Pith. sign in

REVIEW 3 major objections 7 minor 38 references

First Steps Towards Overhearing LLM Agents: A Case Study With Dungeons & Dragons Gameplay

T0 review · 3 major / 7 minor · reviewed 2026-08-07 · deepseek-v4-flash

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

Pith's one-line read Large audio-language models can act as 'overhearing agents' that listen to human-to-human conversation and make background suggestions, with the biggest model using tone, pauses, and speaker identity that transcripts cannot capture.

desk verdict A genuinely new paradigm and a serious empirical study, but the headline audio-vs-text claim is confounded by checkpoint differences in the only models that show the effect. read the letter →

arxiv 2505.22809 v2 pith:GMF3HJWS submitted 2025-05-28 cs.CL cs.AIcs.HC

Andrew Zhu , Evan Osgood , Chris Callison-Burch This is my paper
classification cs.CLcs.AIcs.HC
keywords overhearingagentsaudio-languagemodelsmultimodalLLMtoolcallingDungeons&Dragonsimplicitaudiocueshumanevaluationreal-timeAIassistant
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 introduces "overhearing agents": AI assistants that never speak to their user but listen in on human-to-human conversation and hand helpful suggestions to the background, through tool calls rather than dialogue. To test whether such agents work, the authors built one that assists a Dungeon Master during live D&D games, with three background tasks (retrieving game rules, managing NPC portraits on a virtual stage, and improvising new NPCs), and ran it on 41.5 hours of recorded play. The central claim is that some large audio-language models have an emergent ability for this task: the strongest model, GPT-4o, reached an F1 of 0.369 with audio input versus 0.236 when given only a transcript, and the gap is credited to implicit audio cues such as the DM's tone, speaking rate, and hesitation. A reader should care because, if true, the result shows that passive, non-intrusive assistants for meetings, calendars, and other group settings are within reach of existing models, and that the audio channel carries information that text transcription throws away.

What carries the argument

The load-bearing machinery is a real-time tool-calling loop: gameplay audio is fed to a multimodal model in 10-second intervals appended to one running conversation (capped at 15 minutes), and each round the model must write chain-of-thought reasoning and then either call a tool or output "None." The three tasks are exposed as functions (search D&D data, manage NPCs on a virtual stage, generate an improvised NPC), so the model can only "speak" by making suggestions through those calls. What makes the paradigm work is the reasoning step: the model formulates an internal belief about the conversational goal before acting, and the paper shows that removing that step costs most of the performance. The audio itself carries the decisive cues, speaker identity and in-character voice for the stage-director task, pauses and filler words for NPC generation, which is why transcript-based systems hit a ceiling that the audio models pass.

What would settle it

Take the recorded gameplay audio and strip out exactly the cues the paper says matter, for example flattening pitch and speaking rate or scrambling which voice is the DM's while keeping the words identical, and rerun GPT-4o on the audio-input setting; if its F1 does not drop materially below 0.369, the audio-channel advantage is not being driven by tone, pause, and speaker-identity cues as claimed.

Watch

Extended reading notes

Core claim

On the paper's own terms, the discovery is that overhearing-agent behavior, inferring what a group of people needs and acting on it without joining the conversation, is not something that must be engineered task by task; it emerges in sufficiently large audio-language models. The paper argues this is specifically an audio-channel ability: a model that hears the game can tell when the Dungeon Master is speaking in character versus out of character, can notice pauses and fillers that signal the DM is improvising, and can track long-term conversational goals across 15-minute windows of context. Three lines of evidence carry the claim: the audio-over-text performance gap for the largest models, the collapse (more than 70% average F1 loss) when chain-of-thought reasoning is removed, and the failure-mode analysis showing that text-only pipelines either hallucinate NPC dialogue or slip into a "conversational default" of replying to the players directly. Smaller models and an ASR-pretrained audio model do not show the ability, which the paper attributes to instruction-tuning and pretraining objectives that do not prepare models for the overhearing role.

Load-bearing premise

The evaluation counts a model suggestion as correct when it falls within 300 seconds of a matching human-annotated gold suggestion, and those gold suggestions come from the Dungeon Master's own stopwatch notes plus players' subjective ratings of helpfulness, so the whole "emergent ability" conclusion rests on that window and those human judgments being trustworthy.

