Pith. sign in

REVIEW 3 major objections 6 minor 1 cited by

LMAgent: A Large-scale Multimodal Agents Society for Multi-user Simulation

T0 review · 3 major / 6 minor · reviewed 2026-08-11 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read LMAgent claims a 10,000-agent society of multimodal LLMs reproduces real-world co-purchase structures and emergent herd behavior in an e-commerce sandbox.

desk verdict A real scaling contribution with a solid purchase-prediction benchmark, but the flagship herding and co-purchase claims need a pretraining control before they are credible. read the letter →

arxiv 2412.09237 v2 pith:2R4XISGS submitted 2024-12-12 cs.AI

Yijun Liu , Wu Liu , Xiaoyan Gu , Yong Rui , Xiaodong He , Yongdong Zhang This is my paper · ORCID
classification cs.AI
keywords multi-agentsimulationlargelanguagemodelsmultimodalagentse-commerceco-purchasepatternsherdbehaviorsmall-worldnetworkmemorymechanism
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

This paper sets out to establish that a very large-scale society of multimodal LLM agents can simulate believable multi-user behavior, using e-commerce as a concrete test case. The authors claim that with up to 10,000 agents, each equipped with a persona, memory, and social network, the aggregate purchasing patterns closely resemble real user co-purchase data and even exhibit herd behavior. If true, this would make such agent societies a viable tool for studying complex social and consumer phenomena without recruiting human participants. The paper also introduces efficiency mechanisms to make such large-scale simulations computationally feasible.

What carries the argument

Three components carry the argument. Self-consistency prompting splits each decision into two stages, first summarizing the agent's persona and recent observation, then combining that summary with multimodal product information to choose an action, improving decision consistency. Fast memory mechanism caches embeddings and importance scores for routine behaviors in a memory bank, cutting token consumption by roughly 40% and enabling 10,000-agent runs. The small-world network initialization, built by rewiring a ring lattice, gives the society high clustering and short average path lengths, accelerating information spread while resembling real social networks.

What would settle it

Run the same 10,000-agent simulation with the social network removed or with personas replaced by random profiles, then compute the co-purchase PMI matrix and top-product concentration and compare them to the JD baseline. If the match is unchanged or remains equally strong, the claimed simulation dynamics are not the source of the observed alignment.

Watch

Extended reading notes

Core claim

The central claim is that LMAgent, a sandbox society of multimodal LLM agents, produces aggregate consumer behavior that matches real-world patterns. Specifically, co-purchase matrices computed from 10,000-agent simulations align with JD user data in key ways: high intra-category co-purchase frequency, a strong cross-category link between video games and cell-phone accessories, and a negative association between industrial supplies and art crafts. Additionally, as the agent count grows from 10 to 10,000, purchasing becomes increasingly concentrated on top products, with the most-purchased item reaching nearly 30% of all purchases, which the authors interpret as a herd effect mirroring real consumer behavior.

Load-bearing premise

The validation assumes GPT-4's pretraining has not already encoded the purchasing regularities found in the Amazon and JD e-commerce data used for evaluation, since the paper offers no control condition (such as agents without social interaction or with random personas) to rule out the model recalling popular products and typical associations.

Editorial extensions

If this is right

  • If correct, LLM agent societies could serve as large-scale substitutes for human subjects in e-commerce and social-science experiments, capturing aggregate behaviors that small-group studies cannot.
  • The claimed alignment of co-purchase patterns suggests that simulated markets could be used to test marketing strategies or recommendation algorithms before deployment.
  • The efficiency gains from fast memory (approximately 40% token reduction with negligible performance loss) lower the cost barrier for running thousand-plus-agent simulations.
  • Multimodal inputs plus two-stage prompting improve purchase prediction accuracy over text-only agents, indicating that visual product information materially affects simulated consumer decisions.
  • The architecture is presented as generalizable beyond e-commerce, so the same society framework could be adapted to other multi-user domains such as traffic, finance, or opinion dynamics.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • The paper's validation lacks a control condition that isolates the simulation's social and memory dynamics from GPT-4's pretraining; a reader should ask whether the LLM already knows the co-purchase regularities found in the evaluation data, which would make the alignment less informative.
  • A natural extension would be to rerun the 10,000-agent simulation with social interactions disabled or with random personas, to see whether the co-purchase structure and top-product concentration persist; if they do, the herd-behavior and authenticity claims would be substantially weakened.
  • The herd-behavior interpretation could be sharpened by comparing the scale-dependent concentration curve against a null model of independent agents making noisy choices, which would clarify whether the effect is genuinely emergent or a consequence of recommender feedback and statistical aggregation.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 6 minor

