pith. sign in

arxiv: 2606.11051 · v1 · pith:CPKV6YYTnew · submitted 2026-06-09 · 💻 cs.SE · cs.HC

Making Software Meaningful

Eagon Meng , Abutalib Namazov , Carmel Schare , Alcino Cunha , Daniel Jackson This is my paper

Pith reviewed 2026-06-27 12:16 UTC · model grok-4.3

classification 💻 cs.SE cs.HC
keywords explicit meaningdomain phenomenasoftware behaviorusabilitymodularityaccountabilitystakeholder vocabularysoftware engineering
0
0 comments X

The pith

Committing to explicit meaning—via a shared representation of domain individuals, actions and facts—improves software usability, modularity and accountability.

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

The paper claims that software can be improved across three separate dimensions by adopting one practice: constructing and agreeing on an explicit representation of its behavior as observed in the application domain. This representation uses individuals, the actions they participate in, and the facts produced by those actions as a common vocabulary that grounds all discussion, artifacts and activities among stakeholders. Actions are further partitioned into concepts to create larger units of meaning. The approach is illustrated through three examples: aligning users and designers on the same meaning for usability gains, mapping meaning units to code units when generating software with large language models, and defining agent behavior through a code of conduct expressed in those same terms.

Core claim

Adopting a commitment to explicit meaning entails building and maintaining a representation of software behavior observed in the domain of application, where the phenomena are individuals, actions they participate in, and facts that result from actions; these can be organized into concepts by partitioning the set of actions. This single representation then serves as the grounding vocabulary for all discourse about the software. The paper shows how this commitment produces concrete gains in usability by aligning user and designer views, in modularity by letting large language models generate code that maps directly to meaning units, and in accountability by having agents adhere to a meaning-b

What carries the argument

The explicit meaning representation, consisting of domain individuals, actions, and resulting facts, organized into concepts by partitioning actions, which grounds all stakeholder discourse and artifacts.

If this is right

  • Aligning users and designers on one shared meaning representation improves software usability.
  • Mapping units of meaning directly to units of code lets large language models produce modular and legible implementations.
  • Expressing an agent's intended behavior as a code of conduct in the same meaning vocabulary makes the agent accountable.
  • All artifacts and activities become grounded in the same vocabulary of individuals, actions, and facts.

Where Pith is reading between the lines

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

  • The same vocabulary could reduce ambiguity when multiple teams or organizations must coordinate on requirements.
  • It offers a potential bridge between informal domain descriptions and more formal verification techniques without requiring full logical formalization upfront.
  • Legacy systems might be made more maintainable by first extracting and documenting their implicit meaning representation before refactoring.

Load-bearing premise

Stakeholders can feasibly construct, agree upon, and maintain a shared representation of software behavior using domain individuals, actions, and facts without prohibitive cost, persistent disagreement, or loss of needed expressiveness.

What would settle it

A concrete software project in which stakeholders attempt to build and use such a domain-phenomena representation yet produce no measurable gains in usability, modularity, or accountability, or encounter irresolvable disagreement that halts progress.

read the original abstract

Adopting a single measure can improve the usability, modularity and accountability of software: a commitment to explicit meaning. This entails constructing and agreeing upon a representation of the behavior of the software, as observed in the domain of application. The phenomena comprising this behavior become a vocabulary that grounds all discourse about the software, among all stakeholders, and for all artifacts and activities. These phenomena are individuals; actions they participate in; and facts that result from actions. They can be organized, by partitioning the set of actions, into concepts, offering larger units of meaning. Examples of exploiting meaning are given in three areas: designing for usability (by aligning user and designer on a single shared meaning); generating modular code with LLMs (by mapping units of meaning to units of code, achieving not only modularity but also legibility); and making agents accountable (by having them adhere to a code of conduct that defines their intended behavior).

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

2 major / 2 minor

Summary. The manuscript proposes that committing to explicit meaning representations—constructed from domain phenomena as individuals, actions, and facts, then organized into concepts—can serve as a unifying measure to improve software usability, modularity, and accountability. It illustrates the idea through three application areas: aligning user and designer understandings for usability design, mapping meaning units to code units when generating software with LLMs, and enforcing agent behavior via an explicit code of conduct.

