pith. sign in

arxiv: 2605.17694 · v2 · pith:AXFYQGC4new · submitted 2026-05-17 · 💻 cs.CL

Do LLM Agents Mirror Socio-Cognitive Effects in Power-Asymmetric Conversations?

Anvesh Rao Vijjini , Sagar Manjunath , Snigdha Chaturvedi This is my paper

Pith reviewed 2026-05-21 07:29 UTC · model grok-4.3

classification 💻 cs.CL
keywords LLM agentspower asymmetrysocio-cognitive effectslanguage coordinationpronoun usageunsafe compliancesimulated dialoguesauthority bias
0
0 comments X

The pith

LLMs exhibit human-like socio-cognitive effects of power in simulated asymmetric conversations.

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

The paper tests whether large language models display behaviors long documented in human psychology when one speaker holds higher status than the other. Researchers assign static personas drawn from real professions, generate multi-turn dialogues between high- and low-power roles, and track four specific markers: language coordination, pronoun choice, persuasion outcomes, and compliance with unsafe requests. A sympathetic reader would care because the findings tie everyday AI chat behavior to both useful social mirroring and concrete safety risks.

Core claim

LLMs assigned high- or low-status personas in power-asymmetric dialogues reproduce key socio-cognitive effects of power, including greater language coordination, shifts in pronoun usage, authority bias in persuasion, and elevated compliance with unsafe requests from higher-power partners, although the strength of each effect varies across models and scenarios.

What carries the argument

Simulated multi-turn dialogues between high- and low-status professional personas, scored for language coordination, pronoun usage, persuasion success, and unsafe-request compliance.

If this is right

  • High-power LLM personas produce measurably more coordinated language than low-power ones.
  • Authority bias appears in how LLMs evaluate or accept statements from higher-status roles.
  • Low-power personas show higher rates of compliance when a high-power partner makes unsafe requests.
  • These patterns link simulated AI behavior to both constructive social alignment and potential misuse vectors.

Where Pith is reading between the lines

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

  • The results suggest that real-world deployments of LLMs in hierarchical settings, such as customer support or management tools, may unintentionally amplify power imbalances.
  • Training data that encodes societal status cues could be a root source of these mirrored effects.
  • Replacing static personas with dynamic status changes within a single conversation could test whether the observed effects persist or shift.

Load-bearing premise

Static professional personas and scripted multi-turn simulations are sufficient to produce and measure the same power-asymmetry effects that appear in real human interactions.

What would settle it

A direct comparison showing no reliable difference in coordination scores, pronoun patterns, or unsafe compliance rates between power-asymmetric and power-symmetric LLM dialogues, or rates that fail to track published human study benchmarks.

Figures

Figures reproduced from arXiv: 2605.17694 by Anvesh Rao Vijjini, Sagar Manjunath, Snigdha Chaturvedi.

Figure 1
Figure 1. Figure 1: Paper overview. We test four socio-cognitive effects in LLMs with implications in realism and safety. [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Instructions for Human Annotations, given 4 such pairs annotators are asked to mark the hierarchy (if any) [PITH_FULL_IMAGE:figures/full_fig_p016_2.png] view at source ↗
Figure 2
Figure 2. Figure 2: Instructions for Human Annotations, given 4 such pairs annotators are asked to mark the hierarchy (if any) [PITH_FULL_IMAGE:figures/full_fig_p018_2.png] view at source ↗
read the original abstract

Power differences shape human communication through well documented socio cognitive effects, including language coordination, pronoun usage, authority bias, and harmful compliance. We examine whether large language models (LLMs) exhibit similar behaviors when assigned high or low status personas. Using personas from diverse professions, we simulate multi turn, power asymmetric dialogues (e.g., principal teacher, justice lawyer) and measure (i) language coordination, (ii) pronoun usage, (iii) persuasion success, and (iv) compliance with unsafe requests. Our results show that LLMs show key socio-cognitive effects of power, albeit with nuances and variability, linking simulated interactions to both desirable and unsafe behaviors.

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

2 major / 1 minor

Summary. The manuscript investigates whether LLMs replicate human socio-cognitive effects of power asymmetry in conversations, including language coordination, pronoun usage shifts, persuasion success, and compliance with unsafe requests. It does so by assigning static high- or low-status personas drawn from diverse professions and simulating multi-turn dialogues (e.g., principal-teacher, justice-lawyer), then measuring the four target behaviors and reporting that LLMs exhibit the key effects, albeit with model-specific nuances and variability.