Editorial extensions

If this is right

  • If the claim holds, transcription-based pipelines are not just a practical shortcut but a ceiling: any task that depends on who said something, how they said it, or when they hesitated will need models that hear the audio itself.
  • Real-time overhearing agents must budget for chain-of-thought reasoning, since removing it costs more than 70% of F1 on average; the paper's speed numbers show the two can overlap, but only within a narrow window.
  • The paradigm generalizes by swapping tools: the same passive-listening model, given calendar or meeting tools instead of D&D tools, is claimed to direct the same emergent reasoning toward scheduling and meeting support.
  • Smaller audio models as they exist today are not usable for this task, they degenerate into repeated suggestions or silence, so the paper's implication is that distillation, not just scaling, is the next lever.

Reading between the lines

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

  • Editorial inference: the 300-second matching window is generous enough that near-duplicate suggestions minutes apart count as hits; tightening the window to, say, 60 seconds would likely compress the gap between models and is the single most direct robustness check a skeptic could run.
  • Editorial inference: the audio advantage should be portable to other group settings with high cognitive load, such as negotiation tables, medical consultations, and live interviews, and a cheap test would be a Wizard-of-Oz calendar agent that only ever acts when conversationally appropriate, mirroring the DM-assistant design.
  • Editorial inference: the "conversational default" failure of text models suggests that instruction-tuned models carry a strong prior toward replying; an overhearing agent may need its output channel (tool calls only, never visible text) to be enforced at the decoding level, not just the prompt level.
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 / 7 minor

Summary. The paper introduces "overhearing agents," an interaction paradigm in which an LLM-based agent passively listens to human-to-human conversation and performs background tasks via tool calls, rather than conversing directly with the user. The paradigm is instantiated as a Dungeon Master's assistant for live D&D games, with three tasks: game-data retrieval, NPC stage direction, and improvised NPC generation. The authors recorded 41.5 hours of gameplay, ran five audio-language models in both audio and text-transcription conditions, and evaluated suggestions through stopwatch-based recall annotation and post-hoc human precision annotation. They report F1 scores per model and task, speed measurements, and ablations (removing reasoning, adding transcription). The headline claim is that some large audio-language models have an emergent ability to use implicit audio cues, such as speaker identity, tone, and hesitation, to perform these overhearing tasks. The paper also releases open-source libraries and project code.

Significance. The overhearing-agent paradigm is a genuinely new framing for LLM agent interaction, and the D&D setting is a rich, naturalistic test bed. The study has real strengths: a large novel dataset, human evaluation with two complementary annotation mechanisms (stopwatch recall and post-hoc precision), a moderate inter-annotator agreement of alpha=0.67, multiple model families and sizes, and a public code and library release. If the audio-cue claim were solidly supported, this would be a notable result for multimodal agents and for understanding how audio signals beyond words can be exploited by LLMs. However, as presented, the central evidence for the audio advantage is not yet convincing because the audio and text conditions for the two OpenAI models use different model checkpoints, and the open-weight models, which are checkpoint-matched, show no audio advantage. The evaluation also lacks confidence intervals and significance tests, and the 300-second matching window is loose enough to require sensitivity analysis.