Significance. If the central proposal holds, the work could offer a foundational conceptual tool for software engineering by grounding all artifacts and stakeholder discourse in a shared domain vocabulary, potentially reducing fragmentation across design, implementation, and operation. The paper's conceptual breadth across usability, LLM-assisted development, and accountability is a strength, as is its emphasis on phenomena-based representations rather than ad-hoc abstractions.

major comments (2)
  1. [Abstract] Abstract and the three illustrative areas: the central claim that explicit meaning improves the three properties rests on the unexamined feasibility of stakeholders constructing, agreeing upon, and maintaining a shared representation of behavior; the examples supply no discussion of mechanisms for resolving disagreements or bounding maintenance costs, leaving the load-bearing practical assumption unaddressed.
  2. [LLM code generation area] The LLM code-generation example: the assertion that mapping units of meaning to units of code yields both modularity and legibility is presented without a concrete mapping procedure, comparison to existing modularization techniques, or demonstration that the resulting code remains correct with respect to the original meaning representation.
minor comments (2)
  1. The distinction between 'facts that result from actions' and the concepts formed by partitioning actions could be clarified with a short formal or diagrammatic example early in the text.
  2. The manuscript would benefit from explicit pointers to related work in domain-driven design or ontology-based software engineering to situate the proposal.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the detailed and thoughtful report. We address the two major comments point by point below, clarifying the conceptual scope of the manuscript.

read point-by-point responses

  1. Referee: [Abstract] Abstract and the three illustrative areas: the central claim that explicit meaning improves the three properties rests on the unexamined feasibility of stakeholders constructing, agreeing upon, and maintaining a shared representation of behavior; the examples supply no discussion of mechanisms for resolving disagreements or bounding maintenance costs, leaving the load-bearing practical assumption unaddressed.

    Authors: The manuscript advances a conceptual proposal that explicit meaning representations, once established, can serve as a unifying measure across usability, modularity, and accountability. The central claim concerns the downstream benefits of grounding artifacts and discourse in such a representation; it does not purport to supply or evaluate the upstream processes of construction, stakeholder agreement, or ongoing maintenance. These activities are treated as prerequisites whose detailed mechanisms (including disagreement resolution and cost bounding) lie outside the paper's scope. The examples illustrate exploitation of an already-available meaning vocabulary rather than its creation. revision: no

  2. Referee: [LLM code generation area] The LLM code-generation example: the assertion that mapping units of meaning to units of code yields both modularity and legibility is presented without a concrete mapping procedure, comparison to existing modularization techniques, or demonstration that the resulting code remains correct with respect to the original meaning representation.

    Authors: The LLM example is offered at an illustrative level to show how an explicit vocabulary of individuals, actions, facts, and concepts could supply natural decomposition boundaries for generated code. No concrete mapping algorithm is supplied because the contribution is the organizing principle itself, not an implementation recipe; any such procedure would be domain- and LLM-specific. The paper does not compare against existing modularization methods because it positions meaning-based units as a complementary rather than competing criterion. Correctness is preserved by construction when the meaning representation functions as the authoritative specification and the mapping remains faithful to it. revision: no

Circularity Check

0 steps flagged

No significant circularity

full rationale

The paper is a position paper advancing a conceptual proposal for explicit meaning representations in software. It contains no equations, fitted parameters, predictions of derived quantities, or load-bearing self-citations. The argument is presented as an independent commitment to domain phenomena (individuals, actions, facts, concepts) rather than any derivation that reduces to its own inputs by construction.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

The abstract introduces no explicit free parameters, mathematical axioms, or new postulated entities; the framework is presented as a conceptual commitment rather than a formal system with fitted quantities.

pith-pipeline@v0.9.1-grok · 5688 in / 1123 out tokens · 20858 ms · 2026-06-27T12:16:22.517686+00:00 · methodology

discussion (0)

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

Reference graph

Works this paper leans on

52 extracted references · 22 canonical work pages · 4 internal anchors

  1. [1]

    Jean-Raymond Abrial. 1974. Data Semantics. InData Base Management, IFIP Working Conference (North-Holland Publishing Company), J. W. Klimbie and K. L. Koffeman (Eds.). North-Holland, Amsterdam, 1–60

    1974

  2. [2]

    1996.The B-Book: Assigning Programs to Mean- ings

    Jean-Raymond Abrial. 1996.The B-Book: Assigning Programs to Mean- ings. Cambridge University Press, Cambridge, UK

    1996

  3. [3]

    Patrick Behm, Paul Benoit, Alain Faivre, and Jean-Marc Meynadier

  4. [4]