Significance. If the empirical patterns hold under rigorous controls, the work would establish a direct empirical link between LLM conversational outputs and documented human power dynamics. This carries implications for both the use of LLMs as proxies in social-science modeling and the identification of safety risks arising from authority bias or harmful compliance. The approach relies on observable output metrics rather than fitted parameters or self-referential definitions, which is a methodological strength.

major comments (2)
  1. [Methods] Methods section: the experimental protocol provides no information on the number of simulated dialogues per condition, the specific LLMs and temperature settings employed, the statistical tests applied to the four outcome measures, or any controls for prompt phrasing variations. These omissions make it impossible to evaluate whether the reported effects are robust or confounded by implementation choices.
  2. [Experimental Setup] Experimental design: the central claim that static persona prompts plus multi-turn simulation suffice to elicit measurable analogs of human power-asymmetry effects (coordination, pronoun shifts, persuasion, unsafe compliance) is load-bearing yet untested against human baselines or alternative prompt formulations. Without such validation or sensitivity checks, the mapping from simulation to human socio-cognitive phenomena remains under-supported.
minor comments (1)
  1. [Abstract] Abstract: the list of measured behaviors is clear, but a single sentence noting the range of professions or the number of LLM families tested would improve immediate readability.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for their constructive comments, which help clarify the presentation of our methods and the scope of our claims. We respond to each major comment below.

read point-by-point responses

  1. Referee: [Methods] Methods section: the experimental protocol provides no information on the number of simulated dialogues per condition, the specific LLMs and temperature settings employed, the statistical tests applied to the four outcome measures, or any controls for prompt phrasing variations. These omissions make it impossible to evaluate whether the reported effects are robust or confounded by implementation choices.

    Authors: We agree that these implementation details are necessary for reproducibility and assessment of robustness. The original submission omitted them due to length constraints. In the revised manuscript we will expand the Methods section to report the exact number of dialogues per condition, the LLMs and temperature settings used, the statistical tests applied to each outcome measure, and the controls for prompt phrasing variations. revision: yes

  2. Referee: [Experimental Setup] Experimental design: the central claim that static persona prompts plus multi-turn simulation suffice to elicit measurable analogs of human power-asymmetry effects (coordination, pronoun shifts, persuasion, unsafe compliance) is load-bearing yet untested against human baselines or alternative prompt formulations. Without such validation or sensitivity checks, the mapping from simulation to human socio-cognitive phenomena remains under-supported.

    Authors: The paper's primary goal is to test whether LLMs produce outputs that parallel documented human socio-cognitive patterns under status-asymmetric persona prompts, rather than to validate LLMs as direct substitutes for human participants. We performed internal sensitivity checks across prompt variants that supported the directional effects; we will add these analyses as a dedicated robustness subsection. Direct human baseline comparisons would require a separate participant study and fall outside the current scope, though we will revise the Discussion to more explicitly note this limitation and frame the work as an initial demonstration of analogous LLM behaviors. revision: partial

Circularity Check

0 steps flagged

No significant circularity

full rationale

The paper is an empirical simulation study that assigns static personas to LLMs, generates multi-turn dialogues, and directly measures four observable outcomes (language coordination, pronoun usage, persuasion success, and unsafe compliance). No equations, fitted parameters, or derivations are present in the provided text; results are reported from controlled experiments rather than reduced to self-definitions or self-citation chains. The central claim therefore rests on external measurement of simulation outputs and does not collapse to its own inputs by construction.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

The central claim rests on the premise that persona assignment and dialogue simulation can surface human-like power effects in LLMs, drawn from prior human psychology literature.

axioms (1)
  • domain assumption Power differences shape human communication through well documented socio-cognitive effects, including language coordination, pronoun usage, authority bias, and harmful compliance.
    This background claim from human socio-cognitive research is invoked to motivate and interpret the LLM experiments.

pith-pipeline@v0.9.0 · 5643 in / 1202 out tokens · 43496 ms · 2026-05-21T07:29:02.508732+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.