major comments (3)
  1. [Appendix B, Table 5] The audio-vs-text comparison for GPT-4o and GPT-4o-mini, which is the primary quantitative support for the "implicit audio cues" claim, is confounded by model checkpoint differences. The audio condition uses gpt-4o-realtime-preview-2024-12-17 and gpt-4o-mini-realtime-preview-2024-12-17, while the text condition uses gpt-4o-2024-11-20 and gpt-4o-mini-2024-07-18. These are different model versions with potentially different post-training, tool-calling behavior, and decoding characteristics, so the F1 gaps in Figure 3 (GPT-4o: 0.369 audio vs 0.236 text; GPT-4o-mini: 0.159 vs 0.042) cannot be attributed to audio modality alone. This matters because the open-weight models, which do use identical weights across modalities, show either no audio advantage (Ultravox: text 0.343 vs audio 0.132) or near-zero performance in both modalities (Qwen2.5, Phi-4). The only positive quantitative evidence for the headline claim therefore rests on the confounded comparison. The authors should either run a checkpoint-matched comparison (for example, using a model that accepts both audio and text inputs under identical weights, or matching the text model version to the realtime-preview version if the API permits), or provide a separate controlled experiment that manipulates audio cues (e.g., same transcript with different prosody or speaker diarization) to isolate their effect, and should revise the abstract's claim if such evidence is not available.
  2. [Section 4, Figure 3] The F1 scores are reported as point estimates without confidence intervals or significance tests. Given that the dataset is a single continuous game with 14,939 turns and 940 gold suggestions, the differences between conditions may be within sampling noise; for example, Ultravox-text (0.343) and GPT-4o-audio (0.369) are close, and several small-model scores are near zero. The paper's qualitative conclusions about model ordering and about which models "significantly" benefit from audio would be substantially stronger with bootstrap confidence intervals or an appropriate paired significance test across sessions or time intervals. Without these, the reader cannot tell whether the reported F1 differences are reliable, which is load-bearing for the central claim.
  3. [Section 4, evaluation metrics] The 300-second matching window for a suggestion to be considered correct is very generous relative to both the 10-second input intervals and the real-time nature of the assistive task. Over a 41.5-hour corpus, a model that periodically guesses common entities or repeatedly emits suggestions could match gold events almost by chance, especially for the Generate NPCs task. The paper should report how F1 changes as a function of the matching window (e.g., 30s, 60s, 120s, 300s) and should present precision and recall separately for each task and model, rather than only aggregate F1 and one task-specific precision table. In addition, Section 3.3 is ambiguous about whether the 940 gold suggestions come from the DM's stopwatch events or from filtered positive post-hoc annotations; this distinction should be clarified because it directly affects how recall is computed.
minor comments (7)
  1. [Title page] The title contains a spacing typo: "T owards" should be "Towards."
  2. [Abstract] The phrase "presentoverhearingAI agents" is missing a space; it should be "present overhearing AI agents."
  3. [Section 5] The sentence "the agent in which the agent converses directly with the user" repeats "the agent" and is grammatically broken; it should be "in which the agent converses directly with the user."
  4. [Section 5] "audio-langauge models" is a typo for "audio-language models."
  5. [Section 3.1] The heading "T asks" should read "Tasks."
  6. [Figure 4] The y-axis label "Times Realtime" should be "Times real-time" and the measurement basis (relative to what reference for open-weight models) should be stated more precisely in the caption.
  7. [Table 2] The table caption refers to errors "marked in red," but the reproduced table does not show color; please mark the erroneous span or otherwise indicate it clearly in print.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the overhearing-agent results rest on external human annotations and off-the-shelf model outputs; self-citations are software/background only.

full rationale

The paper makes an empirical claim about off-the-shelf models; there is no derivation chain that reduces to its inputs. Gold suggestions were produced by DM stopwatch annotations and post-hoc player ratings (Section 3.3), independent of model outputs, and the 300-second equivalence window and the 80% string-similarity threshold are evaluation criteria applied symmetrically, not fitted to produce the headline result. The model comparisons use external checkpoints and human labels, so the central F1 scores are not constructed from the claim being tested. Self-citations (Zhu et al. 2023a for Kani; Zhu et al. 2023b for CALYPSO; Callison-Burch et al. 2022 for D&D as a challenge setting) are software/background references and are not load-bearing for the empirical finding. The audio-vs-text comparison for GPT-4o and GPT-4o-mini uses different checkpoint versions (Appendix B, Table 5), which is a potential experimental confound for the 'implicit audio cues' interpretation, but a confound is not circularity: the reported F1 values are not equivalent to the conclusion by construction, and the open-weight models with matched weights show no audio advantage. No quoted step exhibits an equation that reduces to its own input or a fitted parameter renamed as a prediction; therefore the circularity score is 0.

