Papers
Topics
Authors
Recent
Gemini 2.5 Flash
Gemini 2.5 Flash
125 tokens/sec
GPT-4o
47 tokens/sec
Gemini 2.5 Pro Pro
43 tokens/sec
o3 Pro
4 tokens/sec
GPT-4.1 Pro
47 tokens/sec
DeepSeek R1 via Azure Pro
28 tokens/sec
2000 character limit reached

Attaining Human`s Desirable Outcomes in Human-AI Interaction via Structural Causal Games (2405.16588v1)

Published 26 May 2024 in cs.AI, cs.GT, and cs.HC

Abstract: In human-AI interaction, a prominent goal is to attain humans desirable outcome with the assistance of AI agents, which can be ideally delineated as a problem of seeking the optimal Nash Equilibrium that matches the humans desirable outcome. However, reaching the outcome is usually challenging due to the existence of multiple Nash Equilibria that are related to the assisting task but do not correspond to the humans desirable outcome. To tackle this issue, we employ a theoretical framework called structural causal game (SCG) to formalize the human-AI interactive process. Furthermore, we introduce a strategy referred to as pre-policy intervention on the SCG to steer AI agents towards attaining the humans desirable outcome. In more detail, a pre-policy is learned as a generalized intervention to guide the agents` policy selection, under a transparent and interpretable procedure determined by the SCG. To make the framework practical, we propose a reinforcement learning-like algorithm to search out this pre-policy. The proposed algorithm is tested in both gridworld environments and realistic dialogue scenarios with LLMs, demonstrating its adaptability in a broader class of problems and potential effectiveness in real-world situations.

