Pith. sign in

REVIEW 3 major objections 4 minor 1 cited by

Who Speaks Next? Multi-party AI Discussion Leveraging the Systematics of Turn-taking in Murder Mystery Games

T0 review · 3 major / 4 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 Applying human turn-taking rules to multi-agent LLM dialogue sharply reduces breakdowns and improves cooperation, information sharing, and reasoning in murder-mystery games.

desk verdict A clean empirical study of adjacency-pair turn-taking for multi-agent dialogue; the mechanism claim is confounded by a prompt-injection change, but the system works and deserves a proper peer review. read the letter →

arxiv 2412.04937 v2 pith:4S26YS6K submitted 2024-12-06 cs.CL cs.AI

Ryota Nonomura , Hiroki Mori This is my paper · ORCID
classification cs.CLcs.AI
keywords turn-takingconversationanalysismulti-agentsystemsLLMagentsadjacencypairscurrentspeakerselectsnextdialoguebreakdownmurdermystery
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 argues that the turn-taking rules conversation analysts have documented in human talk—especially the 'current speaker selects next' technique, where a question or address creates an obligation for a specific addressee to respond—can be transplanted into multi-agent LLM dialogue to make it more orderly and effective. The authors build 'Murder Mystery Agents,' a framework in which an LLM detects when an utterance is the first half of an adjacency pair (such as a question) and selects the named addressee as the next speaker, adding an explicit '(response)' constraint to that agent's prompt. In a four-player murder mystery discussion, this condition produced significantly fewer dialogue breakdowns, higher judged cooperation and diversity, and better information sharing and reasoning scores than both random turn assignment and pure self-selection. The work matters because it turns a descriptive science of human conversation into a design rule for AI systems, and it gives a concrete, cheap mechanism—detect the adjacency pair, assign the obligation—that other multi-party agent architectures could adopt.

What carries the argument

The central mechanism is the detectDesignation() module coupled with the CSSN-or-SS turn-taking rule. detectDesignation() uses an LLM to classify whether the latest utterance is the first pair part of an adjacency pair (a yes/no question, a wh-question, or a direct address) and to predict the addressee; that addressee becomes the next speaker, and the prompt for the next utterance carries an explicit '(response)' constraint. This enforces the 'current speaker selects next' rule from conversation analysis, so that a question is immediately followed by the answer from the person asked, rather than by a self-selected speaker or a random turn.

What would settle it

Run the same experiment with a fourth condition: select the predicted addressee as next speaker but omit the '(response)' token from their prompt. If breakdown counts and human scores match CSSN-or-SS, the selection mechanism alone is responsible; if they regress toward SS, the prompt-level obligation is the active ingredient.

Watch

Extended reading notes

Core claim

On the authors' own terms, the discovery is that imposing response obligations derived from adjacency pairs makes LLM-based multi-party conversation work better. When an utterance contains a first pair part (a question or a direct address), the detectDesignation() module identifies the type of the pair and predicts the intended next speaker; that speaker's next-generation prompt is then marked with '(response)'. Across 50 ten-turn conversations per condition, this CSSN-or-SS condition reduced dialogue-breakdown utterances by a wide margin compared with EQUAL random turns and SS self-selection, with the Kruskal–Wallis test showing significant differences for breakdowns, cooperation, diversity, and the human murder-mystery criteria. The authors read this as support for the theoretical claim that sequence organization is the source of conversational coherence, now evidenced inside machine-machine dialogue.

Load-bearing premise

The paper never tests next-speaker selection without the explicit '(response)' note in the prompt, so the improvement could come from that nudge rather than from choosing the addressee as the next speaker.