Assumptions & free parameters 2 free parameters · 3 assumptions · 1 invented entities

The central claim rests on the validity of the human-generated gold labels and on the evaluation thresholds used to match suggestions to gold. The models themselves are off-the-shelf, so no parameters are fitted to the central claim; the only hand-chosen numbers are the evaluation thresholds and design constants (context length, interval size).

free parameters (2)
  • Suggestion matching window = 300 seconds
    A model suggestion is counted as a hit if it occurs within 300 seconds of an equivalent gold suggestion (Section 4). This threshold directly determines recall and F1; no sensitivity analysis is reported.
  • NPC speech equivalence threshold = >80% string similarity
    NPC speech suggestions are considered equivalent to gold when string similarity exceeds 80% (Section 4). This hand-chosen tolerance affects precision and recall for the NPC Stage Director task.
assumptions (3)
  • domain assumption Human ratings on a -2 to +2 scale correspond to true helpfulness of a suggestion in context.
    The gold set and precision labels come from subjective ratings by the players; the paper treats these as ground truth without validating against an external objective measure (Section 3.3).
  • domain assumption Dungeons & Dragons gameplay is representative of overhearing-agent use cases.
    The paper generalizes the paradigm beyond D&D (calendar scheduling, meetings), but all data comes from a single D&D campaign with seven participants (Sections 1 and 5).
  • ad hoc to paper The 300-second temporal window and 80% string-similarity threshold define 'correct' suggestions.
    These thresholds are chosen by the authors without sensitivity analysis, yet they determine the reported F1 scores (Section 4).
invented entities (1)
  • Overhearing agent
    purpose: A new category of LLM-based agent that listens to human-human conversation and provides suggestions via tool calls without directly participating.
    This is a conceptual framing introduced by the paper, not a physical or mathematical entity with a falsifiable handle outside the paper.

how reviews work

0 comments
Cite this review

Pith. "Pith review of First Steps Towards Overhearing LLM Agents: A Case Study With Dungeons & Dragons Gameplay." pith.science (2026). https://pith.science/paper/GMF3HJWS

@misc{pith2026250522809,
  author       = {Pith},
  title        = {Pith review of: First Steps Towards Overhearing LLM Agents: A Case Study With Dungeons & Dragons Gameplay},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/GMF3HJWS}},
  note         = {Machine review of arXiv:2505.22809}
}
read the original abstract

Much work has been done on conversational LLM agents which directly assist human users with tasks. We present an alternative paradigm for interacting with LLM agents, which we call "overhearing agents". These overhearing agents do not actively participate in conversation -- instead, they "listen in" on human-to-human conversations and perform background tasks or provide suggestions to assist the user. In this work, we explore the overhearing agents paradigm through the lens of Dungeons & Dragons gameplay. We present an in-depth study using large multimodal audio-language models as overhearing agents to assist a Dungeon Master. We perform a human evaluation to examine the helpfulness of such agents and find that some large audio-language models have the emergent ability to perform overhearing agent tasks using implicit audio cues. Finally, we release Python libraries and our project code to support further research into the overhearing agents paradigm at https://github.com/zhudotexe/overhearing_agents.

Figures

Figures reproduced from arXiv: 2505.22809 by the authors.

Figure 1
Figure 1. Conversational AI agents (top) are presented as a chatbot with which the user con￾verses directly. We present overhearing AI agents (bottom), which instead “listen in” on human￾to-human conversations, and leverage tool call￾ing to provide suggestions to the user without needing to participate in the conversation. There has been a recent uptick in the pop￾ularity of large language model (LLM)- powered AI agents: semi… view at source ↗
Figure 2
Figure 2. Examples of suggestions and corresponding gameplay utterances for which a [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Aggregate F1 scores of each model for each overhearing agent task, separated by [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figures from the paper (4 more)
Figure 4
Figure 4. Figure 4: The relative speed of each model and modality. OpenAI models are measured relative to real time, while other models are measured relative to each other. We evaluate each system on the precision and recall of its generated suggestions, with a suggestion marked as correc…
Figure 5
Figure 5. Figure 5: A timeline of the suggestions gen￾erated by each model, alongside the gold￾labeled suggestions, for the first third of our dataset. Each line represents one suggestion. Orange lines represent Game Data Retrieval suggestions, blue lines represent NPC Stage Director sugg…
Figure 6
Figure 6. Figure 6: The main interface that was used to annotate model suggestions. The annotator [PITH_FULL_IMAGE:figures/full_fig_p022_6.png]
Figure 7
Figure 7. Figure 7: Screenshots of the virtual tabletop software used by DMs to illustrate game scenes. [PITH_FULL_IMAGE:figures/full_fig_p022_7.png]