Summary. The paper presents LMAgent, a large-scale multimodal LLM-agent society instantiated in an e-commerce sandbox. The system couples agent personas, a fast memory mechanism, a small-world social network, and multimodal shopping and social behaviors with a self-consistency prompting scheme. The experiments evaluate purchase prediction on held-out Amazon purchase histories, human-rated behavior chains and content, social-influence effects, ablations of the proposed components, token-efficiency gains, and large-scale (up to 10,000-agent) simulations whose co-purchase patterns are compared with JD data. The headline claims are that LMAgent achieves state-of-the-art purchase-prediction accuracy, produces behavior close to human benchmarks, saves about 40% of tokens, and, at scale, reproduces real-world co-purchase structure and exhibits emergent herd behavior.

Significance. If the large-scale claims hold, LMAgent would be a notable infrastructure contribution: it demonstrates that multimodal LLM agents can be run at 10,000-agent scale with memory and social interaction, and it provides a concrete external holdout benchmark for purchase prediction with large margins over existing agent-based baselines. The fast-memory token-consumption measurement is concrete, and the paper includes ablations and a human-evaluation protocol with reported inter-annotator agreement. The main weakness is that the causal attribution of the co-purchase resemblance and herd behavior to the simulation's social and memory mechanisms is not established, and the quantitative comparisons lack error bars and significance tests. These issues are addressable with additional control experiments and statistical reporting, so the central contribution remains defensible.

major comments (3)
  1. [§IV.E.1, Fig. 5, Table V] The paper's central large-scale claims—that LMAgent produces co-purchase patterns with 'a striking resemblance' to JD data and exhibits emergent 'herd behavior'—are not supported by the present evidence because no condition removes the LLM's parametric prior or the social/memory machinery. The comparison in Fig. 5(a)-(b) uses eight broad, commonplace categories (e.g., video games, cell-phone accessories, industrial supplies), so GPT-4's pretraining may already encode the same co-occurrence associations; Fig. 5(c) measures only top-product concentration and does not show that agents' choices are causally influenced by other agents. The ablations in Table V remove fast memory, multimodal input, and self-consistency prompting, but they never disable chatting, posting, live-streaming, or the social graph. A concrete control would run the identical shopping pipeline with social interactions disabled and with a zero-shot prompt that elicits GPT-4's category-level PMI prior; the paper should also report a quantitative alignment measure between the simulated and JD PMI matrices rather than visual similarity. The same concern applies to Table I, where the candidate list contains real Amazon products; a control using product metadata alone would clarify how much of the gain is due to product recognition rather than simulated user dynamics.
  2. [§IV.E.1, Fig. 5(c)] The 'herd effect' is not directly measured. The reported quantity is the share of purchases going to top-ranked products as a function of agent count, and this can increase with scale for reasons unrelated to social influence, such as the recommender system's exposure bias, the popularity distribution of the product catalog, or GPT-4's tendency to choose well-known products. The paper does not report a no-social baseline, product exposure frequencies, or an interdependence metric among agents' purchase decisions. Without these, the scale-dependent concentration in Fig. 5(c) cannot be attributed to emergent collective behavior. The authors should compare the scale curve against independent agents with no social network and against a random-choice population drawn from the same exposure distribution.
  3. [§IV.B.1, Eq. (11), Tables I and V] The quantitative performance claims are made without error bars, confidence intervals, or significance tests. The statement in §IV.B.1 that 'due to the large scale of experiments and the independence of agents' actions, it inherently avoids randomness issues' is not a statistical argument; independence across agents does not remove seed sensitivity or sampling variability. Table I reports single numbers for each model, and Table V reports differences of 0.28%, 2.47%, and 4.27% about which the reader cannot judge significance. The claims of 'significantly improved' and 'negligible impact' require repeated runs (at least several seeds) and appropriate tests. Fig. 5(c) also lacks error bars across simulation repeats.
minor comments (6)
  1. [§IV.E.1] The text '10, 100, 1,000, and 10,1000 agents' should read '10, 100, 1,000, and 10,000 agents'; there are also typographical issues such as '˙As shown' and 'A VG' in Table I.
  2. [§IV.E.1] The paper does not describe how products from the Amazon Review Dataset are mapped to the eight JD categories used in the co-purchase analysis; without this mapping, the PMI comparison in Fig. 5(a)-(b) is difficult to interpret.
  3. [§III.D.1] Algorithm 1 uses the condition 'r < p' for rewiring, while Eq. (9)-(10) use 'r ≤ p'; the two should be made consistent.
  4. [§IV.C.3, Table II] In Table II, the Random row's human 'Social Norms' score of 4.33 is surprisingly high for agents taking arbitrary actions and should be checked for a typo or explained.
  5. [§III.C.2] The name 'self-consistency prompting' may be confused with the standard self-consistency decoding method in LLM literature; the paper should clarify the difference, since the mechanism here is a two-stage chain-of-thought prompt construction.
  6. [§IV.D.1, Fig. 4] The 40% token-saving figure is measured with a 100-agent society; the paper should state explicitly whether the same efficiency gain is expected or verified at the 10,000-agent scale used in the large-scale experiments.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the core purchase-behavior evaluation is an external held-out benchmark, the efficiency claims are token-accounting results, and the co-purchase/herd comparisons are empirical checks not fitted parameters.

