pith. sign in

arxiv: 2605.21351 · v2 · pith:MCTVU47Inew · submitted 2026-05-20 · 💻 cs.HC

The Human-AI Delegation-Verification Dilemma: Individual Strategies, Collective Equilibria and Sociotechnical Lock-in

Angjelin Hila This is my paper

Pith reviewed 2026-06-30 17:05 UTC · model grok-4.3

classification 💻 cs.HC
keywords human-AI interactiondelegationverificationsociotechnical lock-inprisoner's dilemmacollective actionepistemic standardsgame theory
0
0 comments X

The pith

Individually adaptive delegation to AI without safeguards aggregates into sociotechnical lock-in that degrades shared epistemic standards through a prisoner's dilemma.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

This paper maps decision-theoretic strategies for when users delegate tasks to AI versus verifying outputs, showing how feedback drives transitions to stable individual equilibria. It then scales these strategies to collective levels using non-communicative aggregation, local social signaling, and institutional norms setting. The resulting model identifies sociotechnical lock-in as a macro state in which rational delegation choices create a prisoner's dilemma that erodes collective epistemic standards. The paper claims higher communicative standards and institutional norms can impose commitments that shift payoffs away from this suboptimal equilibrium. A reader would care because the argument explains a possible systemic risk in widespread AI adoption even when every agent acts locally rationally.

Core claim

The paper claims that in the absence of communicative and institutional safeguards, individually adaptive delegation strategies aggregate into a systemic collective action problem modeled as a prisoner's dilemma, resulting in sociotechnical lock-in that degrades shared epistemic standards.

What carries the argument

The delegation-verification dilemma, analyzed through decision-theoretic strategies that reach stable equilibria via interaction feedback and then scaled to collective outcomes by the three extrapolation principles of non-communicative aggregation, local social signaling, and institutional norms setting.

If this is right

  • Adoption under higher communicative standards and institutional norms can mitigate suboptimal collective equilibria by imposing social commitments on individual users.
  • Individually stable delegation strategies reach collective prisoner's dilemma equilibria when aggregation occurs without communication or norm-setting.
  • Sociotechnical lock-in emerges as a macro-behavioral state that degrades shared epistemic standards when individual adaptation is left unchecked.
  • The three extrapolation principles suffice to connect micro-level strategy transitions to macro-level collective action problems.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • Interface designs that make verification decisions visible to others could function as the local social signaling mechanism the model requires to escape lock-in.
  • The argument implies that training data quality may decline over time if verification rates fall system-wide, creating a feedback loop not directly modeled in the paper.
  • Empirical tests could track whether populations with strong professional norms around verification show lower lock-in rates than general populations.
  • The model suggests that regulatory requirements for audit trails on AI outputs might serve as an institutional norm that alters individual payoffs.

Load-bearing premise

The three extrapolation principles of non-communicative aggregation, local social signaling, and institutional norms setting are sufficient to scale individually stable strategies to collective equilibria without requiring additional factors.

What would settle it

Compare verification rates and epistemic quality in matched groups of users: one group with enforced channels for communicating verification decisions versus one without, and check whether the no-communication group shows the predicted prisoner's dilemma degradation.

Figures

Figures reproduced from arXiv: 2605.21351 by Angjelin Hila.

Figure 1
Figure 1. Figure 1: Decision diagram illustrating the two strategy poles: Reflective Aug [PITH_FULL_IMAGE:figures/full_fig_p009_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Sample strategy-evolution state-space diagram for one canonical [PITH_FULL_IMAGE:figures/full_fig_p011_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Canonical Path B (suboptimal): throughput-weighted users converge [PITH_FULL_IMAGE:figures/full_fig_p011_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Canonical Path C (mixed-assurance): low-trust users adopt [PITH_FULL_IMAGE:figures/full_fig_p012_4.png] view at source ↗
read the original abstract