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

38 extracted references · 14 canonical work pages

  1. [1]

    Ontological Overhearing

    Marco Aiello, Paolo Busetta, Antonia Doná, and Luciano Serafini. Ontological Overhearing . In John-Jules Ch. Meyer and Milind Tambe (eds.), Intelligent Agents VIII , pp.\ 175--189, Berlin, Heidelberg, 2002. Springer. ISBN 978-3-540-45448-9. doi:10.1007/3-540-45448-9_13

    work page doi:10.1007/3-540-45448-9_13 2002

  2. [2]

    Homogenization Effects of Large Language Models on Human Creative Ideation

    Barrett R Anderson, Jash Hemant Shah, and Max Kreminski. Homogenization Effects of Large Language Models on Human Creative Ideation . Creativity and Cognition, pp.\ 413--425, 2024. doi:10.1145/3635636.3656204. URL https://dl.acm.org/doi/10.1145/3635636.3656204

    arXiv 2024

  3. [3]

    Investigating Proactive Search Support in Conversations

    Salvatore Andolina, Valeria Orso, Hendrik Schneider, Khalil Klouche, Tuukka Ruotsalo, Luciano Gamberini, and Giulio Jacucci. Investigating Proactive Search Support in Conversations . In Proceedings of the 2018 Designing Interactive Systems Conference , DIS '18, pp.\ 1295--1307, New York, NY, USA, 2018. Association for Computing Machinery. ISBN 978-1-4503-...

    arXiv 2018

  4. [4]

    James, and Nadia Polikarpova

    Shraddha Barke, Michael B. James, and Nadia Polikarpova. Grounded copilot: How programmers interact with code-generating models. Proc. ACM Program. Lang., 7 0 (OOPSLA1), April 2023. doi:10.1145/3586030. URL https://doi-org.proxy.library.upenn.edu/10.1145/3586030

    doi:10.1145/3586030 2023

  5. [5]

    o m, Staffan Jonsson, and Staffan Bj \

    Karl Bergstr \" o m, Staffan Jonsson, and Staffan Bj \" o rk. Undercurrents: A computer-based gameplay tool to support tabletop roleplaying. In Proceedings of Nordic DiGRA 2010 Conference, Nordic DiGRA 2010, Stockholm, Sweden, August 16-17, 2010, 2010. Digital Games Research Association, 2010. URL https://dl.digra.org/index.php/dl/article/view/507

    work page 2010

  6. [6]

    Extending Multi -agent Cooperation by Overhearing

    Paolo Busetta, Luciano Serafini, Dhirendra Singh, and Floriano Zini. Extending Multi -agent Cooperation by Overhearing . In Carlo Batini, Fausto Giunchiglia, Paolo Giorgini, and Massimo Mecella (eds.), Cooperative Information Systems , pp.\ 40--52, Berlin, Heidelberg, 2001. Springer. ISBN 978-3-540-44751-1. doi:10.1007/3-540-44751-2_5

    work page doi:10.1007/3-540-44751-2_5 2001

  7. [7]

    Martin, Daphne Ippolito, Suma Bailis, and David Reitter

    Chris Callison-Burch, Gaurav Singh Tomar, Lara J. Martin, Daphne Ippolito, Suma Bailis, and David Reitter. Dungeons and dragons as a dialog challenge for artificial intelligence. In Yoav Goldberg, Zornitsa Kozareva, and Yue Zhang (eds.), Proceedings of the 2022 Conference on Empirical Methods in Natural Language Processing, pp.\ 9379--9393, Abu Dhabi, Uni...

    doi:10.18653/v1/2022.emnlp-main.637 2022

  8. [8]