Editorial extensions

If this is right

  • Multi-agent LLM systems can reduce dialogue breakdowns by routing turns through the addressee detected in the previous utterance.
  • In collaborative reasoning games, structuring turns by adjacency pairs leads agents to share more mission-relevant facts and to reason more logically, per the paper's human evaluation.
  • The mechanism is cheap: one extra LLM call per turn to detect the first pair part, with no retraining or fine-tuning.
  • The design guidance generalizes beyond games: any chat platform with multiple AI participants could adopt the same obligation-based turn-taking.
  • The result supports the theoretical claim that sequence organization (adjacency pairs) is a source of conversational coherence, now demonstrated empirically inside machine-machine talk.

Reading between the lines

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

  • A natural extension the paper leaves implicit is to use the same adjacency-pair detection to manage topic length: when a sequence is closed with a sequence-closing third such as 'I see, thank you', the system could score the topic as finished and nudge agents toward new information.
  • Because the paper reports long-term memory degradation around 30 turns, the observed turn-taking benefit may shrink in longer dialogues; a direct test is to plot breakdown frequency against turn number and see whether CSSN-or-SS loses its edge.
  • The mechanism's strategic-evasion behavior (partial answers, topic shifts) suggests that response obligations are weighed against character goals; this could be repurposed to create agents with calibrated secrecy—obliged to respond, but free to withhold the core secret.
  • One could also swap the LLM-based detector for a simpler rule-based one in constrained domains; if the gains are mostly from the response constraint, rule-based detection may be sufficient, making the approach even cheaper.
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 / 4 minor

Summary. The paper presents MMAgents, a multi-agent framework for LLM-based Murder Mystery game play, which implements two conversation-analysis-inspired turn-taking mechanisms: a self-selection mechanism based on agent-generated importance scores (think()/selectMostImportant()) and a current-speaker-selects-next mechanism (detectDesignation()) that detects adjacency-pair first parts and designates the addressee as next speaker. The authors compare three conditions (EQUAL, SS, CSSN-or-SS) over 50 ten-turn conversations per condition, using dialogue-breakdown analysis, LLM-as-a-judge scoring, and human evaluation. They report that CSSN-or-SS significantly reduces dialogue breakdowns and improves cooperativeness, diversity, information sharing, and reasoning relative to both baselines, and they attribute these gains to the adjacency-pair-based next-speaker selection mechanism.

Significance. If the reported effects are robust and correctly attributed, this is a useful contribution to multi-agent LLM dialogue control: it operationalizes classic conversation-analysis concepts in an implemented system and provides a concrete, statistically tested comparison. The paper is generally clear, the three-condition design is sensible, and the authors are candid about several limitations (e.g., self-conducted human evaluation, topic-control issues). The main weakness is that the CSSN-or-SS condition differs from SS in two coupled ways, so the central mechanism-level claim is not yet isolated.

major comments (3)
  1. [§3.2.4, §4.1, Figure 4] The comparison between CSSN-or-SS and SS is confounded. In CSSN-or-SS, detectDesignation() both (a) selects the addressee as the next speaker and (b) rewrites the speak() prompt to include an explicit '(response)' marker and the second-pair-part type; SS removes both changes together. The abstract and conclusion attribute the improvements specifically to the next-speaker selection mechanism, but the observed gains could come from the prompt-level response obligation alone, from the selection rule, or from their interaction. An ablation separating these two factors (e.g., SS with the '(response)' prompt injected but without selection, or CSSN-or-SS without the prompt injection) is needed to support the stated mechanism.
  2. [§4.2.3, §6, Figure 7] The human evaluation was performed by the authors themselves using a bespoke rubric, with no second annotator, no inter-annotator reliability statistics, and no blinding to condition. Section 6 acknowledges the subjectivity concern, but the conclusion's claims about 'information sharing capabilities and reasoning abilities' depend substantially on this evaluation. Adding at least one independent annotator and reporting agreement (e.g., Cohen's kappa) is necessary to make this evidence load-bearing.
  3. [§4.2.1–4.2.2, §5] The breakdown analysis and LLM-as-a-judge both rely on GPT-4, the same model family used to generate the conversations (GPT-4o for speak() and detectDesignation()). While this does not invalidate the relative comparisons since all conditions are judged by the same model, it introduces a possible same-family bias that is not discussed as a limitation. A brief note in Section 6 about this circularity risk, and ideally a spot-check with a different judge model, would strengthen the evaluation.
minor comments (4)
  1. [§5] The Kruskal-Wallis and Dunn tests are reported with p-values, but no effect sizes or medians/interquartile ranges are given for the breakdown counts or judge scores; reporting these would make the magnitude of the improvements more interpretable.
  2. [§3.2.4, Figure 5] No accuracy analysis is provided for detectDesignation() (the LLM-based detection of first pair parts and addressee prediction). Since the mechanism's behavior depends on this detector, a small evaluation of its error rate would help readers gauge how noise in detection propagates to the results.
  3. [§4.1] The example conversations are shown in English translation from Japanese; the authors note this in a figure caption, but it would be helpful to state explicitly in the text that all prompts and generated utterances were originally in Japanese.
  4. [§3.1.2, Eq. (2)] The notation for shortTermHistory uses the same index window length k as History, and the later statement in §4.1 confirms k=5; consider defining k once and noting that the same k is used for both memory types.