This paper takes an ecological approach toward large-scale models of hybrid human-AI intelligence. Emerging models of human-AI interaction predominantly advance the complementarity thesis variously dubbed human-AI collaboration and human-AI hybrid intelligence. However, this constitutes an over-simplification of the modalities of human-AI interaction and possibility-space for both individual and collective action that human-AI interaction potentiates. To fill these gaps, this paper develops a decision and game-theoretic approach to the human-AI delegation-verification dilemma. First, we map out canonical decision-theoretic strategies that account for adaptive user trajectories, modeling how agents transition between strategies based on interaction feedback to reach stable equilibria. Second, we scale individually stable strategies to collective equilibria using three extrapolation principles: (a) non-communicative aggregation (b) local social signaling and (c) institutional norms setting. The analysis identifies the emergence of sociotechnical lock-in, a macro-behavioral state where individually adaptive delegation, in the absence of communicative and institutional safeguards, aggregates into a systemic collective action problem modeled as a prisoner's dilemma that degrades shared epistemic standards. We argue that adoption under higher communicative standards and institutional norms can mitigate suboptimal collective equilibria by imposing social commitments on individual users.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

3 major / 2 minor

Summary. The paper develops a decision- and game-theoretic framework for the human-AI delegation-verification dilemma. It first identifies canonical individual strategies and their adaptive transitions to stable equilibria based on interaction feedback. It then scales these strategies to collective equilibria via three extrapolation principles—non-communicative aggregation, local social signaling, and institutional norms setting—thereby identifying the emergence of sociotechnical lock-in, a macro-level prisoner's dilemma that degrades shared epistemic standards in the absence of communicative and institutional safeguards. The paper concludes that higher communicative standards and institutional norms can mitigate these suboptimal collective outcomes.

Significance. If the modeling steps were made explicit and the scaling shown to be robust, the work would supply a useful conceptual bridge between individual human-AI interaction patterns and collective epistemic risks, offering a language for analyzing lock-in phenomena that could inform both system design and policy in hybrid intelligence settings.

major comments (3)
  1. [Abstract, §2–3] Abstract and §2–3: the central claim that individually stable delegation strategies aggregate into a prisoner's dilemma via the three extrapolation principles is asserted without any payoff matrix, transition rules, equilibrium derivation, or simulation output. The collective outcome is therefore not shown to follow from the stated individual-level modeling.
  2. [§4] §4: the three extrapolation principles are introduced as scaling mechanisms, yet no derivation demonstrates that non-communicative aggregation, local social signaling, and institutional norms setting alone suffice to produce the PD; the text provides no robustness check against omitted factors such as agent heterogeneity or interaction topology.
  3. [§5] §5: the mitigation argument (higher communicative standards and institutional norms) is presented as sufficient to escape the lock-in, but no formal condition or comparative equilibrium analysis is supplied showing how these interventions alter the collective payoff structure.
minor comments (2)
  1. [§2] Notation for strategy sets and transition functions is introduced informally; explicit definitions or a table would improve traceability.
  2. [Introduction] The manuscript would benefit from a short related-work subsection contrasting the proposed extrapolation principles with existing multi-agent or evolutionary game models of technology adoption.

Simulated Author's Rebuttal

3 responses · 0 unresolved

We thank the referee for the constructive and precise comments, which correctly identify points where the linkage between individual-level modeling and collective outcomes remains implicit. We agree that greater explicitness would strengthen the contribution and have prepared revisions to address each concern. Our responses below indicate the planned changes.

read point-by-point responses

  1. Referee: [Abstract, §2–3] Abstract and §2–3: the central claim that individually stable delegation strategies aggregate into a prisoner's dilemma via the three extrapolation principles is asserted without any payoff matrix, transition rules, equilibrium derivation, or simulation output. The collective outcome is therefore not shown to follow from the stated individual-level modeling.

    Authors: The manuscript offers a conceptual decision- and game-theoretic framework. Individual strategies and adaptive transitions are described qualitatively in §2–3, with collective aggregation argued via the extrapolation principles. We accept that no explicit payoff matrix, formal transition rules, or equilibrium derivation is supplied. In revision we will insert a schematic payoff matrix in §3 illustrating the collective PD and state the transition conditions as explicit qualitative rules; simulation output lies outside the paper's theoretical scope. revision: yes

  2. Referee: [§4] §4: the three extrapolation principles are introduced as scaling mechanisms, yet no derivation demonstrates that non-communicative aggregation, local social signaling, and institutional norms setting alone suffice to produce the PD; the text provides no robustness check against omitted factors such as agent heterogeneity or interaction topology.

