![](https://images1.caifuzhongwen.com/images/attachement/jpg/site1/20240809/08002780efb72820209108.jpg)
如今,人工智能比以往任何时候都更为先进。然而,能力越大,责任越大。政策制定者、公民社会和行业人士都在寻求治理方式,以最大限度地减少潜在危害,并塑造安全、以人为本的人工智能赋能社会。我对其中一些努力表示赞赏,但对其他努力则持谨慎态度;加州的《前沿人工智能模型安全创新法案》(更为人所知的是SB-1047)就属于后者。这项立法的初衷是好的,但将带来意想不到的严重后果,不仅对加州如此,对整个国家都是如此。
人工智能政策必须鼓励创新,设定适当的限制,并减轻这些限制的影响。否则,相关政策在最好的情况下将无法实现其目标,在最坏的情况下会导致意想不到的严重后果。
如果该法案通过成为法律,SB-1047将损害刚刚起步的人工智能生态系统,尤其是那些相对于当今科技巨头已经处于劣势的部门:公共部门、学术界和“小型科技公司”。SB-1047将不必要地惩罚开发者,扼杀开源社区,阻碍学术人工智能研究,却无法解决它本应解决的实际问题。
首先,SB-1047将过度惩罚开发者并抑制创新。如果人工智能模型被滥用,SB-1047将追究责任方和模型原始开发者的责任。每个人工智能开发者——尤其是初出茅庐的程序员和企业家——都无法预测其模型的所有可能用途。SB-1047将迫使开发者退缩并采取防御措施——这正是我们试图避免的。
其次,SB-1047将束缚开源开发。SB-1047强制规定,所有超过一定阈值的模型都必须包含一个"终止开关",这是一种可以随时关闭程序的机制。如果开发人员担心他们下载和基于其开发的程序会被删除,他们编写代码和合作的意愿就会大打折扣。这个终止开关将摧毁开源社区——无数创新的源泉,不仅在人工智能领域,而且在各个领域,从全球定位系统到磁共振成像再到互联网本身。
第三,SB-1047将削弱公共部门和学术界的人工智能研究。开源开发对私营部门很重要,但对学术界也至关重要,原因是没有合作和对模型数据的访问,学术界就无法取得进步。以研究开放式人工智能模型的计算机科学专业学生为例。如果我们的机构无法获得适当的模型和数据,我们将如何培养下一代人工智能领导者?与大型科技公司相比,这些学生和研究人员在数据和计算方面已经处于劣势,如果再设置"终止开关",将进一步削弱他们的努力。SB-1047将为学术界的人工智能敲响丧钟,而我们本应加大对公共部门人工智能的投资。
最令人担忧的是,该法案并未解决人工智能发展的潜在危害,包括偏见和深度伪造。相反,SB-1047设定了一个任意的阈值,对使用一定算力或训练成本达1亿美元的模型进行监管。这项措施非但不能提供保障,反而会限制包括学术界在内的各个领域的创新。如今,学术界的人工智能模型低于这个阈值,但如果我们要重新平衡对私营和公共部门人工智能的投资,学术界就会受到SB-1047的监管。美国的人工智能生态系统将因此变得更糟。
我们必须反其道而行之。在过去一年与拜登(Biden)总统的多次谈话中,我都表示需要"登月精神"来推动美国的人工智能教育、研究和发展。然而,SB-1047却限制过多、过于武断,不仅会打击加州的人工智能生态系统,还会对全美的人工智能下游生态系统产生令人不安的影响。
我并非反对人工智能治理。立法对于人工智能的安全和有效发展至关重要。但人工智能政策必须促进开源开发,提出统一、合理的规则,并建立消费者信心。SB-1047并未制定相关标准。我向该法案的起草人、参议员斯科特·维纳(Scott Wiener)提出合作意向:让我们共同努力,制定人工智能立法,真正构建技术驱动、以人为本的未来社会。事实上,人工智能的未来取决于此。加州——作为先驱实体和美国最强大的人工智能生态系统的所在地——是人工智能运动的核心;加州的走向决定着全美的走向。(财富中文网)
Fortune.com上发表的评论文章中表达的观点,仅代表作者本人的观点,不代表《财富》杂志的观点和立场。
译者:中慧言-王芳
如今,人工智能比以往任何时候都更为先进。然而,能力越大,责任越大。政策制定者、公民社会和行业人士都在寻求治理方式,以最大限度地减少潜在危害,并塑造安全、以人为本的人工智能赋能社会。我对其中一些努力表示赞赏,但对其他努力则持谨慎态度;加州的《前沿人工智能模型安全创新法案》(更为人所知的是SB-1047)就属于后者。这项立法的初衷是好的,但将带来意想不到的严重后果,不仅对加州如此,对整个国家都是如此。
人工智能政策必须鼓励创新,设定适当的限制,并减轻这些限制的影响。否则,相关政策在最好的情况下将无法实现其目标,在最坏的情况下会导致意想不到的严重后果。
