“AI治理与法律”的维度

[title]AGI万字长文（下）| 2024，分叉与洪流[heading1]下篇：2024，分叉与洪流[heading2]<6>AI地缘政治：从中美相争到“主权AI”？我去年5月份写了篇文章[《AGI|大模型与大国博弈》](http://mp.weixin.qq.com/s?__biz=MzkyMTY1MTM4Mw==&mid=2247483691&idx=2&sn=4968228195e05a00d2357742b202e3a0&chksm=c18110f3f6f699e50b584eef6240abeea6785b700e092219ed77137871b461e3e00c6684ab7b&scene=21#wechat_redirect)，今天来看几个大判断基本是对的：最先进的AGI世界模型不会开源：不仅OpenAI的GPT4，Sora不开源；连原来支持开源的Google-Gemini，Anthropic-Claude，Mistral-Large都没开源；国内的当然也没有人开源。不过，这并不意味着开源生态会失效，大概率会是开源模型会落后闭源一个代际，但会服务更广泛的各种专业应用。美国对于中国的硬件-科技限制进一步升级。不赘述了，这个是割韭菜的自媒体最爱讲的。AGI将对于全行业科技发展起到推动作用。这点目前仅仅是苗头，但AGI在各领域的科研中越来越重要是既定的事实。以及，如果AGI的能力有比较大差距，对于科技进步-经济发展将产生质变性的影响。有更好AGI的国家会有更快的全面技术进步。对于AI的立法、监管、伦理讨论大范围落后于技术发展。全世界主要国家都急于拥有有自己的AI，当然不会有人搬起石头砸自己的脚。目前唯一有点对于AI治理讨论的只有欧洲，而且也都在纸面上。我国相关法规讨论的出发点完全在于“对于舆论的影响”，根本没有触及AGI本身的伦理问题。决策路径大概是政治>经济>>AI伦理。

However,AI can increase the riskof unfair bias or discrimination across a range of indicators or characteristics.Thiscould undermine public trust in AI.Product safety laws ensure that goods manufactured and placed on the market inthe UK are safe.Product-specific legislation(such as for electrical and electronicequipment,56medicaldevices,57and toys58)may apply to some products thatinclude integrated AI.However,safety risks specific to AI technologies should bemonitored closely.As the capability and adoption of AI increases,it may pose newand substantial risks that are unaddressed by existing rules.Global Innovation Index 2022,GII 2022; Global Indicators of Regulatory Governance,World Bank,2023.Demand for AI skills in jobs,OECD Science,Technology and Industry Working Papers,2021.The protected characteristics are age,disability,gender reassignment,marriage and civil partnership,race,religion orbelief,sex,and sexual orientation.Article 5(1)(a)Principles relating to processing of personal data,HM Government,2016.Electrical Equipment(Safety)Regulations,HM Government,2016.Medical Devices Regulation,HM Government,2002.Toys(Safety)Regulations,HM Government,2011.A pro-innovation approach to AI regulationConsumer rightslaw59may protect consumers where they have entered into asales contract for AI-based products and services.Certain contract terms(forexample,that goods are of satisfactory quality,fit for a particular purpose,and asdescribed)are relevant to consumer contracts.Similarly,businesses are prohibitedfrom including certain terms in consumer contracts.Tort law provides acomplementary regime that may provide redress where a civil wrong has causedharm.It is not yet clear whether consumer rights law will provide the right level ofprotection in the context of products that include integrated AI or services based on

事实上，“人工智能”的概念自1956年于美国的达特茅斯学会上被提出后，其所涵盖的理论范围及技术方法随着时代的发展也在不断扩展。如今，人工智能技术也发展出多个技术分支，应用于不同的领域中。相比于《2018年人工智能战略》，2021年《AI法案》提案对于人工智能的定义采取更加宽泛的界定标准。值得注意的是，在2022年《AI法案》妥协版本中，欧盟理事会及欧洲议会对于上述界定的观点有进一步意见，其认为“AI系统”的定义范围应适当缩窄，并侧重强调机器学习的方法。