    Authors: We agree that the sufficiency of the three principles is asserted conceptually without a step-by-step derivation or robustness analysis. The principles are presented as minimal mechanisms that generate the PD when other factors are absent. We will revise §4 to supply a logical derivation showing how each principle maps individual equilibria onto the collective PD and add a paragraph discussing robustness to heterogeneity and topology, treating these as open questions for subsequent formal work. revision: yes

  3. Referee: [§5] §5: the mitigation argument (higher communicative standards and institutional norms) is presented as sufficient to escape the lock-in, but no formal condition or comparative equilibrium analysis is supplied showing how these interventions alter the collective payoff structure.

    Authors: The mitigation claim is advanced at the level of altered social commitments rather than through comparative statics. We acknowledge the absence of formal conditions or equilibrium comparisons. In the revised §5 we will provide a qualitative comparative analysis of the payoff matrix under varying communicative standards and institutional norms, specifying the conditions under which the PD is transformed into a different game form. revision: yes

Circularity Check

1 steps flagged

Sociotechnical lock-in obtained by applying the three extrapolation principles as definitional scaling rules

specific steps
  1. self definitional [Abstract]
    "Second, we scale individually stable strategies to collective equilibria using three extrapolation principles: (a) non-communicative aggregation (b) local social signaling and (c) institutional norms setting. The analysis identifies the emergence of sociotechnical lock-in, a macro-behavioral state where individually adaptive delegation, in the absence of communicative and institutional safeguards, aggregates into a systemic collective action problem modeled as a prisoner's dilemma that degrades shared epistemic standards."

    The emergence of the lock-in state and its characterization as a prisoner's dilemma are presented as the direct output of applying the three extrapolation principles to the individual-level equilibria. The principles function as the authors' chosen scaling rules, so the collective PD outcome is equivalent to the input assumptions by construction.

full rationale

The paper first models individual decision-theoretic strategies reaching stable equilibria via adaptive trajectories and feedback. It then scales these to collective equilibria explicitly 'using three extrapolation principles' and identifies the resulting sociotechnical lock-in (modeled as a prisoner's dilemma) as the aggregate outcome. Because the lock-in state is defined as what follows from applying precisely those three principles (non-communicative aggregation, local social signaling, institutional norms setting) in the absence of safeguards, the macro result reduces to a restatement of the chosen aggregation assumptions rather than an independent derivation. No equations, robustness checks against heterogeneity or topology, or external validation are supplied to show the PD must emerge under only these rules. This constitutes self-definitional circularity at the central claim.

Axiom & Free-Parameter Ledger

0 free parameters · 2 axioms · 2 invented entities

The model rests on standard game-theoretic assumptions about strategy adaptation and equilibrium plus newly introduced conceptual entities without independent empirical grounding or falsifiable predictions.

axioms (2)
  • domain assumption Agents transition between strategies based on interaction feedback to reach stable equilibria.
    Invoked when mapping canonical decision-theoretic strategies and adaptive user trajectories.
  • domain assumption Individual strategies scale to collective equilibria via non-communicative aggregation, local social signaling, and institutional norms setting.
    Used to identify the emergence of sociotechnical lock-in from individual behavior.
invented entities (2)
  • sociotechnical lock-in no independent evidence
    purpose: Macro-behavioral state in which individual delegation aggregates into collective epistemic degradation.
    Introduced as the central emergent outcome of the model.
  • delegation-verification dilemma no independent evidence
    purpose: Individual choice problem between delegating to AI and verifying outputs.
    Frames the decision-theoretic component of the analysis.

pith-pipeline@v0.9.1-grok · 5746 in / 1564 out tokens · 37175 ms · 2026-06-30T17:05:26.401423+00:00 · methodology

Review history (2 revisions) →

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

64 extracted references · 35 canonical work pages · 2 internal anchors

  1. [1]

    Toward an Integrative Study of Human-AI In- teraction

    Mohammed Alsobay. “Toward an Integrative Study of Human-AI In- teraction”. Advisor: Abdullah Almaatouq. PhD thesis. Cambridge, MA: Massachusetts Institute of Technology, Sept. 2025.url:https://hdl. handle.net/1721.1/164569

    2025

  2. [2]