如果该法案通过成为法律,SB-1047将损害刚刚起步的人工智能生态系统,尤其是那些相对于当今科技巨头已经处于劣势的部门:公共部门、学术界和“小型科技公司”。SB-1047将不必要地惩罚开发者,扼杀开源社区,阻碍学术人工智能研究,却无法解决它本应解决的实际问题。
首先,SB-1047将过度惩罚开发者并抑制创新。如果人工智能模型被滥用,SB-1047将追究责任方和模型原始开发者的责任。每个人工智能开发者——尤其是初出茅庐的程序员和企业家——都无法预测其模型的所有可能用途。SB-1047将迫使开发者退缩并采取防御措施——这正是我们试图避免的。
其次,SB-1047将束缚开源开发。SB-1047强制规定,所有超过一定阈值的模型都必须包含一个"终止开关",这是一种可以随时关闭程序的机制。如果开发人员担心他们下载和基于其开发的程序会被删除,他们编写代码和合作的意愿就会大打折扣。这个终止开关将摧毁开源社区——无数创新的源泉,不仅在人工智能领域,而且在各个领域,从全球定位系统到磁共振成像再到互联网本身。
第三,SB-1047将削弱公共部门和学术界的人工智能研究。开源开发对私营部门很重要,但对学术界也至关重要,原因是没有合作和对模型数据的访问,学术界就无法取得进步。以研究开放式人工智能模型的计算机科学专业学生为例。如果我们的机构无法获得适当的模型和数据,我们将如何培养下一代人工智能领导者?与大型科技公司相比,这些学生和研究人员在数据和计算方面已经处于劣势,如果再设置"终止开关",将进一步削弱他们的努力。SB-1047将为学术界的人工智能敲响丧钟,而我们本应加大对公共部门人工智能的投资。
最令人担忧的是,该法案并未解决人工智能发展的潜在危害,包括偏见和深度伪造。相反,SB-1047设定了一个任意的阈值,对使用一定算力或训练成本达1亿美元的模型进行监管。这项措施非但不能提供保障,反而会限制包括学术界在内的各个领域的创新。如今,学术界的人工智能模型低于这个阈值,但如果我们要重新平衡对私营和公共部门人工智能的投资,学术界就会受到SB-1047的监管。美国的人工智能生态系统将因此变得更糟。
我们必须反其道而行之。在过去一年与拜登(Biden)总统的多次谈话中,我都表示需要"登月精神"来推动美国的人工智能教育、研究和发展。然而,SB-1047却限制过多、过于武断,不仅会打击加州的人工智能生态系统,还会对全美的人工智能下游生态系统产生令人不安的影响。
我并非反对人工智能治理。立法对于人工智能的安全和有效发展至关重要。但人工智能政策必须促进开源开发,提出统一、合理的规则,并建立消费者信心。SB-1047并未制定相关标准。我向该法案的起草人、参议员斯科特·维纳(Scott Wiener)提出合作意向:让我们共同努力,制定人工智能立法,真正构建技术驱动、以人为本的未来社会。事实上,人工智能的未来取决于此。加州——作为先驱实体和美国最强大的人工智能生态系统的所在地——是人工智能运动的核心;加州的走向决定着全美的走向。(财富中文网)
Fortune.com上发表的评论文章中表达的观点,仅代表作者本人的观点,不代表《财富》杂志的观点和立场。
译者:中慧言-王芳
Today, AI is more advanced than ever. With great power, though, comes great responsibility. Policymakers, alongside those in civil society and industry, are looking to governance that minimizes potential harm and shapes a safe, human-centered AI-empowered society. I applaud some of these efforts yet caution against others; California’s Safe and Secure Innovation for Frontier Artificial Intelligence Models Act, better known as SB-1047, falls into the latter category. This well-meaning piece of legislation will have significant unintended consequences, not just for California, but for the entire country.