    In FM’99 — Formal Methods: World Congress on Formal Methods in the Development of Computing Systems (Lecture Notes in Computer Science, Vol

    Météor: A Successful Application of B in a Large Project. In FM’99 — Formal Methods: World Congress on Formal Methods in the Development of Computing Systems (Lecture Notes in Computer Science, Vol. 1708), Jeannette M. Wing, Jim Woodcock, and Jim Davies (Eds.). Springer, Berlin, Heidelberg, 369–387. doi:10.1007/3-540-48119-2_22

    work page doi:10.1007/3-540-48119-2_22

  5. [5]

    2019.Race After Technology: Abolitionist Tools for the New Jim Code

    Ruha Benjamin. 2019.Race After Technology: Abolitionist Tools for the New Jim Code. Polity, Cambridge, UK and Medford, MA

    2019

  6. [6]

    Nicolas Bettenburg, Sascha Just, Adrian Schröter, Cathrin Weiss, Rahul Premraj, and Thomas Zimmermann. 2008. What Makes a Good Bug Report?. InProceedings of the 16th ACM SIGSOFT International Sym- posium on Foundations of Software Engineering (SIGSOFT ’08/FSE-16). Association for Computing Machinery, New York, NY, USA, 308–318. doi:10.1145/1453101.1453146

    work page doi:10.1145/1453101.1453146 2008

  7. [7]

    Harry Brignull, Mark Leiser, Cristiana Santos, and Kosha Doshi. 2023. Deceptive Patterns – User Interfaces Designed to Trick You. https: //www.deceptive.design/

    2023

  8. [8]

    Peter P-S Chen. 1976. The entity-relationship model—toward a unified view of data.ACM Transactions on Database Systems (TODS)1, 1 (1976), 9–36

    1976

  9. [9]

    2004.User Stories Applied: For Agile Software Development

    Mike Cohn. 2004.User Stories Applied: For Agile Software Development. Addison-Wesley Professional, Boston, MA

    2004

  10. [10]

    Johan de Kleer and John Seely Brown. 1981. Mental models of phys- ical mechanisms and their acquisition. InCognitive Skills and Their Acquisition, John R. Anderson (Ed.). Lawrence Erlbaum Associates, Hillsdale, NJ, 285–309

    1981

  11. [11]

    Denning (Ed.)

    Peter J. Denning (Ed.). 1989. A Debate on Teaching Computing Science. Commun. ACM32, 12 (Dec. 1989), 1397–1414. doi:10.1145/76380.76381

    work page doi:10.1145/76380.76381 1989

  12. [12]

    Algo- rithms Ruin Everything

    Michael A. DeVito, Darren Gergle, and Jeremy Birnholtz. 2017. “Algo- rithms Ruin Everything”: #RIPTwitter, Folk Theories, and Resistance to Algorithmic Change in Social Media. InProceedings of the 2017 CHI Conference on Human Factors in Computing Systems (CHI ’17). Asso- ciation for Computing Machinery, New York, NY, USA, 3163–3174. doi:10.1145/3025453.3025659

    work page doi:10.1145/3025453.3025659 2017

  13. [13]

    Cory Doctorow. 2023. The Enshittification of TikTok.Wired(23 Jan. 2023). https://www.wired.com/story/tiktok-platforms-cory- doctorow/ Originally appeared in Doctorow’sPluralisticnewslet- ter

    2023

  14. [14]

    I always assumed that I wasn’t really that close to [her]

    Motahhare Eslami, Aimee Rickman, Kristen Vaccaro, Amirhossein Aleyasen, Andy Vuong, Karrie Karahalios, Kevin Hamilton, and Chris- tian Sandvig. 2015. “I Always Assumed That I Wasn’t Really That Close to [Her]”: Reasoning About Invisible Algorithms in News Feeds. InProceedings of the 33rd Annual ACM Conference on Human Factors in Computing Systems (CHI ’15...

    work page doi:10.1145/2702123.2702556 2015

  15. [15]

    2003.Domain-Driven Design: Tackling Complexity in the Heart of Software

    Eric Evans. 2003.Domain-Driven Design: Tackling Complexity in the Heart of Software. Addison-Wesley, Boston, MA, USA

    2003

  16. [16]

    Git Project. [n. d.].git-worktree Documentation. Git. https://git- scm.com/docs/git-worktree Accessed 15 May 2026

    2026

  17. [17]