    Competing Technologies, Increasing Returns, and Lock- In by Historical Events

    W. Brian Arthur. “Competing Technologies, Increasing Returns, and Lock- In by Historical Events”. In:The Economic Journal99.394 (1989), pp. 116– 131.doi:10.2307/2234208

    work page doi:10.2307/2234208 1989

  3. [3]

    On Gen- eralized Urn Schemes of the P´ olya Kind

    W. Brian Arthur, Yuri M. Ermoliev, and Yuri M. Kaniovski. “On Gen- eralized Urn Schemes of the P´ olya Kind”. In:Cybernetics19.5 (1983), pp. 61–71.issn: 0011-4235.doi:10.1007/BF01068569

    work page doi:10.1007/bf01068569 1983

  4. [4]

    Aumann and Michael B

    Robert J. Aumann and Michael B. Maschler.Repeated Games with In- complete Information. MIT Press Classics. Cambridge, MA, USA: MIT Press, 1995

    1995

  5. [5]

    Robert Axelrod.The Evolution of Cooperation. Revised. New York, NY, USA: Basic Books, 2006.isbn: 9780465005644.url:https : / / www . amazon.com/Evolution- Cooperation- Revised- Robert- Axelrod/dp/ 0465005640

    2006

  6. [6]

    Oxford, Eng- land: Oxford University Press, 2016

    Nick Bostrom.Superintelligence: Paths, Dangers, Strategies. Oxford, Eng- land: Oxford University Press, 2016

    2016

  7. [7]

    Improving Human-AI Collaboration With Descriptions of AI Behavior

    ´Angel Alexander Cabrera, Adam Perer, and Jason I. Hong. “Improving Human-AI Collaboration With Descriptions of AI Behavior”. In:Pro- ceedings of the ACM on Human-Computer Interaction7.CSCW1 (2023), pp. 1–21.doi:10.1145/3579612

    work page doi:10.1145/3579612 2023

  8. [8]

    Self-Augmented Preference Alignment for Sycophancy Reduction in LLMs

    Chien Hung Chen, Hen-Hsen Huang, and Hsin-Hsi Chen. “Self-Augmented Preference Alignment for Sycophancy Reduction in LLMs”. In:Proceed- ings of the 2025 Conference on Empirical Methods in Natural Language Processing. Ed. by Christos Christodoulopoulos et al. Suzhou, China: As- sociation for Computational Linguistics, 2025, pp. 12379–12391.doi:10. 18653/v1...

    2025

  9. [9]

    Signaling Games and Stable Equilib- ria

    In-Koo Cho and David M. Kreps. “Signaling Games and Stable Equilib- ria”. In:Quarterly Journal of Economics102.2 (1987), pp. 179–221.doi: 10.2307/1885060. 22

    work page doi:10.2307/1885060 1987

  10. [10]

    Extending minds with generative AI

    Andy Clark. “Extending minds with generative AI”. In:Nature Commu- nications16 (2025), p. 4627.doi:10.1038/s41467-025-59906-9.url: https://www.nature.com/articles/s41467-025-59906-9

    work page doi:10.1038/s41467-025-59906-9.url: 2025

  11. [11]

    Oxford / New York: Oxford University Press, 2003

    Andy Clark.Natural-Born Cyborgs: Minds, Technologies, and the Future of Human Intelligence. Oxford / New York: Oxford University Press, 2003. isbn: 9780195148664

    2003

  12. [12]

    Oxford, UK: Oxford University Press, 2008

    Andy Clark.Supersizing the Mind: Embodiment, Action, and Cognitive Extension. Oxford, UK: Oxford University Press, 2008

    2008

  13. [13]

    Whatever Next? Predictive Brains, Situated Agents, and the Future of Cognitive Science

    Andy Clark. “Whatever Next? Predictive Brains, Situated Agents, and the Future of Cognitive Science”. In:Behavioral and Brain Sciences36.3 (2013), pp. 181–204.doi:10.1017/S0140525X12000477

    work page doi:10.1017/s0140525x12000477 2013

  14. [14]

    Clark, D

    Andy Clark and David J. Chalmers. “The Extended Mind”. In:Analysis 58.1 (1998), pp. 7–19.doi:10.1093/analys/58.1.7

    work page doi:10.1093/analys/58.1.7 1998

  15. [15]