AI policy must encourage innovation, set appropriate restrictions, and mitigate the implications of those restrictions. Policy that doesn’t will at best fall short of its goals, and at worst lead to dire, if unintended, consequences.
If passed into law, SB-1047 will harm our budding AI ecosystem, especially the parts of it that are already at a disadvantage to today’s tech giants: the public sector, academia, and “little tech.” SB-1047 will unnecessarily penalize developers, stifle our open-source community, and hamstring academic AI research, all while failing to address the very real issues it was authored to solve.
First, SB-1047 will unduly punish developers and stifle innovation. In the event of misuse of an AI model, SB-1047 holds liable the party responsible and the original developer of that model. It is impossible for each AI developer—particularly budding coders and entrepreneurs—to predict every possible use of their model. SB-1047 will force developers to pull back and act defensively—precisely what we’re trying to avoid.
Second, SB-1047 will shackle open-source development. SB-1047 mandates that all models over a certain threshold include a “kill switch,” a mechanism by which the program can be shut down at any time. If developers are concerned that the programs they download and build on will be deleted, they will be much more hesitant to write code and collaborate. This kill switch will devastate the open-source community—the source of countless innovations, not just in AI, but across sectors, ranging from GPS to MRIs to the internet itself.
Third, SB-1047 will cripple public sector and academic AI research. Open-source development is important in the private sector, but vital to academia, which cannot advance without collaboration and access to model data. Take computer science students, who study open-weight AI models. How will we train the next generation of AI leaders if our institutions don’t have access to the proper models and data? A kill switch would even further dampen the efforts of these students and researchers, already at such a data and computation disadvantage compared to Big Tech. SB-1047 will deal a death knell to academic AI when we should be doubling down on public-sector AI investment.
Most alarmingly, this bill does not address the potential harms of AI advancement, including bias and deepfakes. Instead, SB-1047 sets an arbitrary threshold, regulating models that use a certain amount of computing power or cost $100 million to train. Far from providing a safeguard, this measure will merely restrict innovation across sectors, including academia. Today, academic AI models fall beneath this threshold, but if we were to rebalance investment in private and public sector AI, academia would fall under SB-1047’s regulation. Our AI ecosystem will be worse for it.
We must take the opposite approach. In various conversations with President Biden over the past year, I have expressed the need for a “moonshot mentality” to spur our country’s AI education, research, and development. SB-1047, however, is overly and arbitrarily restrictive, and will not only chill California’s AI ecosystem but will also have troubling downstream implications for AI across the nation.
I am not anti-AI governance. Legislation is critical to the safe and effective advancement of AI. But AI policy must empower open-source development, put forward uniform and well-reasoned rules, and build consumer confidence. SB-1047 falls short of those standards. I extend an offer of collaboration to Senator Scott Wiener, the bill’s author: Let us work together to craft AI legislation that will truly build the technology-enabled, human-centered society of tomorrow. Indeed, the future of AI depends on it. The Golden State—as a pioneering entity, and home to our country’s most robust AI ecosystem—is the beating heart of the AI movement; as California goes, so goes the rest of the country.
The opinions expressed in Fortune.com commentary pieces are solely the views of their authors and do not necessarily reflect the opinions and beliefs of Fortune.