专题:创新引领 自立自强——打造高质量科技创新策源地

基础智能:从联邦智能到基于TAO的智能系统联邦

  • 王飞跃 ,
  • 缪青海
展开
  • 1. 中国科学院大学人工智能学院,北京 100049
    2. 中国科学院自动化研究所复杂系统管理与控制国家重点实验室,北京 100190
王飞跃,研究员,研究方向为平行系统、社会计算、知识自动化等,电子信箱:feiyue.wang@ia.ac.cn

收稿日期: 2023-07-20

  修回日期: 2023-09-13

  网络出版日期: 2023-10-27

基金资助

国家重点研发计划项目(2018AAA0101502);国家自然科学基金项目(62271485)

Foundation intelligence: From federated intelligence to TAO-based intelligent systems federation

  • WANG Fei-Yue ,
  • MIAO Qinghai
Expand
  • 1. School of Artificial Intelligence, University of Chinese Academy of Sciences, Beijing 100049, China
    2. State Key Laboratory for Management and Control of Complex Systems, Institute of Automation, Chinese Academy of Sciences, Beijing 100190, China

Received date: 2023-07-20

  Revised date: 2023-09-13

  Online published: 2023-10-27

摘要

从早期使用数学工具探索神经活动机理,到采用人工神经网络模型模拟人类智能,再到基于深度网络解决特定问题,人工智能历经80年发展,当前已进入大模型时代,其特点是立足基础大模型、服务广大社会群体。基础大模型的发展面临一系列挑战:一是大模型训练所必需的数据,质与量有待提高;二是高端算力设备受限制,模型训练高电耗、高碳排放带来环境问题;三是大模型的社会化带来歧视与偏见、虚假信息传播、隐私侵犯、影响就业等社会发展问题。在当今去全球化的国际形势下,为克服以上挑战,健全包含联邦数据、联邦控制、联邦管理和联邦服务的联邦生态系统,构建基于区块链保真(trustable+reliable+usable+effective/efficient,TRUE)和去中心化自治组织和分布式自治运营(DAO)的智能系统联邦,发展基于TAO(TRUE DAO)的基础智能(foundation intelligence,FI),为大模型时代人工智能的发展提供必要和有力的支撑。

本文引用格式

王飞跃 , 缪青海 . 基础智能:从联邦智能到基于TAO的智能系统联邦[J]. 科技导报, 2023 , 41(19) : 103 -112 . DOI: 10.3981/j.issn.1000-7857.2023.19.012

Abstract

From the early use of mathematical tools to explore the mechanism of neural activity, to the use of artificial neural network models to simulate human intelligence, to the solution of specific problems based on deep networks, AI has developed over the past 80 years and has now entered the era of foundation intelligence (FI), which is characterized by its foothold in large models, serving the broad social groups. The development of large models faces a series of challenges. First, the quality and quantity of data necessary for large models training need to be improved. Second, high-end computing resource is limited, and model training has high power consumption and high carbon emissions, which brings environmental problems. Third, it is the socialization of large models that brings about social development problems such as discrimination and prejudice, the spread of false information, invasion of privacy, and impact on employment. In order to overcome the aforementioned challenges, it is essential to build a federal ecosystem (including federal data, federal control, federal management and federal services) with support of block chain and decentralized autonomous organizations (TRUE+DAO, TAO), and provide solutions for the development of artificial intelligence in the new era.