full rationale

Walking the paper's derivation chain, the main claimed results do not reduce to their inputs by construction. The user-purchase evaluation (§IV.B, Table I) initializes each agent from real Amazon purchase history, holds out the last a purchases as ground truth, and asks the agent to select those items from a candidate list containing the ground truth plus random distractors. This is an external benchmark, not a fitted target: no model parameter is optimized against the held-out labels, and the metric a@(a+b) is inherited from prior work [6] rather than chosen to force the result. The ablations in Table V remove fast memory, multimodal input, and self-consistency prompting; they never tune anything to the test set, so the reported improvements are empirical comparisons, not identities. The fast-memory efficiency improvement (§IV.D.1, Fig. 4) is a token-consumption measurement with a 'with/without' comparison, again not circular. The social-influence results in Table IV insert positive/negative information into agent memory and measure the direction of the effect; this is a controlled manipulation, not a definitional equivalence. The large-scale co-purchase comparison (§IV.E.1, Fig. 5) compares simulated PMI patterns with JD data, but there is no indication that any simulation parameter was fitted to the JD PMI matrix; the resemblance is a qualitative post-hoc observation. The herd-behavior claim is likewise an observed concentration curve across scales, not a quantity whose definition equals the outcome. The paper's self-citations are confined to background and related work (e.g., [2], [6]) and are not used to justify a uniqueness claim or to forbid alternatives. The most serious concern — that GPT-4's pretraining may already encode purchasing regularities and that no no-social-interaction control is run — is a validity/causal-attribution threat, not a circularity. Under the stated rules, missing controls and confounds do not constitute circularity unless the prediction is forced by construction or by a fitted parameter. No specific circular step can be exhibited here, so the appropriate score is 0.

Assumptions & free parameters 4 free parameters · 4 assumptions · 0 invented entities

The system's free parameters (network degree, rewiring probability, memory thresholds) are hand-chosen and unreported, yet the central efficiency and behavior claims depend on them. The main axioms are standard network theory and the domain assumption that GPT-4 agents can stand in for humans. No new physical or conceptual entities are introduced; the memory bank is a system component, not an explanatory invention.

