Metaphysics · doi:10.5334/met.189
AI deception and moral standing
Skretta, A. (2025) · Philosophical Studies · doi:10.1007/s11098-025-02465-y
“Otherwise, behaviour driven by calculations of possible outcomes through subpersonal mental states, e.g. motor representations in the brain which the subject can neither reflect on nor consciously experience, might count as intentional action” (Dung, 2024 , p. 455–456).”
Agent or hammer?: A philosophical inquiry into machine moral agency
Asif, S. (2025) · AI and Ethics · doi:10.1007/s43681-025-00920-4
“Moral agency involves more than the ability to act; it requires the capacity to act for moral reasons [25].”
Minds of their own? Decoding free will in large language models
Lavazza, A., Sartori, G., Farina, M., Melis, G., & Orrù, G. (2026) · AI & Society · doi:10.1007/s00146-025-02809-6 · Cited by 3
“Agency can be depicted as a multifactorial property (Dung 2024).”
Values in science and AI alignment research
Dung, L. (2026) · Inquiry · doi:10.1080/0020174x.2026.2615773
“Such impacts encompass both current harms as well as anticipated risks, including more speculative catastrophic and existential risks (Bostrom 2014; Dung 2024a).”
Self-citationPhase animalism and artificial intelligence
Sauchelli, A. (2026) · Synthese · doi:10.1007/s11229-026-05496-y
What even are AI-generated responsibility gaps?
Königs, P. (2026) · AI and Ethics · doi:10.1007/s43681-026-01047-w
Cecchini, D. (2026) · Advances in neuroethics · doi:10.1007/978-3-032-20117-1_10
Zhang, X. (2026) · Journal of Innovation & Knowledge · doi:10.1016/j.jik.2026.101050 · Cited by 1
Moral Responsibility Without Moral Agency
Mahant, N. (2026) · Springer eBooks · doi:10.1007/978-3-032-15112-4_18
Glikson, E., Asscher, O., & Harush, R. (2026) · Cross Cultural & Strategic Management · doi:10.1108/ccsm-09-2024-0230
Risks Deriving from the Agential Profiles of Modern AI Systems
Crook, B. (2026) · Synthese Library/Synthese library · doi:10.1007/978-3-032-10073-3_4
Talkative AI and the fiction of artificial minds
Roberts, T. (2026) · Philosophical Studies · doi:10.1007/s11098-026-02560-8
Karimova, G. Z. (2026) · SSRN Electronic Journal · doi:10.2139/ssrn.6990449
How continuity distinguishes autonomy from agency in agentic AI
Kahl, P. (2026) · Discover Artificial Intelligence · doi:10.1007/s44163-026-01675-5 · Cited by 3
Examining the AI’s legal status from legal and theoretical perspectives
Labanieh, M. F., Yusoff, Z. M., Ayub, Z. A., Wahab, H. A., & Shariffuddin, M. D. K. (2026) · Multidisciplinary Science Journal · doi:10.31893/multiscience.2026779 · Cited by 1
Toward an internal moral purpose like normative governance function in artificial agency
Keleş, S. (2026) · AI and Ethics · doi:10.1007/s43681-026-01268-z
“List’s (2021) work on group agency and artificial intelligence, Dung’s [ 11 ] account of artificial agency, and Laukyte’s [23] discussion of artificial agents all help clarify the sense in which artefacts may be treated as agent-like for governance purposes without being persons.”