Circularity Check

0 steps flagged · score 0.0 of 10

No definitional circularity: the evaluation metrics are external and nothing is fitted to the outcomes; the main caveat is an unablated confound between next-speaker selection and the '(response)' prompt insertion, which affects mechanism attribution but is not a circular derivation.

full rationale

The paper's central derivation is an empirical comparison of three turn-taking conditions, not a formal derivation. The outcome metrics are defined independently of the mechanism's internal parameters: dialogue-breakdown counts use Higashinaka et al.'s external taxonomy (Section 4.2.1), LLM-as-a-judge uses 1-5 rubrics for coherence, cooperativeness, and diversity (Section 4.2.2), and human evaluation uses a game-specific scoring rubric keyed to case facts (Section 4.2.3). Importance values produced by think() are generated by the LLM, not fitted to these scores, and the theoretical grounding in Sacks/Schegloff turn-taking systematics is externally cited rather than a self-citation chain. No parameter is fitted to a subset and then reported as a prediction. The strongest concern is a confound, not circularity: Section 4.1 defines SS as the same as CSSN-or-SS except for the detectDesignation() mechanism, while Section 3.2.4 shows that detectDesignation() also injects a '(response)' constraint and second-pair-part type into the next speaker's speak() prompt. Thus the CSSN-or-SS improvement is attributable to the combination of next-speaker selection and prompt-level response obligation, so the conclusion crediting 'the next-speaker selection mechanism' is not fully isolated. However, this does not make the outcome equivalent to the input by definition: the model can still fail to produce an appropriate response, and the diversity, cooperation, and information-sharing metrics are not directly forced by the prompt. Section 6's acknowledgement of possible author subjectivity in human evaluation is a validity limitation, not a circular step. Therefore the paper is self-contained against its benchmarks, with no significant circularity.

Assumptions & free parameters 2 free parameters · 3 assumptions · 0 invented entities

The central claim rests on two imported assumptions: the conversation-analysis premise that adjacency pairs generate coherence, and the reliability of the LLM-based detector for first pair parts and addressees. Two small hyperparameters (memory window sizes) are set by hand. No new entities are introduced.

free parameters (2)
  • history_length_k = 5
    history and shortTermHistory retain 5 turns to fit LLM input limits (Section 4.1).
  • long_term_memory_top_l = 5
    Top 5 knowledge entries from longTermHistory are retrieved by cosine similarity (Section 4.1).
assumptions (3)
  • domain assumption Adjacency pairs are the source of coherence in conversation (Schegloff, 1990)
    The CSSN mechanism is justified by this theory, cited as [21] in Sections 1 and 7.
  • domain assumption LLM-based detectDesignation() correctly identifies first pair parts and their addressees
    The benefit of the mechanism depends on classification reliability (Section 3.2.4); no accuracy metric is reported.
  • domain assumption The importance value from think() is a valid proxy for self-selection priority
    Self-selection is operationalized as the maximum importance (Section 3.2.2); the mapping is assumed, not validated.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Who Speaks Next? Multi-party AI Discussion Leveraging the Systematics of Turn-taking in Murder Mystery Games." pith.science (2026). https://pith.science/paper/4S26YS6K