    Strategic Information Transmis- sion

    Vincent P. Crawford and Joel Sobel. “Strategic Information Transmis- sion”. In:Econometrica50.6 (1982), pp. 1431–1451.doi:10.2307/1913390

    work page doi:10.2307/1913390 1982

  16. [16]

    How the Risk of Exploitation in Human-AI Re- lationships Depends on Relationship Type

    Brian D. Earp et al. “How the Risk of Exploitation in Human-AI Re- lationships Depends on Relationship Type”. In:SSRN Electronic Jour- nal(2025).doi:10 . 2139 / ssrn . 5288102.url:https : / / ssrn . com / abstract=5288102

    2025

  17. [17]

    Helping or herding? reward model ensembles mitigate but do not eliminate reward hacking.arXiv preprint arXiv:2312.09244, 2023

    Jacob Eisenstein et al. “Helping or Herding? Reward Model Ensembles Mitigate but do not Eliminate Reward Hacking”. In:arXiv preprint arXiv:2312.09244 (2023).doi:10.48550/arXiv.2312.09244.url:https://arxiv.org/ abs/2312.09244

    work page doi:10.48550/arxiv.2312.09244.url:https://arxiv.org/ 2023

  18. [18]

    Cheap Talk

    Joseph Farrell and Matthew Rabin. “Cheap Talk”. In:Journal of Eco- nomic Perspectives10.3 (1996), pp. 103–118.doi:10.1257/jep.10.3. 103

    work page doi:10.1257/jep.10.3 1996

  19. [19]

    2024.doi:10.48550/arXiv.2407.19098

    George Fragiadakis et al. “Evaluating Human-AI Collaboration: A Review and Methodological Framework”. In:arXiv preprint arXiv:2407.19098 (2024).doi:10.48550/arXiv.2407.19098.url:https://arxiv.org/ pdf/2407.19098

    work page doi:10.48550/arxiv.2407.19098.url:https://arxiv.org/ 2024

  20. [20]

    2024.doi:10.48550/arXiv.2407.19098

    George Fragiadakis et al.Evaluating Human-AI Collaboration: A Review and Methodological Framework. 2024.doi:10.48550/arXiv.2407.19098. arXiv:2407.19098 [cs.HC]

    work page doi:10.48550/arxiv.2407.19098 2024

  21. [21]

    Human–AI collaboration is not very collaborative yet: A taxonomy of interaction patterns in AI-assisted decision making from a systematic review

    Catalina Gomez et al. “Human–AI collaboration is not very collaborative yet: A taxonomy of interaction patterns in AI-assisted decision making from a systematic review”. In:Frontiers in Computer Science6 (2024), p. 1521066.doi:10.3389/fcomp.2024.1521066

    work page doi:10.3389/fcomp.2024.1521066 2024

  22. [22]

    Threshold Models of Collective Behavior

    Mark S. Granovetter. “Threshold Models of Collective Behavior”. In: American Journal of Sociology83.6 (1978), pp. 1420–1443.doi:10.1086/ 226707.url:https://doi.org/10.1086/226707. 23

    work page doi:10.1086/226707 1978

  23. [23]

    The Informational Role of Warranties and Private Disclosure about Product Quality

    Sanford J. Grossman. “The Informational Role of Warranties and Private Disclosure about Product Quality”. In:Journal of Law and Economics 24.3 (1981), pp. 461–483.doi:10.1086/466995

    work page doi:10.1086/466995 1981

  24. [24]

    J¨ urgen Habermas, Thomas MacCarthy, and J¨ urgen Habermas.Reason and the Rationalization of Society. Nachdr. The Theory of Communicative Action / J¨ urgen Habermas. Transl. by Thomas MacCarthy, Vol. 1. Boston: Beacon, 2007

    2007

  25. [25]

    Baltimore: Johns Hopkins University Press, 1982

    Russell Hardin.Collective Action. Baltimore: Johns Hopkins University Press, 1982

    1982

  26. [26]

    Games with Incomplete Information Played by “Bayesian

    John C. Harsanyi. “Games with Incomplete Information Played by “Bayesian” Players, Part I: The Basic Model”. In:Management Science14.3 (1967), pp. 159–182.doi:10.1287/mnsc.14.3.159

    work page doi:10.1287/mnsc.14.3.159 1967

  27. [27]