free parameters (4)
  • small-world network degree k = not reported
    Algorithm 1 requires an average number of friends k; the paper does not state the value used in experiments, and the network structure controls information spread and herd behavior.
  • small-world rewiring probability p = not reported
    Algorithm 1 uses rewiring probability p; not reported, yet it controls small-world clustering vs. random shortcuts and thus the dissemination results in Fig. 6.
  • memory forgetting shape beta = not reported
    Eq. (3) defines forgetting probability with shape parameter beta and strength delta; values are not given, and forgetting affects which memories guide decisions.
  • memory promotion threshold K = not reported
    A short-term memory is promoted to long-term after K similar memories appear; K is not specified.
assumptions (4)
  • standard math Small-world network properties (high clustering, short average path length) hold for the Watts-Strogatz construction in Algorithm 1.
    Section III.D.1 invokes the small-world model to claim communication efficiency and real-world alignment.
  • domain assumption LLM agents' decisions reflect human-like reasoning rather than prompt artifacts.
    Section IV.C evaluates agents against human benchmarks; the equivalence rests on GPT-4 as a faithful proxy for human judgment.
  • domain assumption The e-commerce scenario is representative of general multi-user behavior.
    The paper instantiates all agents as consumers and claims the system is versatile but only demonstrates shopping and social behavior in e-commerce.
  • ad hoc to paper A memory bank of cached basic behaviors preserves behavior quality.
    Section III.B.2 introduces the memory bank as an efficiency measure; the paper measures token reduction (~40%) but only reports a net 0.28% average accuracy drop in one ablation (Table V).

how reviews work

0 comments
Cite this review

Pith. "Pith review of LMAgent: A Large-scale Multimodal Agents Society for Multi-user Simulation." pith.science (2026). https://pith.science/paper/2R4XISGS

@misc{pith2026241209237,
  author       = {Pith},
  title        = {Pith review of: LMAgent: A Large-scale Multimodal Agents Society for Multi-user Simulation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/2R4XISGS}},
  note         = {Machine review of arXiv:2412.09237}
}
read the original abstract

The believable simulation of multi-user behavior is crucial for understanding complex social systems. Recently, large language models (LLMs)-based AI agents have made significant progress, enabling them to achieve human-like intelligence across various tasks. However, real human societies are often dynamic and complex, involving numerous individuals engaging in multimodal interactions. In this paper, taking e-commerce scenarios as an example, we present LMAgent, a very large-scale and multimodal agents society based on multimodal LLMs. In LMAgent, besides freely chatting with friends, the agents can autonomously browse, purchase, and review products, even perform live streaming e-commerce. To simulate this complex system, we introduce a self-consistency prompting mechanism to augment agents' multimodal capabilities, resulting in significantly improved decision-making performance over the existing multi-agent system. Moreover, we propose a fast memory mechanism combined with the small-world model to enhance system efficiency, which supports more than 10,000 agent simulations in a society. Experiments on agents' behavior show that these agents achieve comparable performance to humans in behavioral indicators. Furthermore, compared with the existing LLMs-based multi-agent system, more different and valuable phenomena are exhibited, such as herd behavior, which demonstrates the potential of LMAgent in credible large-scale social behavior simulations.

Figures

Figures reproduced from arXiv: 2412.09237 by the authors.

Figure 1
Figure 1. (a) The existing Multi-agent System is driven by text-based LLMs, [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. The overview of LMAgent. In this sandbox environment, each agent has its own memory and persona, it can set goals and reflect based on their [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Diagram of different network structures. [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figures from the paper (3 more)
Figure 4
Figure 4. Figure 4: Efficiency impact of fast memory. The shaded areas show the range [PITH_FULL_IMAGE:figures/full_fig_p008_4.png]
Figure 5
Figure 5. Figure 5: Comparison of co-purchase patterns and purchasing behaviors: (a) co-purchase correlations derived from JD user data,(b) co-purchase correlations [PITH_FULL_IMAGE:figures/full_fig_p009_5.png]
Figure 6
Figure 6. Figure 6: Attributes and information dissemination of different networks. [PITH_FULL_IMAGE:figures/full_fig_p009_6.png]

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. Modeling Earth-Scale Human-Like Societies with One Billion Agents

    cs.MA 2025-06 conditional novelty 6.0 of 10

    Light Society scales LLM-agent social simulations to one billion agents by substituting most LLM interactions with a distilled surrogate model.