@misc{pith2026241204937,
  author       = {Pith},
  title        = {Pith review of: Who Speaks Next? Multi-party AI Discussion Leveraging the Systematics of Turn-taking in Murder Mystery Games},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/4S26YS6K}},
  note         = {Machine review of arXiv:2412.04937}
}
read the original abstract

Multi-agent systems utilizing large language models (LLMs) have shown great promise in achieving natural dialogue. However, smooth dialogue control and autonomous decision making among agents still remain challenges. In this study, we focus on conversational norms such as adjacency pairs and turn-taking found in conversation analysis and propose a new framework called "Murder Mystery Agents" that applies these norms to AI agents' dialogue control. As an evaluation target, we employed the "Murder Mystery" game, a reasoning-type table-top role-playing game that requires complex social reasoning and information manipulation. In this game, players need to unravel the truth of the case based on fragmentary information through cooperation and bargaining. The proposed framework integrates next speaker selection based on adjacency pairs and a self-selection mechanism that takes agents' internal states into account to achieve more natural and strategic dialogue. To verify the effectiveness of this new approach, we analyzed utterances that led to dialogue breakdowns and conducted automatic evaluation using LLMs, as well as human evaluation using evaluation criteria developed for the Murder Mystery game. Experimental results showed that the implementation of the next speaker selection mechanism significantly reduced dialogue breakdowns and improved the ability of agents to share information and perform logical reasoning. The results of this study demonstrate that the systematics of turn-taking in human conversation are also effective in controlling dialogue among AI agents, and provide design guidelines for more advanced multi-agent dialogue systems.

Figures

Figures reproduced from arXiv: 2412.04937 by the authors.

Figure 1
Figure 1. Example of character information. The original text is in Japanese. The same [PITH_FULL_IMAGE:figures/full_fig_p008_1.png] view at source ↗
Figure 2
Figure 2. Example of normalizing utterances into knowledge or information and storing [PITH_FULL_IMAGE:figures/full_fig_p010_2.png] view at source ↗
Figure 3
Figure 3. Example output of think(). that Kozue Taniguchi judged her utterance to be significant for the conver￾sation’s development. 3.2.2. selectMostImportant() The selectMostImportant(agents) is a speaker selection algorithm that implements the Self-Selection mechanism. This algorithm processes differ￾ently based on the number of agents who have selected “speak.” When only one agent selects “speak”, that agent naturally be… view at source ↗
Figures from the paper (13 more)
Figure 4
Figure 4. Figure 4: Example of prompt for speak(). 15 [PITH_FULL_IMAGE:figures/full_fig_p015_4.png]
Figure 5
Figure 5. Figure 5: Example of detectDesignation(). Murder Case” begins with a story of former college tennis team members reuniting on an isolated island after three years. The scenario features the following four characters: • Kozue Taniguchi (female): A boyish character with a straight…
Figure 6
Figure 6. Figure 6: Prompt for the analysis of dialogue breakdown. [PITH_FULL_IMAGE:figures/full_fig_p019_6.png]
Figure 7
Figure 7. Figure 7: Guidelines for human evaluation. 22 [PITH_FULL_IMAGE:figures/full_fig_p022_7.png]
Figure 8
Figure 8. Figure 8: Example conversation in EQUAL condition. [PITH_FULL_IMAGE:figures/full_fig_p023_8.png]
Figure 9
Figure 9. Figure 9: Example conversation in SS condition. 23 [PITH_FULL_IMAGE:figures/full_fig_p023_9.png]
Figure 10
Figure 10. Figure 10: Example conversation in CSSN-or-SS condition. [PITH_FULL_IMAGE:figures/full_fig_p024_10.png]
Figure 11
Figure 11. Figure 11: The number of utterances that lead to dialogue breakdowns within 10 turns. [PITH_FULL_IMAGE:figures/full_fig_p025_11.png]
Figure 12
Figure 12. Figure 12: Result of LLM judge. 26 [PITH_FULL_IMAGE:figures/full_fig_p026_12.png]
Figure 13
Figure 13. Figure 13: Result of Human evaluation. mation sharing, showing a wide distribution centered around score 8. In contrast, the EQUAL and SS conditions exhibited dialogue breakdowns such as monopolization of speech by specific agents and consecutive questioning, resulting in narrow…
Figure 14
Figure 14. Figure 14: The classified types of utterances that lead to dialogue breakdowns [PITH_FULL_IMAGE:figures/full_fig_p028_14.png]
Figure 15
Figure 15. Figure 15: Example of Ignore expectation. 29 [PITH_FULL_IMAGE:figures/full_fig_p029_15.png]
Figure 16
Figure 16. Figure 16: Example of low score conversation. points even in the CSSN-or-SS condition indicates that there is still room for improvement in the proposed method. Analysis of low-scoring conversation examples, as shown in [PITH_FULL_IMAGE:figures/full_fig_p031_16.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. Multi-Party Conversational Agents: A Survey