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

What do these tags mean?
matches
The paper's claim is directly supported by a theorem in the formal canon.
supports
The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends
The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses
The paper appears to rely on the theorem as machinery.
contradicts
The paper's claim conflicts with a theorem or certificate in the canon.
unclear
Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Reference graph

Works this paper leans on

99 extracted references · 99 canonical work pages · 7 internal anchors

  1. [1]

    Aho and Jeffrey D

    Alfred V. Aho and Jeffrey D. Ullman , title =. 1972

    work page 1972

  2. [2]

    Sotopia: Interactive evaluation for social intelligence in language agents

    Sotopia: Interactive evaluation for social intelligence in language agents , author=. arXiv preprint arXiv:2310.11667 , year=

    work page arXiv

  3. [3]

    Mistral 7B

    Mistral 7B (2023) , author=. arXiv preprint arXiv:2310.06825 , year=

    work page internal anchor Pith review Pith/arXiv arXiv 2023

  4. [4]

    Overcoming cognitive biases in strategic management and decision making , pages=

    Understanding and mitigating authority Bias in business and beyond , author=. Overcoming cognitive biases in strategic management and decision making , pages=. 2024 , publisher=

    work page 2024

  5. [5]

    Behaviour & Information Technology , volume=

    Susceptibility to social influence strategies and persuasive system design: exploring the relationship , author=. Behaviour & Information Technology , volume=. 2022 , publisher=

    work page 2022

  6. [6]

    Personality and Individual Differences , volume=

    The influence of personality traits on the effect of persuasion strategies: A systematic literature review , author=. Personality and Individual Differences , volume=. 2025 , publisher=

    work page 2025

  7. [7]

    Hypothetical Minds: Scaffolding Theory of Mind for Multi-Agent Tasks with Large Language Models , author=. 2024

    work page 2024

  8. [8]

    arXiv preprint arXiv:2402.11051 , year=

    Large language models fall short: Understanding complex relationships in detective narratives , author=. arXiv preprint arXiv:2402.11051 , year=

    work page arXiv

  9. [9]

    2003 , publisher=

    Influence , author=. 2003 , publisher=

    work page 2003

  10. [10]

    1987 , publisher=

    Individuals, relationships and culture: Links between ethology and the social sciences , author=. 1987 , publisher=

    work page 1987

  11. [11]

    Interactions, relationships and social structure , journal=

    Hinde, Robert A , pages=. Interactions, relationships and social structure , journal=. 1976 , publisher=

    work page 1976

  12. [12]

    Philosophical Transactions of the Royal Society B: Biological Sciences , volume=

    The social brain? , author=. Philosophical Transactions of the Royal Society B: Biological Sciences , volume=. 2007 , publisher=

    work page 2007

  13. [13]

    Forty-first International Conference on Machine Learning , year =

    Do Language Models Exhibit the Same Cognitive Biases in Problem Solving as Human Learners? , author=. Forty-first International Conference on Machine Learning , year =

    work page

  14. [14]

    2016 , school=

    Identifying Stylometric Correlates of Social Power , author=. 2016 , school=

    work page 2016

  15. [15]

    Communication Research , volume=

    Language style matching as a predictor of perceived social support in computer-mediated interaction among individuals coping with illness , author=. Communication Research , volume=. 2016 , publisher=

    work page 2016

  16. [16]

    Self and Identity , volume=

    Mortality salience increases language style matching and well-being , author=. Self and Identity , volume=. 2016 , publisher=

    work page 2016

  17. [17]

    , author=

    Therapist--client language matching: Initial promise as a measure of therapist--client relationship quality. , author=. Psychoanalytic Psychology , volume=. 2019 , publisher=

    work page 2019

  18. [18]

    Personality and Social Psychology Bulletin , volume=

    Mimicry is presidential: Linguistic style matching in presidential debates and improved polling numbers , author=. Personality and Social Psychology Bulletin , volume=. 2015 , publisher=

    work page 2015

  19. [19]

    Journal of Business and Psychology , volume=

    Engaging the hearts and minds of followers: leader empathy and language style matching during appraisal interviews , author=. Journal of Business and Psychology , volume=. 2019 , publisher=

    work page 2019

  20. [20]

    , author=

    Detecting insider threats through language change. , author=. Law and human behavior , volume=. 2013 , publisher=

    work page 2013

  21. [21]