    Cambridge, MA, USA: MIT Press, 2009.isbn: 9780262582247.url:https://mitpress

    Joseph Heath.Communicative Action and Rational Choice. Cambridge, MA, USA: MIT Press, 2009.isbn: 9780262582247.url:https://mitpress. mit . edu / 9780262582247 / communicative - action - and - rational - choice/

    2009

  28. [28]

    Toronto, ON, Canada: HarperCollins, 2014

    Joseph Heath.Enlightenment 2.0: Restoring Sanity to Our Politics, Our Economy, and Our Lives. Toronto, ON, Canada: HarperCollins, 2014. isbn: 9781443422536.url:https://www.amazon.com/Enlightenment- 2-0-Joseph-Heath/dp/1443422533

    work page arXiv 2014

  29. [29]

    New York, NY: Oxford University Press, 2011

    Joseph Heath.Following the Rules. New York, NY: Oxford University Press, 2011

    2011

  30. [30]

    An Enactivist Approach to Human-Computer Interac- tion: Bridging the Gap Between Human Agency and Affordances

    Angjelin Hila. “An Enactivist Approach to Human-Computer Interac- tion: Bridging the Gap Between Human Agency and Affordances”. In: HCI International 2025 – Late Breaking Papers: 27th International Con- ference on Human-Computer Interaction, HCII 2025, Gothenburg, Swe- den, June 22–27, 2025, Proceedings, Part I. Ed. by Masaaki Kurosu and Ayako Hashizume. ...

    work page doi:10.1007/978-3-032-12657- 2025

  31. [31]

    The epistemological consequences of large language mod- els: Rethinking collective intelligence and institutional knowledge

    Angjelin Hila. “The epistemological consequences of large language mod- els: Rethinking collective intelligence and institutional knowledge”. In:AI & Society(2025).doi:10.1007/s00146-025-02426-3

    work page doi:10.1007/s00146-025-02426-3 2025

  32. [32]

    Moral Hazard and Observability

    Bengt Holmstr¨ om. “Moral Hazard and Observability”. In:Bell Journal of Economics10.1 (1979), pp. 74–91.doi:10.2307/3003320

    work page doi:10.2307/3003320 1979

  33. [33]

    The Influence of AI Literacy on User’s Trust in AI in Practical Scenarios: A Digital Divide Pilot Study

    Kuo Ting Huang and Christopher Ball. “The Influence of AI Literacy on User’s Trust in AI in Practical Scenarios: A Digital Divide Pilot Study”. In:Proceedings of the Association for Information Science and Technology 61.1 (2024), pp. 937–939.doi:10 . 1002 / pra2 . 1146.url:https : / / experts . illinois . edu / en / publications / the - influence - of - a...

    2024

  34. [34]

    Ar- tificial intelligence, human intelligence and hybrid intelligence based on mutual augmentation

    Mohammad Hossein Jarrahi, Christoph Lutz, and Gemma Newlands. “Ar- tificial intelligence, human intelligence and hybrid intelligence based on mutual augmentation”. In:Big Data & Society9.2 (2022), pp. 1–6.doi: 10.1177/20539517221142824.url:https://journals.sagepub.com/ doi/full/10.1177/20539517221142824

    work page doi:10.1177/20539517221142824.url:https://journals.sagepub.com/ 2022

  35. [35]

    Theory of the Firm: Man- agerial Behavior, Agency Costs and Ownership Structure

    Michael C. Jensen and William H. Meckling. “Theory of the Firm: Man- agerial Behavior, Agency Costs and Ownership Structure”. In:Journal of Financial Economics3.4 (1976), pp. 305–360.doi:10 . 1016 / 0304 - 405X(76)90026-X

    1976

  36. [36]

    Jiaming Ji et al.AI Alignment: A Comprehensive Survey. 2023. arXiv: 2310.19852 [cs.AI]

    work page internal anchor Pith review Pith/arXiv arXiv 2023

  37. [37]

    Bayesian Persuasion

    Emir Kamenica and Matthew Gentzkow. “Bayesian Persuasion”. In:Amer- ican Economic Review101.6 (2011), pp. 2590–2615.doi:10.1257/aer. 101.6.2590

    work page doi:10.1257/aer 2011

  38. [38]

