时间、信息与人工智能

祁晓亮

doi:10.7693/wl20240601

时间、信息与人工智能

祁晓亮^1,2, ,

1 斯坦福大学物理系美国斯坦福 CA 94305;
2 路径积分科技有限公司美国库比蒂诺 CA 95014

详细信息

通讯作者:
祁晓亮,email:xlqi@stanford.edu

计量
- 文章访问数: 825
- HTML全文浏览量: 141
- PDF下载量: 2156
出版历程
- 收稿日期: 2024-05-27
- 网络出版日期: 2024-06-14

Time, information and artificial intelligence

QI Xiao-Liang^1,2, ,

1 Department of Physics, Stanford University, Stanford CA 94305, USA;
2 Path Integral Technology, Inc., Cupertino CA 95014, USA

摘要

摘要: 近年来，人工智能(AI)大语言模型取得了突飞猛进的发展，将人工智能对人类社会的影响也拓宽到了前所未有的范围。文章将从与物理学有关的两个角度——信息和时间尺度，来谈谈作者对大语言模型带来的人工智能革命的一些不成熟的见解。文中首先回顾大语言模型的基本原理和近期发展，再讨论从信息的动力学和复杂度的角度如何看待大语言模型的意义。基于人工智能模型和人类认知系统的比较，也会探讨人工智能的下一步发展方向，以及AI智能体方面的探索和发展。
- 大语言模型 /
- 人工智能 /
- 信息 /
- 复杂性 /
- 系统1 /
- 系统2
Abstract: In recent years, the rapid advances in large language models have expanded the impact of artificial intelligence (AI) on human society to an unprecedented extent. This article will discuss my preliminary insights into the AI revolution brought about by large language models from two physics-related perspectives—information and time scales. I will first review the basic principles and recent developments of large language models, and then discuss their significance from the perspective of information dynamics and complexity. Based on the comparison between AI models and the human cognitive system, I will explore the next direction for AI, as well as the exploration and development of AI agents.
- large language models /
- artificial intelligence /
- information /
- complexity /
- system1 /
- system2

HTML全文

参考文献(13)

[1]	Shannon's Source Coding Theorem. https://web.archive.org/web/20090216231139/;http://plan9.belllabs.com//cm//ms//what//shannonday//shannon1948.pdf
[2]	Vaswani A,Shazeer N,Parmar N et al. Attention Is All You Need. 2023,arXiv:1706.03762
[3]	祁晓亮. 人工智能的黎明:从信息动力学的角度看ChatGPT. https://mp.weixin.qq.com/s/DJRSqwo0cWGOAgZM4As-OQ
[4]	Kahneman D. Thinking,Fast and Slow. Macmillan,2011
[5]	Steven P. Psychon. Bull. Rev.,2014,21(5):1112
[6]	Wei J et al. Chain-of-thought Prompting Elicits Reasoning in Large Language Models. In:Advances in Neural Information Processing Systems 35,2022
[7]	Yao S Y et al. Tree of Thoughts:Deliberate Problem Solving with Large Language Models. In:Advances in Neural Information Processing Systems 36,2024
[8]	Besta M et al. Graph of Thoughts:Solving Elaborate Problems with Large Language Models. In:Proceedings of the AAAI Conference on Artificial Intelligence,2024,38(16):17682
[9]	Park J S et al. Generative Agents:Interactive Simulacra of Human Behavior. 2023,arXiv:2304.03442
[10]	Yang H,Yue S F,He Y Z. Auto-gpt for Online Decision Making:Benchmarks and Additional Opinions. 2023,arXiv:2306.02224
[11]	Wu Q Y et al. AutoGen:Enabling Next-gen LLM Applications via Multiagent Conversation Framework. 2023,arXiv:2308. 08155
[12]	Pan H N et al. Quantum Many-Body Physics Calculations with Large Language Models. 2024,arXiv:2403.03154
[13]	Andrew Ng. What's next for AI agentic workflows. https://www.youtube.com/watch?v=sal78ACtGTc