    cs.CL 2025-05 conditional novelty 4.0 of 10

    A survey of multi-party conversational AI that organizes tasks into state-of-mind modeling, semantic understanding, and action modeling, and argues that theory of mind is the key missing ingredient.

Reference graph

Works this paper leans on

43 extracted references · 19 canonical work pages · cited by 1 Pith paper

  1. [1]

    T. B. Brown, B. Mann, N. Ryder, M. Subbiah, J. Kaplan, P. Dhariwal, A. Neelakantan, P. Shyam, G. Sastry, A. Askell, S. Agarwal, A. Herbert- Voss, G. Krueger, T. Henighan, R. Child, A. Ramesh, D. M. Ziegler, J. Wu, C. Winter, C. Hesse, M. Chen, E. Sigler, M. Litwin, S. Gray, B. Chess, J. Clark, C. Berner, S. McCandlish, A. Radford, I. Sutskever, D. Amodei,...

    arXiv 2020

  2. [2]

    [Preprint]

    OpenAI, GPT-4 Technical ReportArXiv. [Preprint]. Available at: https://arxiv.org/abs/2303.08774 (2023). arXiv:2303.08774

    arXiv 2023

  3. [3]

    Touvron, T

    H. Touvron, T. Lavril, G. Izacard, X. Martinet, M.-A. Lachaux, T. Lacroix, B. Rozière, N. Goyal, E. Hambro, F. Azhar, A. Rodriguez, A. Joulin, E. Grave, G. Lample, LLaMA: Open and efficient foundation language modelsArXiv. [Preprint]. Available at: https://arxiv.org/ abs/2302.13971 (2023). arXiv:2302.13971

    arXiv 2023

  4. [4]

    Hugo, et al., Llama 2: Open foundation and fine-tuned chat model- 33 sArXiv

    T. Hugo, et al., Llama 2: Open foundation and fine-tuned chat model- 33 sArXiv. [Preprint]. Available at:https://arxiv.org/abs/2307.09288 (2023). arXiv:2307.09288

    arXiv 2023

  5. [5]

    ask me anything

    S. Rome, T. Chen, R. Tang, L. Zhou, F. Ture, "ask me anything": How comcast uses LLMs to assist agents in real time, in: Proceed- ings of the 47th International ACM SIGIR Conference on Research and Development in Information Retrieval, ACM, 2024, pp. 2827–2831. doi:10.1145/3626772.3661345

    arXiv 2024

  6. [6]