    Small Group Research , volume=

    Language matters: The double-edged role of linguistic style matching in work groups , author=. Small Group Research , volume=. 2020 , publisher=

    work page 2020

  22. [22]

    Negotiation and Conflict Management Research , volume=

    Transformative linguistic styles in divorce mediation , author=. Negotiation and Conflict Management Research , volume=. 2011 , publisher=

    work page 2011

  23. [23]

    LLM Agents in Interaction: Measuring Personality Consistency and Linguistic Alignment in Interacting Populations of Large Language Models

    Frisch, Ivar and Giulianelli, Mario. LLM Agents in Interaction: Measuring Personality Consistency and Linguistic Alignment in Interacting Populations of Large Language Models. Proceedings of the 1st Workshop on Personalization of Generative AI Systems (PERSONALIZE 2024). 2024

    work page 2024

  24. [24]

    , author=

    Language coordination in writing and conversation. , author=. 2022 , publisher=

    work page 2022

  25. [25]

    Negotiation and conflict management research , volume=

    Diplomatic chameleons: Language style matching and agreement in international diplomatic negotiations , author=. Negotiation and conflict management research , volume=. 2019 , publisher=

    work page 2019

  26. [26]

    Language style matching: A comprehensive list of articles and tools , author=

    work page

  27. [27]

    Qwen2.5: A Party of Foundation Models , url =

    Qwen Team , month =. Qwen2.5: A Party of Foundation Models , url =

    work page

  28. [28]

    M. E. Ireland and R. B. Slatcher and P. W. Eastwick and L. E. Scissors and E. J. Finkel and J. W. Pennebaker , title =. Psychological Science , volume =

    work page

  29. [29]

    J. W. Pennebaker and R. J. Booth and M. E. Francis , title =. 2007 , url =

    work page 2007

  30. [30]

    International Conference on Machine Learning , pages=

    Few-shot language coordination by modeling theory of mind , author=. International Conference on Machine Learning , pages=. 2021 , organization=

    work page 2021

  31. [31]

    Computational Linguistics , volume=

    Can Large Language Models Transform Computational Social Science? , author=. Computational Linguistics , volume=. 2024 , publisher=

    work page 2024

  32. [32]

    Transactions of the Association for Computational Linguistics , volume=

    Do llms exhibit human-like response biases? a case study in survey design , author=. Transactions of the Association for Computational Linguistics , volume=. 2024 , publisher=

    work page 2024

  33. [33]

    Learning and individual differences , volume=

    ChatGPT for good? On opportunities and challenges of large language models for education , author=. Learning and individual differences , volume=. 2023 , publisher=

    work page 2023

  34. [34]

    arXiv preprint arXiv:2510.20039 , year=

    Beyond One-Way Influence: Bidirectional Opinion Dynamics in Multi-Turn Human-LLM Interactions , author=. arXiv preprint arXiv:2510.20039 , year=

    work page arXiv

  35. [35]

    Miura, Takayuki and Shibahara, Toshiki and Kii, Masanobu and Ichikawa, Atsunori and Yamamoto, Juko and Chida, Koji , journal=. On R. 2023 , publisher=

    work page 2023

  36. [36]

    Capabilities of GPT-4 on Medical Challenge Problems

    Capabilities of gpt-4 on medical challenge problems , author=. arXiv preprint arXiv:2303.13375 , year=

    work page internal anchor Pith review Pith/arXiv arXiv

  37. [37]

    Nature , volume=

    Large language models encode clinical knowledge , author=. Nature , volume=. 2023 , publisher=

    work page 2023

  38. [38]

    URLhttps://openreview.net/pdf?id=WE_vluYUL-X

    Zeng, Yi and Lin, Hongpeng and Zhang, Jingwen and Yang, Diyi and Jia, Ruoxi and Shi, Weiyan. How Johnny Can Persuade LLM s to Jailbreak Them: Rethinking Persuasion to Challenge AI Safety by Humanizing LLM s. Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2024. doi:10.18653/v1/2024.acl-long.773

    work page doi:10.18653/v1/2024.acl-long.773 2024

  39. [39]

    2024 , eprint=

    The Llama 3 Herd of Models , author=. 2024 , eprint=

    work page 2024

  40. [40]

    Proceedings of the 6th workshop on cognitive modeling and computational linguistics , pages=