    Mark Chen, Jerry Tworek, Heewoo Jun, Qiming Yuan, Henrique Ponde de Oliveira Pinto, Jared Kaplan, Harri Edwards, Yuri Burda, Nicholas Joseph, Greg Brockman, Alex Ray, Raul Puri, Gretchen Krueger, Michael Petrov, Heidy Khlaaf, Girish Sastry, Pamela Mishkin, Brooke Chan, Scott Gray, Nick Ryder, Mikhail Pavlov, Alethea Power, Lukasz Kaiser, Mohammad Bavarian...

    arXiv 2021

Show all 38 references
  1. [9]

    D&D Basic Rules

    Jeremy Crawford, Mike Mearls, and Christopher Perkins. D&D Basic Rules. Wizards of the Coast, Renton, WA, 2018. URL https://media.wizards.com/2018/dnd/downloads/DnD_BasicRules_2018.pdf

    2018

  2. [10]

    Ferreira, Levi H

    Lucas N. Ferreira, Levi H. S. Lelis, and Jim Whitehead. Computer-generated music for tabletop role-playing games. In Proceedings of the Sixteenth AAAI Conference on Artificial Intelligence and Interactive Digital Entertainment , AIIDE '20, pp.\ 59--65. AAAI Press, 2020. ISBN 9...

    2020

  3. [11]

    Introducing ultravox v0.5: Taking the lead in speech understanding, 2025

    Fixie AI . Introducing ultravox v0.5: Taking the lead in speech understanding, 2025. URL https://www.ultravox.ai/blog/ultravox-v0-5-taking-the-lead-in-speech-understanding. Accessed: 2025-05-07

    2025

  4. [12]

    Yannakakis

    Roberto Gallotta, Graham Todd, Marvin Zammit, Sam Earle, Antonios Liapis, Julian Togelius, and Georgios N. Yannakakis. Large Language Models and Games : A Survey and Roadmap . IEEE Transactions on Games, pp.\ 1--18, 2024. ISSN 2475-1510. doi:10.1109/TG.2024.3461510. URL https:...

    2024

  5. [13]

    The curious case of neural text degeneration

    Ari Holtzman, Jan Buys, Li Du, Maxwell Forbes, and Yejin Choi. The curious case of neural text degeneration. In 8th International Conference on Learning Representations, ICLR 2020, Addis Ababa, Ethiopia, April 26-30, 2020 . OpenReview.net, 2020. URL https://openreview.net/foru...

    2020

  6. [14]

    Creative Writing with an AI - Powered Writing Assistant : Perspectives from Professional Writers , 2022

    Daphne Ippolito, Ann Yuan, Andy Coenen, and Sehmon Burnam. Creative Writing with an AI - Powered Writing Assistant : Perspectives from Professional Writers , 2022. URL https://arxiv.org/abs/2211.05030

    2022 arXiv

  7. [15]

    Towards Computational Support with Language Models for TTRPG Game Masters

    Jack Kelly, Michael Mateas, and Noah Wardrip-Fruin. Towards Computational Support with Language Models for TTRPG Game Masters . In Proceedings of the 18th International Conference on the Foundations of Digital Games , FDG '23, pp.\ 1--4, New York, NY, USA, 2023. Association fo...

    2023

  8. [16]

    A review of prominent paradigms for LLM -based agents: Tool use, planning (including RAG ), and feedback learning

    Xinzhe Li. A review of prominent paradigms for LLM -based agents: Tool use, planning (including RAG ), and feedback learning. In Owen Rambow, Leo Wanner, Marianna Apidianaki, Hend Al-Khalifa, Barbara Di Eugenio, and Steven Schockaert (eds.), Proceedings of the 31st Internation...

    2025

  9. [17]

    Moira McGregor and John C. Tang. More to Meetings : Challenges in Using Speech - Based Technology to Support Meetings . Proceedings of the 2017 ACM Conference on Computer Supported Cooperative Work and Social Computing, pp.\ 2208--2220, 2017. doi:10.1145/2998181.2998335. URL h...