Zhu, Y., Wang, C., & Chu, J. (2026) · International Journal of Human-Computer Interaction · doi:10.1080/10447318.2026.2709241 · Cited by 1
Normative competence and the two faces of AI agency
Gibson, Q. H. (2026) · Inquiry · doi:10.1080/0020174x.2026.2720905
Investigating the workplace networks of Chinese in-house translators in the GenAI era
Hao, X. (2026) · Translation Spaces · doi:10.1075/ts.25065.hao
Skorburg, J. A., & White, D. (2026) · Cambridge University Press eBooks · doi:10.1017/9781009639057.029
Perry, A., & Cameron, C. D. (2026) · Cambridge University Press eBooks · doi:10.1017/9781009639057.024
Big AI-Based Mathematics and the Redistribution of Epistemic Labor
Friedman, M., & Bar-On, K. K. (2026) · Social Studies of Science · doi:10.1177/03063127261482050
The room of amnesiacs: AI individuation and reference
Cappelen, H., & Hawthorne, J. (2026) · Inquiry · doi:10.1080/0020174x.2026.2723988
A Scientific Theory of Consciousness
Goldstein, S., & Kirk‐Giannini, C. D. (2026) · Oxford University Press eBooks · doi:10.1093/9780197840375.003.0009
Goldstein, S., & Kirk‐Giannini, C. D. (2026) · Oxford University Press eBooks · doi:10.1093/9780197840375.003.0002
Goldstein, S., & Kirk‐Giannini, C. D. (2026) · Oxford University Press eBooks · doi:10.1093/9780197840375.003.0008
Goldstein, S., & Kirk‐Giannini, C. D. (2026) · Oxford University Press eBooks · doi:10.1093/9780197840375.003.0001
Goldstein, S., & Kirk‐Giannini, C. D. (2026) · Oxford University Press eBooks · doi:10.1093/9780197840375.003.0010
Goldstein, S., & Kirk‐Giannini, C. D. (2026) · Oxford University Press eBooks · doi:10.1093/9780197840375.003.0011
Goldstein, S., & Kirk‐Giannini, C. D. (2026) · Oxford University Press eBooks · doi:10.1093/9780197840375.003.0004
Goldstein, S., & Kirk‐Giannini, C. D. (2026) · Oxford University Press eBooks · doi:10.1093/9780197840375.003.0013
Goldstein, S., & Kirk‐Giannini, C. D. (2026) · Oxford University Press eBooks · doi:10.1093/9780197840375.003.0006
Goldstein, S., & Kirk‐Giannini, C. D. (2026) · Oxford University Press eBooks · doi:10.1093/9780197840375.003.0017
Goldstein, S., & Kirk‐Giannini, C. D. (2026) · Oxford University Press eBooks · doi:10.1093/9780197840375.003.0015
Goldstein, S., & Kirk‐Giannini, C. D. (2026) · Oxford University Press eBooks · doi:10.1093/9780197840375.003.0014
Belief and Desire in AI Systems
Goldstein, S., & Kirk‐Giannini, C. D. (2026) · Oxford University Press eBooks · doi:10.1093/9780197840375.003.0005
Goldstein, S., & Kirk‐Giannini, C. D. (2026) · Oxford University Press eBooks · doi:10.1093/9780197840375.003.0016
AI Welfare: Agency, Consciousness, Sentience
Goldstein, S., & Kirk‐Giannini, C. D. (2026) · doi:10.1093/9780197840375.001.0001 · Cited by 5
A Case for AI Consciousness: Language Agents and Global Workspace Theory
Goldstein, S., & Kirk-Giannini, C. D. (2024) · arXiv · doi:10.48550/arxiv.2410.11407 · Cited by 4
Metadata
- Authors
- Dung, L.
- Affiliations
- Friedrich-Alexander-Universität Erlangen-Nürnberg (DE)
- Corresponding author
- Dung, Leonard
- ORCID
- Dung, Leonard
- Venue
- The Philosophical Quarterly
- Publisher
- Oxford University Press (OUP)
- Published
- 2024-02-05
- Volume
- 75
- Issue
- 2
- Pages
- 450-472
- ISSN
- 0031-8094 (ISSN-L)
- 1467-9213
- OpenAlex
- W4391592270
- Cited by
- 69
- References
- 75
- Citations by year
- 2026: 45 · 2025: 20 · 2024: 4
- FWCI
- 28.73
- Type
- journal article
- Language
- en
- Open access
- green
- philarchive.org
- Licence
- Locations
- The Philosophical Quarterly (publishedVersion)
- PhilPapers (PhilPapers Foundation) (submittedVersion, other-oa)
- PhilPapers (PhilPapers Foundation) (submittedVersion)
- Funders
- BMBF (033KI216)
- Topics
- Physical Sciences › Computer Science › Artificial Intelligence › Ethics and Social Impacts of AI
- Social Sciences › Law › Literary and Philosophical Studies
- Social Sciences › Arts and Humanities › History and Philosophy of Science › Cybernetics and Technology in Society
- Keywords
- artificial agency, goal-directed behavior, autonomy, long-term planning, AI ethics, existential risk, artificial intelligence
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.
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
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.