Human-level play in the game of <i>Diplomacy</i> by combining language models with strategic reasoning-Reference-Cited by-同舟云学术

Human-level play in the game of Diplomacy by combining language models with strategic reasoning

Published:2022-12-09 Issue:6624 Volume:378 Page:1067-1074
ISSN:0036-8075
Container-title:Science
language:en
Short-container-title:Science

Author:

,Bakhtin Anton¹^ORCID,Brown Noam¹^ORCID,Dinan Emily¹^ORCID,Farina Gabriele¹^ORCID,Flaherty Colin¹^ORCID,Fried Daniel¹²^ORCID,Goff Andrew¹^ORCID,Gray Jonathan¹^ORCID,Hu Hengyuan¹³^ORCID,Jacob Athul Paul¹⁴^ORCID,Komeili Mojtaba¹,Konath Karthik¹,Kwon Minae¹³^ORCID,Lerer Adam¹^ORCID,Lewis Mike¹^ORCID,Miller Alexander H.¹^ORCID,Mitts Sasha¹,Renduchintala Adithya¹^ORCID,Roller Stephen¹,Rowe Dirk¹,Shi Weiyan¹⁵^ORCID,Spisak Joe¹,Wei Alexander¹⁶^ORCID,Wu David¹^ORCID,Zhang Hugh¹⁷^ORCID,Zijlstra Markus¹^ORCID

Affiliation:

1. Meta AI, 1 Hacker Way, Menlo Park, CA, USA.

2. Language Technologies Institute, Carnegie Mellon University, Pittsburgh, PA, USA.

3. Department of Computer Science, Stanford University, Stanford, CA, USA.

4. Computer Science and Artificial Intelligence Laboratory, Massachusetts Insititute of Technology, Cambridge, MA, USA.

5. Department of Computer Science, Columbia University, New York, NY, USA.

6. Department of Computer Science, University of California, Berkeley, Berkeley, CA, USA.

7. EconCS Group, Harvard University, Cambridge, MA, USA.

Abstract

Despite much progress in training artificial intelligence (AI) systems to imitate human language, building agents that use language to communicate intentionally with humans in interactive environments remains a major challenge. We introduce Cicero, the first AI agent to achieve human-level performance in Diplomacy , a strategy game involving both cooperation and competition that emphasizes natural language negotiation and tactical coordination between seven players. Cicero integrates a language model with planning and reinforcement learning algorithms by inferring players’ beliefs and intentions from its conversations and generating dialogue in pursuit of its plans. Across 40 games of an anonymous online Diplomacy league, Cicero achieved more than double the average score of the human players and ranked in the top 10% of participants who played more than one game.

Publisher

American Association for the Advancement of Science (AAAS)

Subject

Multidisciplinary

Reference89 articles.

1. Language models are few-shot learners;Brown T.;Adv. Neural Inf. Process. Syst.,2020

2. Deep Blue

3. Mastering the game of Go with deep neural networks and tree search

4. Superhuman AI for multiplayer poker

5. S. Kraus D. Lehmann Diplomat an agent in a multi agent environment: An overview in IEEE International Performance Computing and Communications Conference (IEEE Computer Society 1988) pp. 434–435.

Cited by 68 articles. 订阅此论文施引文献订阅此论文施引文献，注册后可以免费订阅5篇论文的施引文献，订阅后可以查看论文全部施引文献

1. Taking into Account Opponent’s Arguments in Human-Agent Negotiations;ACM Transactions on Interactive Intelligent Systems;2024-09-10

2. Prospects and challenges of electrochemical random-access memory for deep-learning accelerators;Current Opinion in Solid State and Materials Science;2024-09

3. Strategy Game-Playing with Size-Constrained State Abstraction;2024 IEEE Conference on Games (CoG);2024-08-05

4. Missed Connections: Lateral Thinking Puzzles for Large Language Models;2024 IEEE Conference on Games (CoG);2024-08-05

5. Human-powered AI Gym: Lessons Learned as the Test and Evaluation Team for the DARPA SHADE Program: Human-powered AI Gym;Practice and Experience in Advanced Research Computing 2024: Human Powered Computing;2024-07-17