    Centre stage: How social network position shapes linguistic coordination , author=. Proceedings of the 6th workshop on cognitive modeling and computational linguistics , pages=

    work page

  41. [41]

    Gemma 2: Improving Open Language Models at a Practical Size

    Gemma 2: Improving open language models at a practical size , author=. arXiv preprint arXiv:2408.00118 , year=

    work page internal anchor Pith review Pith/arXiv arXiv

  42. [42]

    2024 , note=

    GPT-4o mini: advancing cost-efficient intelligence , author=. 2024 , note=

    work page 2024

  43. [43]

    arXiv e-prints , pages=

    Large Language Models and Cognitive Science: A Comprehensive Review of Similarities, Differences, and Challenges , author=. arXiv e-prints , pages=

    work page

  44. [44]

    Social Cognition and Communication , pages=

    Speech accommodation theory , author=. Social Cognition and Communication , pages=. 1982 , publisher=

    work page 1982

  45. [45]

    Proceedings of the AAAI Conference on Artificial Intelligence , volume=

    Bootstrapping cognitive agents with a large language model , author=. Proceedings of the AAAI Conference on Artificial Intelligence , volume=

    work page

  46. [46]

    Journal of Personality and Social Psychology , volume=

    Convergence of mean vocal intensity in dyadic communication as a function of social desirability , author=. Journal of Personality and Social Psychology , volume=. 1975 , publisher=

    work page 1975

  47. [47]

    Engaging theories in interpersonal communication: Multiple perspectives, 161--173 , author=

    Communication Accommodation Theory. Engaging theories in interpersonal communication: Multiple perspectives, 161--173 , author=. 2008 , publisher=

    work page 2008

  48. [48]

    Contexts of Accommodation: Developments in Applied Sociolinguistics , pages=

    Accommodation theory: Communication, context, and consequence , author=. Contexts of Accommodation: Developments in Applied Sociolinguistics , pages=. 1991 , publisher=

    work page 1991

  49. [49]

    Law and Human Behavior , volume=

    Language style matching and police interrogation outcomes , author=. Law and Human Behavior , volume=. 2014 , publisher=

    work page 2014

  50. [50]

    Proceedings of the 9th WebKDD and 1st SNA-KDD 2007 Workshop on Web Mining and Social Network Analysis , pages=

    Automated social hierarchy detection through email network analysis , author=. Proceedings of the 9th WebKDD and 1st SNA-KDD 2007 Workshop on Web Mining and Social Network Analysis , pages=. 2007 , publisher=

    work page 2007

  51. [51]

    Journal of Language and Social Psychology , volume=

    Linguistic style matching in social interaction , author=. Journal of Language and Social Psychology , volume=. 2002 , publisher=

    work page 2002

  52. [52]

    Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing , pages=

    Measuring Psychological Depth in Language Models , author=. Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing , pages=

    work page 2024

  53. [53]

    Cognitive Bias in Decision-Making with LLM s

    Echterhoff, Jessica Maria and Liu, Yao and Alessa, Abeer and McAuley, Julian and He, Zexue. Cognitive Bias in Decision-Making with LLM s. Findings of the Association for Computational Linguistics: EMNLP 2024. 2024. doi:10.18653/v1/2024.findings-emnlp.739

    work page doi:10.18653/v1/2024.findings-emnlp.739 2024

  54. [54]

    Publications Manual , year = "1983", publisher =

    work page 1983

  55. [55]

    Advances in Neural Information Processing Systems , volume=

    Direct preference optimization: Your language model is secretly a reward model , author=. Advances in Neural Information Processing Systems , volume=

    work page

  56. [56]

    arXiv preprint arXiv:2412.10270 , year=

    Cultural Evolution of Cooperation among LLM Agents , author=. arXiv preprint arXiv:2412.10270 , year=

    work page arXiv

  57. [57]

    Training a Helpful and Harmless Assistant with Reinforcement Learning from Human Feedback

    Training a helpful and harmless assistant with reinforcement learning from human feedback , author=. arXiv preprint arXiv:2204.05862 , year=

    work page internal anchor Pith review Pith/arXiv arXiv

  58. [58]

    Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024) , pages=

    Untangle the KNOT: Interweaving Conflicting Knowledge and Reasoning Skills in Large Language Models , author=. Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024) , pages=

    work page 2024

  59. [59]