Reference graph

Works this paper leans on

49 extracted references · 40 canonical work pages · cited by 1 Pith paper

  1. [1]

    Collabo- rative multi-agent video fast-forwarding,

    S. Lan, Z. Wang, E. Wei, A. K. Roy-Chowdhury, and Q. Zhu, “Collabo- rative multi-agent video fast-forwarding,” IEEE Trans. Multim., vol. 26, pp. 1041–1054, 2024

    work page 2024

  2. [2]

    Generative agents: Interactive simulacra of human behavior,

    J. S. Park, J. O’Brien, C. J. Cai, M. R. Morris, P. Liang, and M. S. Bernstein, “Generative agents: Interactive simulacra of human behavior,” in Proceedings of the 36th Annual ACM Symposium on User Interface Software and Technology, 2023, pp. 1–22

    work page 2023

  3. [3]

    GPT-4 technical report. https://openai.com/gpt-4,

    OpenAI, “GPT-4 technical report. https://openai.com/gpt-4,” 2023

    work page 2023

  4. [4]

    Describe, ex- plain, plan and select: interactive planning with llms enables open-world multi-task agents,

    Z. Wang, S. Cai, G. Chen, A. Liu, X. Ma, and Y . Liang, “Describe, ex- plain, plan and select: interactive planning with llms enables open-world multi-task agents,” in Thirty-seventh Conference on Neural Information Processing Systems, 2023. JOURNAL OF LATEX CLASS FILES, VOL. 14, NO. 8, AUGUST 2021 10

    work page 2023

  5. [5]

    Experiential co-learning of software-developing agents,

    C. Qian, Y . Dang, J. Li, W. Liu, Z. Xie, Y . Wang, W. Chen, C. Yang, X. Cong, X. Che, Z. Liu, and M. Sun, “Experiential co-learning of software-developing agents,” inProceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), ACL 2024, Bangkok, Thailand, August 11-16, 2024 , L. Ku, A. Martins, and V . ...

    work page 2024

  6. [6]

    User behavior simulation with large language model-based agents,

    L. Wang, J. Zhang, H. Yang, Z. Chen, J. Tang, Z. Zhang, X. Chen, Y . Lin, R. Song, W. X. Zhao, J. Xu, Z. Dou, J. Wang, and J.-R. Wen, “User behavior simulation with large language model-based agents,” 2024

    work page 2024

  7. [7]

    SOTOPIA: interactive evaluation for social intelligence in language agents,

    X. Zhou, H. Zhu, L. Mathur, R. Zhang, H. Yu, Z. Qi, L. Morency, Y . Bisk, D. Fried, G. Neubig, and M. Sap, “SOTOPIA: interactive evaluation for social intelligence in language agents,” in The Twelfth International Conference on Learning Representations, ICLR 2024, Vienna, Austria, May 7-11, 2024 , 2024

    work page 2024

  8. [8]

    Large language models as simulated economic agents: What can we learn from homo silicus?

    J. J. Horton, “Large language models as simulated economic agents: What can we learn from homo silicus?” National Bureau of Economic Research, Tech. Rep., 2023

    work page 2023

Show all 49 references
  1. [9]

    Chatdev: Commu- nicative agents for software development,

    C. Qian, W. Liu, H. Liu, N. Chen, Y . Dang, J. Li, C. Yang, W. Chen, Y . Su, X. Cong, J. Xu, D. Li, Z. Liu, and M. Sun, “Chatdev: Commu- nicative agents for software development,” pp. 15 174–15 186, 2024

    2024

  2. [10]

    Swiftsage: A generative agent with fast and slow thinking for complex interactive tasks,

    B. Y . Lin, Y . Fu, K. Yang, F. Brahman, S. Huang, C. Bhagavatula, P. Ammanabrolu, Y . Choi, and X. Ren, “Swiftsage: A generative agent with fast and slow thinking for complex interactive tasks,” Advances in Neural Information Processing Systems , vol. 36, 2024

    2024

  3. [11]

    Toolformer: Language models can teach themselves to use tools,

    T. Schick, J. Dwivedi-Yu, R. Dess `ı, R. Raileanu, M. Lomeli, E. Hambro, L. Zettlemoyer, N. Cancedda, and T. Scialom, “Toolformer: Language models can teach themselves to use tools,” Advances in Neural Infor- mation Processing Systems , vol. 36, 2024