    Karger, and David D

    Theia Henderson, David R. Karger, and David D. Clark. 2025. Graf- fiti: Enabling an Ecosystem of Personalized and Interoperable Social Applications. InProceedings of the 38th Annual ACM Symposium on User Interface Software and Technology (UIST ’25). Association for Computing Machinery, New York, NY, USA, Article 202, 21 pages. doi:10.1145/3746059.3747627

    work page doi:10.1145/3746059.3747627 2025

  18. [18]

    C. A. R. Hoare. 1985.Communicating Sequential Processes. Prentice Hall, Englewood Cliffs, NJ, USA

    1985

  19. [19]

    Hutchins, James D

    Edwin L. Hutchins, James D. Hollan, and Donald A. Norman. 1985. Direct Manipulation Interfaces.Human-Computer Interaction1, 4 (1985), 311–338. doi:10.1207/s15327051hci0104_2

    work page doi:10.1207/s15327051hci0104_2 1985

  20. [20]

    2011.Software Abstractions: Logic, Language, and Analysis(revised ed.)

    Daniel Jackson. 2011.Software Abstractions: Logic, Language, and Analysis(revised ed.). MIT Press, Cambridge, MA, USA

    2011

  21. [21]

    2021.The Essence of Software: Why Concepts Matter for Great Design

    Daniel Jackson. 2021.The Essence of Software: Why Concepts Matter for Great Design. Princeton University Press

    2021

  22. [22]

    Daniel Jackson. 2026. Why Concepts Aren’t Objects. Blog post. https: //essenceofsoftware.com/posts/concepts-and-oop/

    2026

  23. [23]

    Michael Jackson. 1995. World and the machine.Proceedings - In- ternational Conference on Software Engineering(01 1995), 283–292. doi:10.1145/225014.225041

    work page doi:10.1145/225014.225041 1995

  24. [24]

    Michael A. Jackson. 2001.Problem Frames: Analysing and Structuring Software Development Problems. Addison-Wesley, Boston, MA, USA

    2001

  25. [25]

    Andrej Karpathy. 2026. autoresearch: AI Agents Running Research Ex- periments on Code. GitHub repository. https://github.com/karpathy/ autoresearch

    2026

  26. [26]

    Andrew J. Ko, Robin Abraham, Laura Beckwith, Alan Blackwell, Mar- garet Burnett, Martin Erwig, Christopher Scaffidi, Joseph Lawrance, Henry Lieberman, Brad Myers, Mary Beth Rosson, Gregg Rothermel, Mary Shaw, and Susan Wiedenbeck. 2011. The State of the Art in End-User Software Engineering.Comput. Surveys43, 3, Article 21 (2011), 44 pages. doi:10.1145/192...

    work page doi:10.1145/1922649.1922658 2011

  27. [27]

    Ko, Robert DeLine, and Gina Venolia

    Andrew J. Ko, Robert DeLine, and Gina Venolia. 2007. Information Needs in Collocated Software Development Teams. InProceedings of the 29th International Conference on Software Engineering (ICSE ’07). IEEE Computer Society, 344–353. doi:10.1109/ICSE.2007.45

    work page doi:10.1109/icse.2007.45 2007

  28. [28]

    Chris Lu, Cong Lu, Robert Tjarko Lange, Jakob Foerster, Jeff Clune, and David Ha. 2024. The AI Scientist: Towards Fully Automated Open- Ended Scientific Discovery.arXiv preprint arXiv:2408.06292(2024). arXiv:2408.06292 [cs.AI] doi:10.48550/arXiv.2408.06292

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.2408.06292 2024

  29. [29]

    Friedman, Eli Lucherini, Jonathan Mayer, Marshini Chetty, and Arvind Narayanan

    Arunesh Mathur, Gunes Acar, Michael J. Friedman, Elena Lucherini, Jonathan Mayer, Marshini Chetty, and Arvind Narayanan. 2019. Dark Patterns at Scale: Findings from a Crawl of 11K Shopping Websites. Proceedings of the ACM on Human-Computer Interaction3, CSCW, Article 81 (2019), 32 pages. doi:10.1145/3359183

    work page doi:10.1145/3359183 2019

  30. [30]

    Eagon Meng and Daniel Jackson. 2025. What You See Is What It Does: A Structural Pattern for Legible Software. InProceedings of the 2025 ACM SIGPLAN International Symposium on New Ideas, New Paradigms, and Reflections on Programming and Software (Onward! ’25). Association for Computing Machinery, New York, NY, USA, 178–193. doi:10.1145/3759429.3762628

    work page doi:10.1145/3759429.3762628 2025

  31. [31]