Definition Search Book Streamline Icon: https://streamlinehq.com
References (66)
  1. Gpt-4 technical report. arXiv preprint arXiv:2303.08774, 2023.
  2. Guidelines for human-ai interaction. Proceedings of the 2019 CHI Conference on Human Factors in Computing Systems, 2019.
  3. A multitask, multilingual, multimodal evaluation of chatgpt on reasoning, hallucination, and interactivity, 2023.
  4. Updates in human-ai teams: Understanding and addressing the performance/compatibility tradeoff. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 33, pages 2429–2437, 2019.
  5. How well can llms negotiate? negotiationarena platform and analysis. arXiv, 2024.
  6. Large language models can implement policy iteration. In Advances in Neural Information Processing Systems, 2023.
  7. Understanding incentives: Mechanism design becomes algorithm design. In 2013 IEEE 54th Annual Symposium on Foundations of Computer Science, pages 618–627. IEEE, 2013.
  8. Best-response bayesian reinforcement learning with bayes-adaptive pomdps for centaurs. In Piotr Faliszewski, Viviana Mascardi, Catherine Pelachaud, and Matthew E. Taylor, editors, 21st International Conference on Autonomous Agents and Multiagent Systems, AAMAS 2022, Auckland, New Zealand, May 9-13, 2022, pages 235–243. International Foundation for Autonomous Agents and Multiagent Systems (IFAAMAS), 2022.
  9. A survey on evaluation of large language models. arXiv preprint arXiv:2307.03109, 2023.
  10. Introspective tips: Large language model for in-context decision making. arXiv preprint arXiv:2305.11598, 2023.
  11. Minigrid & miniworld: Modular & customizable reinforcement learning environments for goal-oriented tasks. CoRR, abs/2306.13831, 2023.
  12. Pareto actor-critic for equilibrium selection in multi-agent reinforcement learning, 2023.
  13. Receive, reason, and react: Drive as you say with large language models in autonomous vehicles. arXiv preprint arXiv:2310.08034, 2023.
  14. Evaluation of gpt-3.5 and gpt-4 for supporting real-world information needs in healthcare delivery. ArXiv, abs/2304.13714, 2023.
  15. A Philip Dawid. Influence diagrams for causal modelling and inference. International Statistical Review, 70(2):161–189, 2002.
  16. Zero-shot assistance in sequential decision problems. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 37, pages 11551–11559, 2023.
  17. Agent incentives: A causal perspective, 2021.
  18. Joseph Farrell. Communication, coordination and nash equilibrium. Economics Letters, 27(3):209–214, 1988.
  19. Online meta-learning. In International conference on machine learning, pages 1920–1930. PMLR, 2019.
  20. Constrained environment optimization for prioritized multi-agent navigation, 2023.
  21. Is gpt-4 a reliable rater? evaluating consistency in gpt-4 text ratings. arXiv preprint arXiv:2308.02575, 2023.
  22. Reasoning about causality in games. Artificial Intelligence, 320:103919, 2023.
  23. A general theory of equilibrium selection in games. MIT Press Books, 1, 1988.
  24. GPT-4 as an effective zero-shot evaluator for scientific figure captions. In Houda Bouamor, Juan Pino, and Kalika Bali, editors, Findings of the Association for Computational Linguistics: EMNLP 2023, pages 5464–5474, Singapore, December 2023. Association for Computational Linguistics. doi: 10.18653/v1/2023.findings-emnlp.363.
  25. Language instructed reinforcement learning for human-ai coordination. arXiv preprint arXiv:2304.07297, 2023.
  26. Discovering agents. Artificial Intelligence, page 103963, 2023.
  27. Multi-agent influence diagrams for representing and solving games. Games and economic behavior, 45(1):181–221, 2003.
  28. Reward design with language models, 2023.
  29. A systematic study and comprehensive evaluation of chatgpt on benchmark datasets, 2023.
  30. Convolutional networks for images, speech, and time series. The handbook of brain theory and neural networks, 3361(10):1995, 1995.
  31. Strategic behavior of large language models: Game structure vs. contextual framing. arXiv preprint arXiv:2309.05898, 2023.
  32. Peter JF Lucas and Linda C Van Der Gaag. Principles of expert systems. Addison Wesley Longman, 1991.
  33. Characterising decision theories with mechanised causal graphs. arXiv preprint arXiv:2307.10987, 2023.
  34. Gpteval: A survey on assessments of chatgpt and gpt-4, 2023.
  35. Pipeline psro: A scalable approach for finding approximate nash equilibria in large games. Advances in Neural Information Processing Systems, 33:20238–20248, 2020.
  36. The em algorithm—an old folk-song sung to a fast new tune. Journal of the Royal Statistical Society Series B: Statistical Methodology, 59(3):511–567, 1997.
  37. Ignorable information in multi-agent scenarios. 2008.
  38. Human-in-the-loop machine learning: a state of the art. Artificial Intelligence Review, 56(4):3005–3054, 2023.
  39. Automated evaluation of written discourse coherence using gpt-4. In Proceedings of the 18th Workshop on Innovative Use of NLP for Building Educational Applications (BEA 2023), pages 394–403, 2023.
  40. John F Nash et al. Non-cooperative games. 1950.
  41. Algorithmic mechanism design. In Proceedings of the thirty-first annual ACM symposium on Theory of computing, pages 129–140, 1999.
  42. Gpt as a financial advisor. Available at SSRN 4384861, 2023.
  43. Pareto optimality, game theory and equilibria, volume 17. Springer Science & Business Media, 2008.
  44. Judea Pearl. Causality. Cambridge university press, 2009.
  45. Learning to locomote: Understanding how environment design matters for deep reinforcement learning. In Proceedings of the 13th ACM SIGGRAPH Conference on Motion, Interaction and Games, pages 1–10, 2020.
  46. Reinhard Selten. Spieltheoretische behandlung eines oligopolmodells mit nachfrageträgheit: Teil i: Bestimmung des dynamischen preisgleichgewichts. Zeitschrift für die gesamte Staatswissenschaft/Journal of Institutional and Theoretical Economics, (H. 2):301–324, 1965.
  47. Reinhard Selten. Reexamination of the perfectness concept for equilibrium points in extensive games. International Journal of Game Theory, 4, 1975.
  48. Languagempc: Large language models as decision makers for autonomous driving. arXiv preprint arXiv:2310.03026, 2023.
  49. Reflexion: Language agents with verbal reinforcement learning. In Advances in Neural Information Processing Systems, 2023.
  50. Practical bayesian optimization of machine learning algorithms, 2012.
  51. Reasoning about reasoning by nested conditioning: Modeling theory of mind with probabilistic programs. Cognitive Systems Research, 28:80–99, 2014.
  52. Sequence to sequence learning with neural networks. Advances in Neural Information Processing Systems, 27, 2014.
  53. Hindsight expectation maximization for goal-conditioned reinforcement learning. In International Conference on Artificial Intelligence and Statistics, pages 2863–2871. PMLR, 2021.
  54. A survey on human-ai teaming with large pre-trained models, 2024.
  55. A survey on large language model based autonomous agents. arXiv preprint arXiv:2308.11432, 2023a.
  56. Influencing towards stable multi-agent interactions. In Conference on robot learning, pages 1132–1143. PMLR, 2022.
  57. Reinforcement learning to play an optimal nash equilibrium in team markov games. Advances in Neural Information Processing Systems, 15, 2002.
  58. Chatgpt for design, manufacturing, and education. Procedia CIRP, 119:7–14, 2023b.
  59. Honesty is the best policy: defining and mitigating ai deception. In Advances in Neural Information Processing Systems, 2023.
  60. The reasons that agents act: Intention and instrumental goals, 2024.
  61. Ai chains: Transparent and controllable human-ai interaction by chaining large language model prompts. Proceedings of the 2022 CHI Conference on Human Factors in Computing Systems, 2021.
  62. The rise and potential of large language model based agents: A survey, 2023.
  63. Re-examining whether, why, and how human-ai interaction is uniquely difficult to design. Proceedings of the 2020 CHI Conference on Human Factors in Computing Systems, 2020.
  64. An overview of multi-agent reinforcement learning from game theoretical perspective. arXiv preprint arXiv:2011.00583, 2020.
  65. A general approach to environment design with one agent. In Proceedings of the 21st International Joint Conference on Artificial Intelligence, pages 2002–2008, 2009.
  66. A survey of large language models. arXiv preprint arXiv:2303.18223, 2023.

Summary

We haven't generated a summary for this paper yet.

Ai Sparkles Streamline Icon: https://streamlinehq.com Summarize Now