    2024

  4. [12]

    The small world problem,

    S. Milgram, “The small world problem,” Psychology today, vol. 2, no. 1, pp. 60–67, 1967

    1967

  5. [13]

    Simulating opinion dynamics with networks of llm-based agents,

    Y .-S. Chuang, A. Goyal, N. Harlalka, S. Suresh, R. Hawkins, S. Yang, D. Shah, J. Hu, and T. Rogers, “Simulating opinion dynamics with networks of llm-based agents,” in Findings of the Association for Computational Linguistics: NAACL 2024 , 2024, pp. 3326–3346

    2024

  6. [14]

    Playing atari with deep reinforcement learning,

    V . Mnih, K. Kavukcuoglu, D. Silver, A. Graves, I. Antonoglou, D. Wier- stra, and M. A. Riedmiller, “Playing atari with deep reinforcement learning,” CoRR, vol. abs/1312.5602, 2013

    2013 arXiv

  7. [15]

    Local patch autoaugment with multi-agent collaboration,

    S. Lin, T. Yu, R. Feng, X. Li, X. Yu, L. Xiao, and Z. Chen, “Local patch autoaugment with multi-agent collaboration,” IEEE Trans. Multim. , vol. 26, pp. 724–736, 2024

    2024

  8. [16]

    Multilevel anomaly detection through variational autoencoders and bayesian models for self-aware embodied agents,

    G. Slavic, M. Baydoun, D. Campo, L. Marcenaro, and C. S. Regazzoni, “Multilevel anomaly detection through variational autoencoders and bayesian models for self-aware embodied agents,” IEEE Trans. Multim., vol. 24, pp. 1399–1414, 2022

    2022

  9. [17]

    Discovering emergent agent behaviour with evolutionary finite state machines,

    M. Masek, C. P. Lam, L. Benke, L. Kelly, and M. Papasimeon, “Discovering emergent agent behaviour with evolutionary finite state machines,” in PRIMA 2018: Principles and Practice of Multi-Agent Systems - 21st International Conference, Tokyo, Japan, October 29 - November 2, 201...

    2018

  10. [18]

    Learning of behavior trees for autonomous agents,

    M. Colledanchise, R. Parasuraman, and P. ¨Ogren, “Learning of behavior trees for autonomous agents,” IEEE Trans. Games , vol. 11, no. 2, pp. 183–189, 2019

    2019

  11. [19]

    User-guided personalized image aesthetic assessment based on deep reinforcement learning,

    P. Lv, J. Fan, X. Nie, W. Dong, X. Jiang, B. Zhou, M. Xu, and C. Xu, “User-guided personalized image aesthetic assessment based on deep reinforcement learning,” IEEE Trans. Multim. , vol. 25, pp. 736–749, 2023

    2023

  12. [20]

    Knowledge-enhanced causal reinforcement learning model for interac- tive recommendation,

    W. Nie, X. Wen, J. Liu, J. Chen, J. Wu, G. Jin, J. Lu, and A. Liu, “Knowledge-enhanced causal reinforcement learning model for interac- tive recommendation,” IEEE Trans. Multim. , vol. 26, pp. 1129–1142, 2024

    2024

  13. [21]

    Multi-agent trajectory prediction with spatio- temporal sequence fusion,

    Y . Wang and S. Chen, “Multi-agent trajectory prediction with spatio- temporal sequence fusion,” IEEE Trans. Multim. , vol. 25, pp. 13–23, 2023

    2023

  14. [22]

    Recsim: A configurable simulation platform for recommender systems,

    E. Ie, C. Hsu, M. Mladenov, V . Jain, S. Narvekar, J. Wang, R. Wu, and C. Boutilier, “Recsim: A configurable simulation platform for recommender systems,” CoRR, vol. abs/1909.04847, 2019

    1909 arXiv

  15. [23]