    J. Jeon, S. Lee, Large language models in education: A focus on the complementary relationship between human teachers and Chat- GPT, Education and Information Technologies 28 (2023) 15873–15892. doi:10.1007/s10639-023-11834-1

    doi:10.1007/s10639-023-11834-1 2023

  7. [7]

    B. Hu, L. Zheng, J. Zhu, L. Ding, Y. Wang, X. Gu, Teaching plan generationandevaluationwithGPT-4: UnleashingthepotentialofLLM in instructional design, IEEE Transactions on Learning Technologies 17 (2024) 1471–1485. doi:10.1109/TLT.2024.3384765

    arXiv 2024

  8. [8]

    Zhang, D

    Z. Zhang, D. Zhang-Li, J. Yu, L. Gong, J. Zhou, Z. Liu, L. Hou, J. Li, Simulating classroom education with LLM-empowered agentsArXiv. [Preprint]. Available at: https://arxiv.org/abs/2406.19226 (2024). arXiv:2406.19226

    arXiv 2024

Show all 43 references
  1. [9]

    OpenAI, Introducing ChatGPT[Accessed Feb 17, 2025] (2022)

    2022

  2. [10]

    Anthropic, Claude 3.5 Sonnet[Accessed Feb 17, 2025] (2024)

    2024

  3. [11]

    J. S. Park, J. O’Brien, C. J. Cai, M. R. Morris, P. Liang, M. S. Bernstein, Generative agents: Interactive simulacra of human behavior, in: UIST 2023: Proceedings of the 36th Annual ACM Symposium on User Inter- face Software and Technology, Association for Computing Machinery,...

    2023

  4. [12]

    Y. Lan, Z. Hu, L. Wang, Y. Wang, D. Ye, P. Zhao, E.-P. Lim, H. Xiong, H. Wang, LLM-based agent society investigation: Collabora- tion and confrontation in Avalon gameplayArXiv. [Preprint]. Available at: https://arxiv.org/abs/2310.14985 (2023). arXiv:2310.14985

    2023 arXiv

  5. [13]

    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, M. Sun, ChatDev: Communicative 34 agents for software development, in: L.-W. Ku, A. Martins, V. Srikumar (Eds.), Proceedings of the 62nd Annual Meeting of the Association ...

    2024 doi

  6. [14]

    Z. Gu, X. Zhu, H. Guo, L. Zhang, Y. Cai, H. Shen, J. Chen, Z. Ye, Y. Dai, Y. Gao, Y. Hu, H. Feng, Y. Xiao, AgentGroupChat: An inter- active group chat simulacra for better eliciting emergent behaviorArXiv. [Preprint]. Available at: https://arxiv.org/abs/2403.13433 (2024). arXi...

    2024 arXiv

  7. [15]

    Q. Wu, G. Bansal, J. Zhang, Y. Wu, B. Li, E. E. Zhu, L. Jiang, X. Zhang, S. Zhang, A. Awadallah, R. W. White, D. Burger, C. Wang, Autogen: Enabling next-gen llm applications via multi-agent conversa- tion, in: COLM 2024, 2024. URL https://www.microsoft.com/en-us/research/publi...

    2024

  8. [16]

    Herring, Interactional coherence in CMC, in: HICSS-32: Proceed- ings of the 32nd Annual Hawaii International Conference on Systems Sciences, 1999, pp

    S. Herring, Interactional coherence in CMC, in: HICSS-32: Proceed- ings of the 32nd Annual Hawaii International Conference on Systems Sciences, 1999, pp. 1–13.doi:10.1109/HICSS.1999.772674

    1999

  9. [17]

    Bailis, J

    S. Bailis, J. Friedhoff, F. Chen, Werewolf Arena: A case study in LLM evaluation via social deductionArXiv. [Preprint]. Available at:https: //arxiv.org/abs/2407.13943 (2024). arXiv:2407.13943