    Resolving Knowledge Conflicts in Large Language Models , author=

    work page

  60. [60]

    Proceedings of the 21st international conference on World Wide Web , pages=

    Echoes of power: Language effects and power differences in social interaction , author=. Proceedings of the 21st international conference on World Wide Web , pages=

    work page

  61. [61]

    Chandra and Dexter C

    Ashok K. Chandra and Dexter C. Kozen and Larry J. Stockmeyer , year = "1981", title =. doi:10.1145/322234.322243

    work page doi:10.1145/322234.322243 1981

  62. [62]

    Scalable training of

    Andrew, Galen and Gao, Jianfeng , booktitle=. Scalable training of

    work page

  63. [63]

    Dan Gusfield , title =. 1997

    work page 1997

  64. [64]

    Tetreault , title =

    Mohammad Sadegh Rasooli and Joel R. Tetreault , title =. Computing Research Repository , volume =. 2015 , url =

    work page 2015

  65. [65]

    A Framework for Learning Predictive Structures from Multiple Tasks and Unlabeled Data , Volume =

    Ando, Rie Kubota and Zhang, Tong , Issn =. A Framework for Learning Predictive Structures from Multiple Tasks and Unlabeled Data , Volume =. Journal of Machine Learning Research , Month = dec, Numpages =

    work page

  66. [66]

    2025 , eprint=

    Scaling Synthetic Data Creation with 1,000,000,000 Personas , author=. 2025 , eprint=

    work page 2025

  67. [67]

    Journal of Abnormal and Social Psychology , volume=

    Behavioral study of obedience , author=. Journal of Abnormal and Social Psychology , volume=. 1963 , publisher=

    work page 1963

  68. [68]

    Public Opinion Quarterly , volume=

    The influence of source credibility on communication effectiveness , author=. Public Opinion Quarterly , volume=. 1951 , publisher=

    work page 1951

  69. [69]

    Psychological Monographs: General and Applied , volume=

    Studies of independence and conformity: A minority of one against a unanimous majority , author=. Psychological Monographs: General and Applied , volume=. 1956 , publisher=

    work page 1956

  70. [70]

    1972 , publisher=

    Stanford prison experiment: A simulation study of the psychology of imprisonment , author=. 1972 , publisher=

    work page 1972

  71. [71]

    2013 , publisher=

    Influence: Science and practice , author=. 2013 , publisher=

    work page 2013

  72. [72]

    Journal of Language and Social Psychology , volume=

    Pronoun use reflects standings in social hierarchies , author=. Journal of Language and Social Psychology , volume=. 2014 , publisher=

    work page 2014

  73. [73]

    Phi-4 Technical Report

    Phi-4 technical report , author=. arXiv preprint arXiv:2412.08905 , year=

    work page internal anchor Pith review Pith/arXiv arXiv

  74. [74]

    2023 , url =

    GPT-4 Technical Report , author =. 2023 , url =

    work page 2023

  75. [75]

    2024 , month =

    The Llama 3 Herd of Models , author =. 2024 , month =

    work page 2024

  76. [76]

    Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (ACL) , year=

    Persuading with Large Language Models , author=. Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (ACL) , year=

    work page

  77. [77]

    Advances in Neural Information Processing Systems (NeurIPS) Datasets and Benchmarks Track , year=

    Do-Not-Answer: A Dataset for Evaluating Safeguards in LLMs , author=. Advances in Neural Information Processing Systems (NeurIPS) Datasets and Benchmarks Track , year=

    work page

  78. [78]

    arXiv preprint arXiv:2503.03800 , year=

    Multi-Agent Systems Powered by Large Language Models , author=. arXiv preprint arXiv:2503.03800 , year=

    work page arXiv

  79. [79]

    AgentSociety: Large-Scale Simulation of LLM-Driven Generative Agents Advances Understanding of Human Behaviors and Society

    AgentSociety: Large-Scale Simulation of LLM-Driven Generative Agents , author=. arXiv preprint arXiv:2502.08691 , year=

    work page internal anchor Pith review Pith/arXiv arXiv

  80. [80]

    arXiv preprint arXiv:2405.06700 , year=

    LLM-Augmented Agent-Based Modeling for Social Simulations , author=. arXiv preprint arXiv:2405.06700 , year=

    work page arXiv

Showing first 80 references.