    Merrill and Will Oremus

    Jeremy B. Merrill and Will Oremus. 2021. Five Points for Anger, One for a ‘Like’: How Facebook’s Formula Fostered Rage and Misinformation. The Washington Post(26 Oct. 2021). https://www.washingtonpost. com/technology/2021/10/26/facebook-angry-emoji-algorithm/

    2021

  32. [32]

    Bertrand Meyer. 2026. AI for Software Engineering: From Proba- ble to Provable.Commun. ACM(2026). arXiv:2511.23159 [cs.SE] doi:10.48550/arXiv.2511.23159 To appear; preprint available as arXiv:2511.23159. 14 Making Software Meaningful

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.2511.23159 2026

  33. [33]

    Michael J. Muller. 2002. Participatory Design: The Third Space in HCI. InThe Human-Computer Interaction Handbook: Fundamentals, Evolving Technologies and Emerging Applications, Julie A. Jacko and Andrew Sears (Eds.). Lawrence Erlbaum Associates, Mahwah, NJ, 1051–1068

    2002

  34. [34]

    Bonnie A. Nardi. 1993.A Small Matter of Programming: Perspectives on End User Computing. MIT Press, Cambridge, MA

    1993

  35. [35]

    Donald A. Norman. 2013.The Design of Everyday Things(revised and expanded ed.). Basic Books, New York, NY, USA

    2013

  36. [36]

    MemGPT: Towards LLMs as Operating Systems

    Charles Packer, Sarah Wooders, Kevin Lin, Vivian Fang, Shishir G. Patil, Ion Stoica, and Joseph E. Gonzalez. 2023. MemGPT: Towards LLMs as Operating Systems.arXiv preprint arXiv:2310.08560(2023). arXiv:2310.08560 [cs.AI] doi:10.48550/arXiv.2310.08560

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.2310.08560 2023

  37. [37]

    O’Brien, Carrie J

    Joon Sung Park, Joseph C. O’Brien, Carrie J. Cai, Meredith Ringel Mor- ris, Percy Liang, and Michael S. Bernstein. 2023. Generative Agents: Interactive Simulacra of Human Behavior. InProceedings of the 36th Annual ACM Symposium on User Interface Software and Technology (UIST ’23). Association for Computing Machinery, New York, NY, USA, Article 2, 22 pages...

    work page doi:10.1145/3586183.3606763 2023

  38. [38]

    Ernst, and Jonathan Jacky

    Stuart Pernsteiner, Calvin Loncaric, Emina Torlak, Zachary Tatlock, Xi Wang, Michael D. Ernst, and Jonathan Jacky. 2016. Investigating Safety of a Radiotherapy Machine Using System Models with Pluggable Checkers. InComputer Aided Verification: 28th International Conference, CA V 2016, Toronto, ON, Canada, July 17–23, 2016, Proceedings, Part II (Lecture No...

    2016

  39. [39]

    Gordon, Carina Negreanu, Christian Poelitz, Sruti Srinivasa Ragavan, and Ben Zorn

    Advait Sarkar, Andrew D. Gordon, Carina Negreanu, Christian Poelitz, Sruti Srinivasa Ragavan, and Ben Zorn. 2022. What Is It Like to Program with Artificial Intelligence?. InProceedings of the 33rd Annual Conference of the Psychology of Programming Interest Group (PPIG 2022). https://ppig.org/files/2022-PPIG-33rd-sarkar.pdf

    2022

  40. [40]

    D., Boyd, d., Friedler, S

    Andrew D. Selbst, Danah Boyd, Sorelle A. Friedler, Suresh Venkata- subramanian, and Janet Vertesi. 2019. Fairness and Abstraction in Sociotechnical Systems. InProceedings of the Conference on Fairness, Ac- countability, and Transparency (FAT* ’19). Association for Computing Machinery, New York, NY, USA, 59–68. doi:10.1145/3287560.3287598

    work page doi:10.1145/3287560.3287598 2019

  41. [41]

    2012.Routledge Interna- tional Handbook of Participatory Design

    Jesper Simonsen and Toni Robertson (Eds.). 2012.Routledge Interna- tional Handbook of Participatory Design. Routledge, New York, NY. doi:10.4324/9780203108543

    work page doi:10.4324/9780203108543 2012

  42. [42]