    2024 arXiv

  10. [18]

    E. A. Schegloff, H. Sacks, Opening up closings, Semiotica 8 (1973) 289– 327

    1973

  11. [19]

    E. A. Schegloff, Sequencing in conversational openings, American An- thropologist 70 (6) (1968) 1075–1095. doi:10.1525/aa.1968.70.6. 02a00030

    1968 doi

  12. [20]

    Sacks, E

    H. Sacks, E. A. Schegloff, G. Jefferson, A simplest systematics for the organization of turn-taking for conversation, Language 50 (1974) 696– 735. 35

    1974

  13. [21]

    E. A. Schegloff, On the organization of sequences as a source of ‘coher- ence’ in talk-in-interaction, in: B. Dorval (Ed.), Conversational Organi- zation and its Development, Ablex, Norwood, NJ, 1990, pp. 51–77

    1990

  14. [22]

    L. Wang, C. Ma, X. Feng, Z. Zhang, H. Yang, J. Zhang, Z. Chen, J. Tang, X. Chen, Y. Lin, W. X. Zhao, Z. Wei, J.-R. Wen, A survey on large language model based autonomous agents, Frontiers on Com- puter Science 18 (2024) 1–26, article number 186345. doi:10.1007/ s11704-024-40231-1

    2024

  15. [23]

    Significant Gravitas, AutoGPT [software] (2023)

    2023

  16. [24]

    Yohei Nakajima, BabyAGI [software] (2023)

    2023

  17. [25]

    S. Hu, T. Huang, F. Ilhan, S. Tekin, G. Liu, R. Kompella, L. Liu, A survey on large language model-based game agentsArXiv. [Preprint]. Available at: https://arxiv.org/abs/2404.02039 (2024). arXiv: 2404.02039

    2024 arXiv

  18. [26]

    Filippas, J

    A. Filippas, J. J. Horton, B. S. Manning, Large language models as simulated economic agents: What can we learn from homo silicus?, in: Proceedings of the 25th ACM Conference on Economics and Compu- tation, EC ’24, Association for Computing Machinery, New York, NY, USA, 2024, ...

    2024

  19. [27]

    G. Li, H. A. Al Kader Hammoud, H. Itani, D. Khizbullin, B. Ghanem, Camel: communicative agents for "mind" exploration of large language model society, in: Proceedings of the 37th International Conference on Neural Information Processing Systems, NIPS ’23, Curran Associates Inc...

    2023

  20. [28]

    Junprung, Exploring the intersection of large language models and agent-based modeling via prompt engineeringArXiv

    E. Junprung, Exploring the intersection of large language models and agent-based modeling via prompt engineeringArXiv. [Preprint]. Avail- able at: https://arxiv.org/abs/2308.07411 (2023). arXiv:2308. 07411

    2023 arXiv

  21. [29]

    R. Yang, J. Chen, Y. Zhang, S. Yuan, A. Chen, K. Richardson, Y. Xiao, D. Yang, SelfGoal: Your language agents already know how to achieve high-level goalsArXiv. [Preprint]. Available at: https://arxiv.org/ abs/2406.04784 (2024). arXiv:2406.04784. 36

    2024 arXiv

  22. [30]

    J.Wei, X.Wang, D.Schuurmans, M.Bosma, B.Ichter, F.Xia, E.H.Chi, Q. V. Le, D. Zhou, Chain-of-thought prompting elicits reasoning in large language models, in: Proceedings of the 36th International Conference on Neural Information Processing Systems, NIPS ’22, Curran Associates ...