    Cambridge, UK: Cambridge University Press, 1982.isbn: 9780521288843

    John Maynard Smith.Evolution and the Theory of Games. Cambridge, UK: Cambridge University Press, 1982.isbn: 9780521288843

    1982

  39. [39]

    Game Theory and the Evolution of Fighting

    John Maynard Smith. “Game Theory and the Evolution of Fighting”. In: On Evolution. Edinburgh, UK: Edinburgh University Press, 1972, pp. 8– 28

    1972

  40. [40]

    The Logic of Animal Con- flict

    John Maynard Smith and George R. Price. “The Logic of Animal Con- flict”. In:Nature246.5427 (1973), pp. 15–18.doi:10.1038/246015a0

    work page doi:10.1038/246015a0 1973

  41. [41]

    Cambridge, MA: MIT Press, 2010

    Richard Menary, ed.The Extended Mind. Cambridge, MA: MIT Press, 2010

    2010

  42. [42]

    Good News and Bad News: Representation Theorems and Applications

    Paul R. Milgrom. “Good News and Bad News: Representation Theorems and Applications”. In:Bell Journal of Economics12.2 (1981), pp. 380– 391.doi:10.2307/3003562

    work page doi:10.2307/3003562 1981

  43. [43]

    The Optimal Structure of Incentives and Author- ity Within an Organization

    James A. Mirrlees. “The Optimal Structure of Incentives and Author- ity Within an Organization”. In:Bell Journal of Economics7.1 (1976), pp. 105–131.doi:10.2307/3003186

    work page doi:10.2307/3003186 1976

  44. [44]

    4E cognition and the coevolution of human–AI interaction

    J¨ org Noller. “4E cognition and the coevolution of human–AI interaction”. In:Discover Artificial Intelligence5.1 (2025), p. 323.doi:10 . 1007 / s44163 - 025 - 00595 - 0.url:https : / / doi . org / 10 . 1007 / s44163 - 025-00595-0

    2025

  45. [45]

    Cambridge, MA, USA: Harvard University Press, 1965

    Mancur Olson.The Logic of Collective Action: Public Goods and the The- ory of Groups. Cambridge, MA, USA: Harvard University Press, 1965

    1965

  46. [46]

    Cambridge, UK: Cambridge University Press, 1990

    Elinor Ostrom.Governing the Commons: The Evolution of Institutions for Collective Action. Cambridge, UK: Cambridge University Press, 1990. isbn: 9780521405998.url:https : / / www . amazon . com / Governing - Commons-Evolution-Institutions-Collective/dp/0521405998. 25

    work page arXiv 1990

  47. [47]

    Hybrid collective intelligence in a human– AI society

    Marieke M. M. Peeters et al. “Hybrid collective intelligence in a human– AI society”. In:AI & SOCIETY36.1 (2021), pp. 217–238.doi:10.1007/ s00146-020-01005-y.url:https://doi.org/10.1007/s00146-020- 01005-y

    work page doi:10.1007/s00146-020- 2021

  48. [48]

    A Multifaceted Vision of the Human-AI Collaboration: A Comprehensive Review

    Maite Puerta-Beldarrain et al. “A Multifaceted Vision of the Human-AI Collaboration: A Comprehensive Review”. In:IEEE Access13 (2025), pp. 21876–21903.doi:10.1109/ACCESS.2025.3536095

    work page doi:10.1109/access.2025.3536095 2025

  49. [49]

    The Economic Theory of Agency: The Principal’s Prob- lem

    Stephen A. Ross. “The Economic Theory of Agency: The Principal’s Prob- lem”. In:American Economic Review63.2 (1973), pp. 134–139

    1973

  50. [50]

    Enough With

    Advait Sarkar. “Enough With ”Human-AI Collaboration””. In:Extended Abstracts of the 2023 CHI Conference on Human Factors in Computing Systems (CHI EA ’23)(2023), pp. 1–8.doi:10.1145/3544549.3582735

    work page doi:10.1145/3544549.3582735 2023

  51. [51]

    Savage.The Foundations of Statistics

    Leonard J. Savage.The Foundations of Statistics. New York: John Wiley & Sons, 1954.isbn: 9780486623499

    1954

  52. [52]

    Schelling.Micromotives and Macrobehavior

    Thomas C. Schelling.Micromotives and Macrobehavior. New York, NY, USA: W. W. Norton, 1978