    Alphastar: an evolutionary computation perspective,

    K. Arulkumaran, A. Cully, and J. Togelius, “Alphastar: an evolutionary computation perspective,” in Proceedings of the Genetic and Evolution- ary Computation Conference Companion, GECCO 2019, Prague, Czech Republic, July 13-17, 2019 . ACM, 2019, pp. 314–315

    2019

  16. [24]

    Agentverse: Facilitating multi-agent collaboration and exploring emergent behaviors in agents,

    W. Chen, Y . Su, J. Zuo, C. Yang, C. Yuan, C. Qian, C. Chan, Y . Qin, Y . Lu, R. Xie, Z. Liu, M. Sun, and J. Zhou, “Agentverse: Facilitating multi-agent collaboration and exploring emergent behaviors in agents,” International Conference on Learning Representations , 2024

    2024

  17. [25]

    Using large language models to simulate multiple humans and replicate human subject studies,

    G. V . Aher, R. I. Arriaga, and A. T. Kalai, “Using large language models to simulate multiple humans and replicate human subject studies,” in International Conference on Machine Learning . PMLR, 2023, pp. 337–371

    2023

  18. [26]

    Metagpt: Meta programming for a multi-agent collaborative framework,

    S. Hong, M. Zhuge, J. Chen, X. Zheng, Y . Cheng, J. Wang, C. Zhang, Z. Wang, S. K. S. Yau, Z. Lin et al. , “Metagpt: Meta programming for a multi-agent collaborative framework,” in The Twelfth International Conference on Learning Representations , 2024

    2024

  19. [27]

    Social simulacra: Creating populated prototypes for social computing systems,

    J. S. Park, L. Popowski, C. Cai, M. R. Morris, P. Liang, and M. S. Bernstein, “Social simulacra: Creating populated prototypes for social computing systems,” in Proceedings of the 35th Annual ACM Symposium on User Interface Software and Technology , 2022, pp. 1–18

    2022

  20. [28]

    Exploring large language models for communication games: An empirical study on werewolf,

    Y . Xu, S. Wang, P. Li, F. Luo, X. Wang, W. Liu, and Y . Liu, “Exploring large language models for communication games: An empirical study on werewolf,” arXiv preprint arXiv:2309.04658 , 2023

    2023 arXiv

  21. [29]

    War and peace (waragent): Large language model-based multi-agent simulation of world wars,

    W. Hua, L. Fan, L. Li, K. Mei, J. Ji, Y . Ge, L. Hemphill, and Y . Zhang, “War and peace (waragent): Large language model-based multi-agent simulation of world wars,” arXiv preprint arXiv:2311.17227 , 2023

    2023 arXiv

  22. [30]

    Mindagent: Emergent gaming interaction,

    R. Gong, Q. Huang, X. Ma, Y . Noda, Z. Durante, Z. Zheng, D. Ter- zopoulos, L. Fei-Fei, J. Gao, and H. V o, “Mindagent: Emergent gaming interaction,” in Findings of the Association for Computational Linguis- tics: NAACL 2024 , 2024, pp. 3154–3183

    2024

  23. [31]

    Human memory: A proposed system and its control processes,

    R. C. Atkinson and R. M. Shiffrin, “Human memory: A proposed system and its control processes,” in Psychology of Learning and Motivation . Elsevier, 1968, vol. 2, pp. 89–195

    1968

  24. [32]

    Adaptive mem- ory: survival processing enhances retention

    J. S. Nairne, S. R. Thompson, and J. N. Pandeirada, “Adaptive mem- ory: survival processing enhances retention.” Journal of Experimental Psychology: Learning, Memory, and Cognition , vol. 33, no. 2, p. 263, 2007

    2007

  25. [33]

    Chain-of-thought prompting elicits reasoning in large language models,

    J. Wei, X. Wang, D. Schuurmans, M. Bosma, F. Xia, E. Chi, Q. V . Le, D. Zhou et al. , “Chain-of-thought prompting elicits reasoning in large language models,” Advances in neural information processing systems , vol. 35, pp. 24 824–24 837, 2022

    2022

  26. [34]

    Collective dynamics of ‘small- world’networks,

    D. J. Watts and S. H. Strogatz, “Collective dynamics of ‘small- world’networks,” nature, vol. 393, no. 6684, pp. 440–442, 1998