    2022

  23. [31]

    S. Yao, J. Zhao, D. Yu, N. Du, I. Shafran, K. Narasimhan, Y. Cao, ReAct: Synergizing reasoning and acting in language modelsArXiv. [Preprint]. Available at: https://arxiv.org/abs/2210.03629 (2022). arXiv:2210.03629

    2022 arXiv

  24. [32]

    Shanahan, K

    M. Shanahan, K. McDonell, L. Reynolds, Role play with large language models, Nature 623 (7987) (2023) 493–498

    2023

  25. [33]

    Y. Shao, L. Li, J. Dai, e. . B. H. Qiu, Xipeng", J. Pino, K. Bali, Character-LLM: A trainable agent for role-playing, in: Proceedings of the 2023 Conference on Empirical Methods in Natural Language Pro- cessing, Association for Computational Linguistics, Singapore, 2023, pp. 1...

    2023

  26. [35]

    K. Lu, B. Yu, C. Zhou, J. Zhou, Large language models are superposi- tions of all characters: Attaining arbitrary role-play via self-alignment, in: L.-W. Ku, A. Martins, V. Srikumar (Eds.), Proceedings of the 62nd Annual Meeting of the Association for Computational Linguis- ti...

    2024 doi

  27. [36]

    Zhang, X

    Z. Zhang, X. Bo, C. Ma, R. Li, X. Chen, Q. Dai, J. Zhu, Z. Dong, J.- R. Wen, A survey on the memory mechanism of large language model based agentsArXiv. [Preprint]. Available at:https://arxiv.org/abs/ 2404.13501 (2024). arXiv:2404.13501

    2024 arXiv

  28. [37]

    Zhong, L

    W. Zhong, L. Guo, Q. Gao, H. Ye, Y. Wang, Memorybank: Enhancing large language models with long-term memory, in: Proceedings of the AAAI Conference on Artificial Intelligence, Vol. 38, 2024, pp. 19724– 19731

    2024

  29. [38]

    Modarressi, A

    A. Modarressi, A. Imani, M. Fayyaz, H. Schütze, RET-LLM: To- wards a general read-write memory for large language modelsArXiv. [Preprint]. Available at: https://arxiv.org/abs/2305.14322 (2023). arXiv:2305.14322

    2023 arXiv

  30. [39]

    Higashinaka, M

    R. Higashinaka, M. Araki, H. Tsukahara, M. Mizukami, Classification of utterances that lead to dialogue breakdowns in chat-oriented dialogue systems, Journal of Natural Language Processing 29 (2022) 443–466. doi:10.5715/jnlp.29.443

    2022 doi

  31. [40]

    Al-Onaizan, M

    Z.Li, X.Xu, T.Shen, C.Xu, J.-C.Gu, Y.Lai, C.Tao, S.Ma, Leveraging large language models for NLG evaluation: Advances and challenges, in: Y. Al-Onaizan, M. Bansal, Y.-N. Chen (Eds.), Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing, Associa...

    2024 doi

  32. [41]

    Kocmi, C

    T. Kocmi, C. Federmann, Large language models are state-of-the-art evaluators of translation quality, in: M. Nurminen, J. Brenner, M. Ko- ponen, S. Latomaa, M. Mikhailov, F. Schierl, T. Ranasinghe, E. Van- massenhove, S. A. Vidal, N. Aranberri, M. Nunziatini, C. P. Escartín, M...

    2023

  33. [42]

    M. Gao, J. Ruan, R. Sun, X. Yin, S. Yang, X. Wan, Human-like sum- 38 marization evaluation with ChatGPTArXiv. [Preprint]. Available at: https://arxiv.org/abs/2304.02554 (2023). arXiv:2304.02554

    2023 arXiv

  34. [43]

    Chiang, H.-y

    C.-H. Chiang, H.-y. Lee, Can large language models be an alternative to human evaluations?, in: A. Rogers, J. Boyd-Graber, N. Okazaki (Eds.), Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), Association for Comput...

    2023 doi

  35. [44]

    E. A. Schegloff, Sequence Organization in Interaction: A Primer in Conversation Analysis, Vol. 1, Cambridge University Press, Cambridge, 2007. 39

    2007

Pith tools

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