参考文献

[1] Zhou C, Li Q, Li C, et al. A comprehensive survey on pretrained foundation models: A history from BERT to ChatGPT[J]. arXiv, 2023, arXiv:2302.09419.
[2] Wang F Y, Zhang J J, Hang X, et al. Where does AlphaGo go: From Church-Turing thesis to AlphaGo thesis and beyond[J]. IEEE/CAA Journal of Automatica Sinica, 2016, 3(2): 113-120.
[3] Wang F Y, Miao Q H, Li X, et al. What does ChatGPT say: The DAO from algorithmic intelligence to linguistic intelligence[J]. IEEE/CAA Journal of Automatica Sinica, 2023, 10(3): 575-579.
[4] Ouyang L, Wu J, Jiang X, et al. Training language models to follow instructions with human feedback[J]. arXiv, 2022, arXiv:2203.02155.
[5] Touvron H, Lavril T, Izacard G, et al. LLaMA: Open and efficient foundation language models[J]. arXiv, 2023, arXiv:2302.13971.
[6] Wang F Y, Ding W W, Wang X, et al. The DAO to DeSci: AI for free, fair, and responsibility sensitive sciences [J]. IEEE Intelligent Systems, 2022, 37(2): 16-22.
[7] Ding W W, Hou J, Li J J, et al. DeSci based on Web3 and DAO: A comprehensive overview and reference model [J]. IEEE Transactions on Computational Social Systems, 2022, 9(5):1563-1573.
[8] 李娟娟, 秦蕊, 丁文文, 等 . 基于 Web3的去中心化自治组织与运营新框架[J]. 自动化学报, 2023, 49(5): 985-998.
[9] Miao Q H, Zheng W B, Lv Y S, et al. DAO to HANOI via DeSci: AI paradigm shifts from AlphaGo to ChatGPT [J]. IEEE/CAA Journal of Automatica Sinica, 2023, 10(4): 877-897.
[10] 缪青海, 王雨桐, 吕宜生, 等. McCulloch-Pitts人工神经元模型80周年纪念:思想、方法与意义[J]. 智能科学与技术学报, 2023, 5(2): 133-142.
[11] 王飞跃, 缪青海. 人工智能驱动的科学研究新范式:从 AI4S 到智能科学[J]. 中国科学院院刊, 2023, 38(4):536-540.
[12] 杨静, 王晓, 王雨桐, 等 . 平行智能与 CPSS: 三十年发展的回顾与展望[J]. 自动化学报, 2023, 49(3): 614-634
[13] Miao Q H, Lv Y S, Huang M, et al. Overview and perspective for computational learning across Syn2Real and Sim2Real[J]. IEEE/CAA Journal of Automatica Sinica, 2023, 10(3): 603-631.
[14] Wang F Y. Parallel intelligence in Metaverses: Welcome to Hanoi[J]. IEEE Intelligent Systems, 2022, 37(1): 1620.
[15] Li X, Tian Y L, Ye P J, et al. A novel scenarios engineering methodology for foundation models in Metaverse [J]. IEEE Transactions on Systems, Man, and Cybernetics: Systems, 2023, 53(4): 2148-2159.
[16] Wang F Y. New control paradigm for Industry 5.0: From big models to foundation control and management[J]. IEEE/CAA Journal of Automatica Sinica, 2023, 10(8): 1643-1646.
[17] 卢经纬, 郭超, 戴星原, 等. 问答ChatGPT之后: 超大预训练模型的机遇和挑战[J]. 自动化学报, 2023, 49(4):705-717.
[18] Wang F Y, Qin R, Chen Y Z, et al. Federated ecology:Steps toward confederated intelligence[J]. IEEE Transactions on Computational Social Systems, 2021, 8(2): 271278.
[19] Wang F Y, Zhang W S, Tian Y L, et al. Federated data: Toward new generation of credible and trustable artificial intelligence[J]. IEEE Transactions on Computational Social Systems, 2021, 8(3): 538-545.
[20] 李娟娟, 王晓, 袁勇, 等 . 基于区块链技术的数据共享方法研究进展[J]. 复杂性与智能化, 2021, 17(3): 30-37.
[21] Wang F Y, Zhu J, Qin R, et al. Federated control: Toward information security and rights protection[J]. IEEE Transactions on Computational Social Systems, 2021, 8 (4): 793-798.
[22] 朱静, 王飞跃, 田永林, 等. 联邦控制: 面向信息安全和权益保护的分布式控制方法[J].自动化学报, 2021, 47(8): 1912-1920.
[23] Wang F Y. The DAO to MetaControl for MetaSystems in Metaverses: The system of parallel control systems for knowledge automation and control intelligence in CPSS [J]. IEEE/CAA Journal of Automatica Sinica, 2022, 9 (11): 1899-1908.
[24] Wang F Y, Qin R, Li J J, et al. Federated management: Toward federated services and federated security in federated ecology[J]. Transactions on Computational Social Systems, 2021, 8(6): 1283-1290.
[25] Li J J, Qin R, Wang F Y. The future of management: DAO to smart organizations and intelligent operations[J]. IEEE Transactions on Systems, Man, and Cybernetics: Systems, 2023, 53(6): 3389-3399.
[26] Wang X X, Yang J, Wang Y T, et al. Steps toward Industry 5.0: Building“6S”parallel industries with cyberphysical-social intelligence[J]. IEEE/CAA Journal of Automatica Sinica, 2023, 10(8): 1692-1703.
[27] Lin Y L, Na X X, Wang D, et al. Mobility 5.0: Smart logistics and transportation services in Cyber-Physical-Social systems[J]. IEEE Transactions Intelligent Vehicels, 2023, 8(6): 3527-3532.
[28] Ye P J, Wang F Y. Parallel population and parallel human-A Cyber-Physical social approach[J]. IEEE Intelligent Systems, 2022, 37(5): 19-27.
[29] Wang F Y, Qin R, Wang X, et al. MetaSocieties in Metaverse: MetaEconomics and MetaManagement for MetaEnterprises and MetaCities[J]. Transactions on Computational Social Systems, 2022, 9(1): 2-7.
文章导航

/