    1978

  53. [53]

    Towards Understanding Sycophancy in Language Models

    Mrinank Sharma et al. “Towards Understanding Sycophancy in Language Models”. In:arXiv preprint arXiv:2310.13548(2023).doi:10 . 48550 / arXiv.2310.13548.url:https://arxiv.org/abs/2310.13548

    work page internal anchor Pith review Pith/arXiv arXiv 2023

  54. [54]

    Human-AI Interaction from an Evolutionary Neuroso- ciological Perspective

    Yulia S. Shkurko. “Human-AI Interaction from an Evolutionary Neuroso- ciological Perspective”. In:Handbook of Neurosociology. Ed. by David D. Franks et al. Handbooks of Sociology and Social Research. Springer Nature Switzerland, 2025, pp. 387–409.doi:10.1007/978-3-031-95615-7_21. url:https://doi.org/10.1007/978-3-031-95615-7_21

    work page doi:10.1007/978-3-031-95615-7_21 2025

  55. [55]

    Simon.The Sciences of the Artificial

    Herbert A. Simon.The Sciences of the Artificial. 3rd ed. Cambridge, MA: MIT Press, 1996

    1996

  56. [56]

    Cambridge, UK: Cambridge University Press, 1996.isbn: 9780521555839

    Brian Skyrms.The Evolution of the Social Contract. Cambridge, UK: Cambridge University Press, 1996.isbn: 9780521555839

    1996

  57. [57]

    The Quarterly Journal of Economics , volume =

    Michael Spence. “Job Market Signaling”. In:Quarterly Journal of Eco- nomics87.3 (1973), pp. 355–374.doi:10.2307/1882010

    work page doi:10.2307/1882010 1973

  58. [58]

    Evolutionary Stable Strategies and Game Dynamics

    Peter D. Taylor and Leo B. Jonker. “Evolutionary Stable Strategies and Game Dynamics”. In:Mathematical Biosciences40.1-2 (1978), pp. 145– 156.doi:10.1016/0025-5564(78)90077-9

    work page doi:10.1016/0025-5564(78)90077-9 1978

  59. [59]

    Thompson.Mind in Life

    E. Thompson.Mind in Life. London, England: Belknap Press, 2010

    2010

  60. [60]

    From Human–Human Collaboration to Human–AI Collaboration: Designing AI Systems That Can Work Together with Peo- ple

    Dakuo Wang et al. “From Human–Human Collaboration to Human–AI Collaboration: Designing AI Systems That Can Work Together with Peo- ple”. In:Extended Abstracts of the 2020 CHI Conference on Human Fac- tors in Computing Systems. 2020, pp. 1–6.doi:10 . 1145 / 3334480 . 3381069.url:https : / / dl . acm . org / doi / pdf / 10 . 1145 / 3334480 . 3381069

    2020

  61. [61]

    Joel Watson.Strategy: An Introduction to Game Theory. 3rd. New York: W. W. Norton & Company, 2013.isbn: 9780393918380. 26

    2013

  62. [62]

    Exploiting Human-AI Dependence for Learning to Defer

    Zixi Wei, Yuzhou Cao, and Lei Feng. “Exploiting Human-AI Dependence for Learning to Defer”. In:Proceedings of the 41st International Confer- ence on Machine Learning. Vol. 235. Proceedings of Machine Learning Research. PMLR, 2024, pp. 52201–52218.url:https : / / openreview . net/forum?id=9H7WzF6qjK

    2024

  63. [63]

    Original work published 1977

    Langdon Winner.Autonomous Technology: Technics-out-of-Control as a Theme in Political Thought. Original work published 1977. MIT Press, 1978

    1977

  64. [64]

    How Does Collective Self-esteem Influence Confrontational Be- haviors and Emotional Responses in Human-AI Competition?

    Li Zhu. “How Does Collective Self-esteem Influence Confrontational Be- haviors and Emotional Responses in Human-AI Competition?” In:Com- puters in Human Behavior: Artificial Humans3.1 (2025), p. 100155.doi: 10.1016/j.chbah.2025.100155.url:https://doi.org/10.1016/j. chbah.2025.100155. 27

    work page doi:10.1016/j.chbah.2025.100155.url:https://doi.org/10.1016/j 2025