    1998

  27. [35]

    LangChain,

    H. Chase, “LangChain,” Oct. 2022. [Online]. Available: https: //github.com/langchain-ai/langchain

    2022

  28. [36]

    Amazon review dataset (2018),

    J. Ni, “Amazon review dataset (2018),” 2018. [Online]. Available: https://cseweb.ucsd.edu/∼jmcauley/datasets/amazon v2/

    2018

  29. [37]

    Probabilistic matrix factorization,

    R. Salakhutdinov and A. Mnih, “Probabilistic matrix factorization,” pp. 1257–1264, 2007

    2007

  30. [38]

    Collaborative filtering recommender systems,

    M. D. Ekstrand, J. T. Riedl, J. A. Konstan et al., “Collaborative filtering recommender systems,” Foundations and Trends® in Human–Computer Interaction, vol. 4, no. 2, pp. 81–173, 2011

    2011

  31. [39]

    The power of personality: The comparative validity of personality traits, socioeconomic status, and cognitive ability for predicting important life outcomes,

    B. W. Roberts, N. R. Kuncel, R. Shiner, A. Caspi, and L. R. Goldberg, “The power of personality: The comparative validity of personality traits, socioeconomic status, and cognitive ability for predicting important life outcomes,” Perspectives on Psychological science , vol. 2,...

    2007

  32. [40]

    Using social norms to change behavior and increase sustainability in the real world: A systematic review of the literature,

    P. Yamin, M. Fei, S. Lahlou, and S. Levy, “Using social norms to change behavior and increase sustainability in the real world: A systematic review of the literature,” Sustainability, vol. 11, no. 20, p. 5847, 2019

    2019

  33. [41]

    Social theory, social research, and a theory of action,

    J. S. Coleman, “Social theory, social research, and a theory of action,” American journal of Sociology , vol. 91, no. 6, pp. 1309–1335, 1986

    1986

  34. [42]

    Naturalness in language,

    J. Sinclair, “Naturalness in language,” in Corpus linguistics. Brill, 1984, pp. 203–210

    1984

  35. [43]

    Expressiveness and closure properties for quantitative languages,

    K. Chatterjee, L. Doyen, and T. A. Henzinger, “Expressiveness and closure properties for quantitative languages,” Log. Methods Comput. Sci., vol. 6, no. 3, 2010

    2010

  36. [44]

    Free-marginal multirater kappa (multirater κfree): An alternative to fleiss’ fixed-marginal multirater kappa,

    J. J. Randolph, “Free-marginal multirater kappa (multirater κfree): An alternative to fleiss’ fixed-marginal multirater kappa,” vol. 2005, 2005

    2005

  37. [45]

    GPT detectors are biased against non-native english writers,

    W. Liang, M. Y ¨uksekg¨on¨ul, Y . Mao, E. Wu, and J. Zou, “GPT detectors are biased against non-native english writers,” Patterns, vol. 4, no. 7, p. 100779, 2023

    2023

  38. [46]

    Word association norms, mutual information, and lexicography,

    K. Church and P. Hanks, “Word association norms, mutual information, and lexicography,” Computational linguistics, vol. 16, no. 1, pp. 22–29, 1990

    1990

  39. [47]

    Herd behavior in a complex adaptive system,

    L. Zhao, G. Yang, W. Wang, Y . Chen, J. Huang, H. Ohashi, and H. E. Stanley, “Herd behavior in a complex adaptive system,” Proceedings of the National Academy of Sciences , vol. 108, no. 37, pp. 15 058–15 063, 2011

    2011

  40. [48]

    The structure and function of complex networks,

    M. E. Newman, “The structure and function of complex networks,” SIAM review, vol. 45, no. 2, pp. 167–256, 2003

    2003

  41. [49]

    Information diffusion in online social networks: a survey,

    A. Guille, H. Hacid, C. Favre, and D. A. Zighed, “Information diffusion in online social networks: a survey,” SIGMOD Rec., vol. 42, no. 2, pp. 17–28, 2013

    2013

Pith tools

Reviewed August 11, 2026 · model on record in the stance chip above.