    2017

  10. [18]

    Phi-4-mini technical report: Compact yet powerful multimodal language models via mixture-of-loras, 2025

    Microsoft , Abdelrahman Abouelenin, Atabak Ashfaq, Adam Atkinson, Hany Awadalla, Nguyen Bach, Jianmin Bao, Alon Benhaim, Martin Cai, Vishrav Chaudhary, Congcong Chen, Dong Chen, Dongdong Chen, Junkun Chen, Weizhu Chen, Yen-Chun Chen, Yi ling Chen, Qi Dai, Xiyang Dai, Ruchao Fa...

    2025 arXiv

  11. [19]

    Novick and Karen Ward

    David G. Novick and Karen Ward. Mutual Beliefs of Multiple Conversants : A Computational Model of Collaboration in Air Traffic Control . In Proceedings of the Eleventh National Conference on Artificial Intelligence, AAAI'93, pp.\ 196–201. AAAI Press, 1993. ISBN 0262510715

    1993

  12. [20]

    Gpt-4o system card, 2024

    OpenAI . Gpt-4o system card, 2024. URL https://arxiv.org/abs/2410.21276

    2024 arXiv

  13. [21]

    Bardo: Emotion - Based Music Recommendation for Tabletop Role - Playing Games

    Rafael Padovani, Lucas Ferreira, and Levi Lelis. Bardo: Emotion - Based Music Recommendation for Tabletop Role - Playing Games . In Proceedings of the AAAI Conference on Artificial Intelligence and Interactive Digital Entertainment , volume 13, pp.\ 214--220, 2017. doi:10.1609...

    2017 doi

  14. [22]

    Robust speech recognition via large-scale weak supervision

    Alec Radford, Jong Wook Kim, Tao Xu, Greg Brockman, Christine McLeavey, and Ilya Sutskever. Robust speech recognition via large-scale weak supervision. In Andreas Krause, Emma Brunskill, Kyunghyun Cho, Barbara Engelhardt, Sivan Sabato, and Jonathan Scarlett (eds.), Internation...

    2023

  15. [23]

    Toolformer: Language models can teach themselves to use tools

    Timo Schick, Jane Dwivedi - Yu, Roberto Dess \` , Roberta Raileanu, Maria Lomeli, Eric Hambro, Luke Zettlemoyer, Nicola Cancedda, and Thomas Scialom. Toolformer: Language models can teach themselves to use tools. In Alice Oh, Tristan Naumann, Amir Globerson, Kate Saenko, Morit...

    2023

  16. [24]

    Understanding by addressees and overhearers

    Michael F Schober and Herbert H Clark. Understanding by addressees and overhearers. Cognitive Psychology, 21 0 (2): 0 211--232, 1989. ISSN 0010-0285. doi:10.1016/0010-0285(89)90008-X. URL https://www.sciencedirect.com/science/article/pii/001002858990008X

    1989

  17. [25]

    Abigail See, Aneesh Pappu, Rohun Saxena, Akhila Yerukola, and Christopher D. Manning. Do massively pretrained language models make better storytellers? In Mohit Bansal and Aline Villavicencio (eds.), Proceedings of the 23rd Conference on Computational Natural Language Learning...

    2019 doi

  18. [26]

    Investigating Users ' Preferences and Expectations for Always - Listening Voice Assistants

    Madiha Tabassum, Tomasz Kosiński, Alisa Frik, Nathan Malkin, Primal Wijesekera, Serge Egelman, and Heather Richter Lipford. Investigating Users ' Preferences and Expectations for Always - Listening Voice Assistants . Proceedings of the ACM on Interactive, Mobile, Wearable and ...

    2019 doi

  19. [27]

    Chi, Quoc V

    Jason Wei, Xuezhi Wang, Dale Schuurmans, Maarten Bosma, Brian Ichter, Fei Xia, Ed H. Chi, Quoc V. Le, and Denny Zhou. Chain-of-thought prompting elicits reasoning in large language models. In Sanmi Koyejo, S. Mohamed, A. Agarwal, Danielle Belgrave, K. Cho, and A. Oh (eds.), Ad...