    Brian Cantwell Smith. 2002. The Foundations of Computing. In Computationalism: New Directions, Matthias Scheutz (Ed.). MIT Press, Cambridge, MA, USA, 23–58. https://inferenceproject.yale.edu/sites/ default/files/bcs_foundations_of_computing.pdf

    2002

  43. [43]

    2026.Computational Reflections

    Brian Cantwell Smith. 2026.Computational Reflections. MIT Press, Cambridge, MA, USA

    2026

  44. [44]

    Lucy Suchman. 2002. Located Accountabilities in Technology Produc- tion.Scandinavian Journal of Information Systems14, 2 (2002), 91–105. https://aisel.aisnet.org/sjis/vol14/iss2/7/

    2002

  45. [45]

    Federal Trade Commission

    U.S. Federal Trade Commission. 2024. Negative Option Rule. Final rule, 89 Fed. Reg. 90476, 16 C.F.R. Part 425. https://www.federalregister.gov/ documents/2024/11/15/2024-25534/negative-option-rule The 2024 amended rule was vacated by the U.S. Court of Appeals for the Eighth Circuit inCustom Communications, Inc. v. FTC, No. 24-3137 (8th Cir. July 8, 2025)

    2024

  46. [46]

    Keith Edwards, Mark W

    Stephen Voida, W. Keith Edwards, Mark W. Newman, Rebecca E. Grin- ter, and Nicolas Ducheneaut. 2006. Share and Share Alike: Exploring the User Interface Affordances of File Sharing. InProceedings of the SIGCHI Conference on Human Factors in Computing Systems (CHI ’06). Association for Computing Machinery, New York, NY, USA, 221–230. doi:10.1145/1124772.1124806

    work page doi:10.1145/1124772.1124806 2006

  47. [47]

    Guanzhi Wang, Yuqi Xie, Yunfan Jiang, Ajay Mandlekar, Chaowei Xiao, Yuke Zhu, Linxi Fan, and Anima Anandkumar. 2024. Voy- ager: An Open-Ended Embodied Agent with Large Language Mod- els.Transactions on Machine Learning Research(2024). https: //openreview.net/forum?id=ehfRiF0R3a

    2024

  48. [48]

    Steve Whittaker. 2011. Personal Information Management: From Information Consumption to Curation.Annual Review of Informa- tion Science and Technology45, 1 (2011), 1–62. doi:10.1002/aris.2011. 1440450108

    work page doi:10.1002/aris.2011 2011

  49. [49]

    1986.Understanding Comput- ers and Cognition: A New Foundation for Design

    Terry Winograd and Fernando Flores. 1986.Understanding Comput- ers and Cognition: A New Foundation for Design. Ablex Publishing Corporation, Norwood, NJ

    1986

  50. [50]

    AutoGen: Enabling Next-Gen LLM Applications via Multi-Agent Conversation

    Qingyun Wu, Gagan Bansal, Jieyu Zhang, Yiran Wu, Beibin Li, Erkang Zhu, Li Jiang, Xiaoyun Zhang, Shaokun Zhang, Jiale Liu, Ahmed Has- san Awadallah, Ryen W. White, Doug Burger, and Chi Wang. 2023. Au- toGen: Enabling Next-Gen LLM Applications via Multi-Agent Conver- sation.arXiv preprint arXiv:2308.08155(2023). arXiv:2308.08155 [cs.AI] doi:10.48550/arXiv....

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.2308.08155 2023

  51. [51]

    Shunyu Yao, Jeffrey Zhao, Dian Yu, Nan Du, Izhak Shafran, Karthik Narasimhan, and Yuan Cao. 2023. ReAct: Synergizing Reason- ing and Acting in Language Models. InProceedings of the 11th International Conference on Learning Representations (ICLR 2023). arXiv:2210.03629 [cs.CL] https://arxiv.org/abs/2210.03629

    Pith/arXiv arXiv 2023

  52. [52]

    Matteo Zignani, Sabrina Gaito, and Gian Paolo Rossi. 2018. Follow the Mastodon: Structure and Evolution of a Decentralized Online Social Network. InProceedings of the International AAAI Conference on Web and Social Media, Vol. 12. 541–550. doi:10.1609/icwsm.v12i1.14988 15

    work page doi:10.1609/icwsm.v12i1.14988 2018