    2022

  20. [28]

    Willard and Rémi Louf

    Brandon T. Willard and Rémi Louf. Efficient guided generation for large language models, 2023. URL https://arxiv.org/abs/2307.09702

    2023 arXiv

  21. [29]

    Qwen2.5-omni technical report, 2025

    Jin Xu, Zhifang Guo, Jinzheng He, Hangrui Hu, Ting He, Shuai Bai, Keqin Chen, Jialin Wang, Yang Fan, Kai Dang, Bin Zhang, Xiong Wang, Yunfei Chu, and Junyang Lin. Qwen2.5-omni technical report, 2025. URL https://arxiv.org/abs/2503.20215

    2025 arXiv

  22. [30]

    Narasimhan, and Yuan Cao

    Shunyu Yao, Jeffrey Zhao, Dian Yu, Nan Du, Izhak Shafran, Karthik R. Narasimhan, and Yuan Cao. React: Synergizing reasoning and acting in language models. In The Eleventh International Conference on Learning Representations, ICLR 2023, Kigali, Rwanda, May 1-5, 2023 . OpenRevie...

    2023

  23. [31]

    Wordcraft: Story writing with large language models

    Ann Yuan, Andy Coenen, Emily Reif, and Daphne Ippolito. Wordcraft: Story writing with large language models. In Proceedings of the 27th International Conference on Intelligent User Interfaces, IUI '22, pp.\ 841–852, New York, NY, USA, 2022. Association for Computing Machinery....

    2022

  24. [32]

    I cast detect thoughts: Learning to converse and guide with intents and theory-of-mind in dungeons and dragons

    Pei Zhou, Andrew Zhu, Jennifer Hu, Jay Pujara, Xiang Ren, Chris Callison-Burch, Yejin Choi, and Prithviraj Ammanabrolu. I cast detect thoughts: Learning to converse and guide with intents and theory-of-mind in dungeons and dragons. In Anna Rogers, Jordan Boyd-Graber, and Naoak...

    2023

  25. [33]

    Kani: A lightweight and highly hackable framework for building language model applications

    Andrew Zhu, Liam Dugan, Alyssa Hwang, and Chris Callison-Burch. Kani: A lightweight and highly hackable framework for building language model applications. In Liling Tan, Dmitrijs Milajevs, Geeticka Chauhan, Jeremy Gwinnup, and Elijah Rippeth (eds.), Proceedings of the 3rd Wor...

    2023 doi

  26. [34]

    CALYPSO : LLMs as dungeon masters' assistants

    Andrew Zhu, Lara Martin, Andrew Head, and Chris Callison-Burch. CALYPSO : LLMs as dungeon masters' assistants. In Proceedings of the Nineteenth AAAI Conference on Artificial Intelligence and Interactive Digital Entertainment , volume 19 of AIIDE '23 , pp.\ 380--390, Salt Lake ...

    2023 doi

  27. [35]

    write newline

    " write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 gl...

  28. [36]

    @esa (Ref

    \@ifxundefined[1] #1\@undefined \@firstoftwo \@secondoftwo \@ifnum[1] #1 \@firstoftwo \@secondoftwo \@ifx[1] #1 \@firstoftwo \@secondoftwo [2] @ #1 \@temptokena #2 #1 @ \@temptokena \@ifclassloaded agu2001 natbib The agu2001 class already includes natbib coding, so you should ...

  29. [37]

    \@lbibitem[] @bibitem@first@sw\@secondoftwo \@lbibitem[#1]#2 \@extra@b@citeb \@ifundefined br@#2\@extra@b@citeb \@namedef br@#2 \@nameuse br@#2\@extra@b@citeb \@ifundefined b@#2\@extra@b@citeb @num @parse #2 @tmp #1 NAT@b@open@#2 NAT@b@shut@#2 \@ifnum @merge>\@ne @bibitem@firs...

  30. [38]

    @open @close @open @close and [1] URL: #1 \@ifundefined chapter * \@mkboth \@ifxundefined @sectionbib * \@mkboth * \@mkboth\@gobbletwo \@ifclassloaded amsart * \@ifclassloaded amsbook * \@ifxundefined @heading @heading NAT@ctr thebibliography [1] @ \@biblabel @NAT@ctr \@bibset...

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.