前财政部长罗伯特·鲁宾（Robert Rubin）警告称，人工智能投资热潮可能会带来巨大的生产力提升，但也可能产生市场可能并未完全定价的金融和社会风险。
Treasury Secretary Robert Rubin warned that the artificial-intelligence investment boom could bring major productivity gains but also create financial and social risks that markets may not be fully pricing.
曾在20世纪90年代末互联网泡沫时期担任财政部长的鲁宾表示，他尤其担心人工智能生态系统中的“循环风险”。该术语是指供应商、客户和投资者之间的承诺出现重叠，例如芯片制造商和软件公司之间的重叠。
Rubin, who served as Treasury chief during the internet boom in the late 1990s, said he’s particularly concerned about “circularity risk” in the AI ecosystem. The term refers to the overlapping of commitments among suppliers, customers and investors — such as between chipmakers and software firms.
鲁宾周二在格林威治经济论坛（Greenwich Economic Forum）接受采访时表示：“其中一些非常大的人工智能公司做出了巨额承诺，然后还有很多供应商，许多供应商以这些承诺作抵押进行了借贷。如果他们无法兑现这些承诺，或者所有以这些承诺作抵押的借贷出现问题，会发生什么？这就叫做循环风险。”
“Some of these very big AI companies have enormous commitments, and then there are a lot of suppliers, and a lot of suppliers have borrowed against those commitments,” Rubin said in an interview at the Greenwich Economic Forum Tuesday. “What happens if they can’t fulfill those commitments or all those borrowed against them? It’s called circularity risk.”
鲁宾认为，出现某种中断的风险——即一方打破承诺导致连锁崩溃——是切切实存在的。这位高盛前联席董事长将其描述为“并非接近于零”。今年建设数据中心和开发人工智能软件的公司发行的债务规模如此之大，以至于许多市场参与者表示，这导致了全球借贷成本的上升。本周基准政府债券收益率进一步攀升，10年期美国国债利率创下2002年以来的最高水平。
The risk of some disruption — where one party breaks its commitments, causing cascading failures — is meaningful, according to Rubin. The former Goldman Sachs co-chairman described it as not “near zero.”The scale of debt issuance this year by firms building data centers and developing AI software has been such that many market participants say it’s contributed to an increase in borrowing costs globally. Benchmark government bond yields have climbed further this week, with 10-year Treasury rates hitting their highest level since 2002.
这反过来又推高了政府债务偿还成本，加剧了人们对美国、法国和其他国家财政可持续性的担忧。鲁宾加入克林顿总统政府时，各方对预算赤字的担忧普遍存在，他表示，对借贷规模的焦虑已经开始显现。
That’s in turn propelling government debt servicing costs, stoking concerns about fiscal sustainability in the US, France and elsewhere. Rubin, who joined President Bill Clinton’s administration at a time when worries over budget deficits were widely shared, said angst over the scale of borrowing is beginning to be evident.
他说：“我认为，目前发生的事情是，人们对我们财政状况的一些认识开始以多年来未曾有过的方式影响市场。”
“I think what’s happening right now is some realization about our fiscal situation beginning to affect markets in a way it hasn’t for a long time,” he said.
他还指出，通胀以及公众对政府解决财政问题能力的信心普遍下降。“当财政状况不利或不稳定时，可能会更广泛地影响人们对你政府处理这一问题的信心，”他说。
He also pointed to inflation and a broader loss of confidence in the government’s ability to address fiscal problems. “When you have adverse or unstable fiscal conditions, it can affect confidence more broadly in the ability of your government to deal with this problem,” he said.
他表示，这种动态最终也可能影响股市，因为不确定性往往会抑制投资和生产率。
That dynamic could ultimately matter for equity markets as well, he said — because uncertainty tends to weigh on investment and productivity.
鲁宾反驳了那种认为人工智能驱动的生产率增长能让美国通过经济增长摆脱财政困境的观点。这位前财政部长表示，虽然生产率的加速可能会推动国内生产总值（GDP）增长，但这种收益可能伴随着大量失业，尤其是白领阶层的失业。
Rubin pushed back against the argument that AI-driven productivity growth could allow the US to effectively grow its way out of its fiscal problems. While it’s possible that a productivity acceleration boosts GDP gains, the benefits could be accompanied by substantial job losses, particularly among white-collar workers, according to the former Treasury chief.
“这将给知识工作者带来沉重打击，”他预测。他说，律师、会计师和电视行业从业者都可能面临被取代的风险，这引发了关于人们如何过渡到新岗位的棘手问题。他表示，美国目前缺乏有效的项目来应对此类冲击。
“This is going to hit knowledge workers very hard,” he predicted. Lawyers, accountants and television workers could all face displacement, he said, raising difficult questions about how people transition into new roles. The US currently lacks effective programs to deal with such disruptions, he said.
“那么，人工智能能否带来更高的增长？可以——我们很可能确实能做到，”他说。“但我不认为它能做到我们所需的那样，从而彻底消除财政问题。”
“So could we have higher growth from AI? We could — we probably could well,” he said. “But I don’t think it’s going to do what we need to do to zero-off fiscal problem.”
他还质疑了对人工智能的大规模投资最终是否能产生足够的回报。“它会成功吗？还是不会？我不知道，但其他人也不知道，”鲁宾说。“其中一些结果会非常棒，一些公司会表现出色，而另一大批公司则会表现糟糕。”鲁宾表示，他的担忧不仅限于市场和经济，还涉及日益强大的人工智能系统可能带来的潜在安全风险。
He also questioned whether the massive investment in AI will ultimately generate sufficient returns. “Will it pay off, will it not? I don’t know, but neither does anybody else,” Rubin said. “Some of this will turn out to be terrific and some companies will do great, and a bunch of other companies are going to do badly.”Rubin said his concerns extend beyond markets and the economy to the potential safety risks posed by increasingly powerful AI systems.
他表示，他将人工智能安全和气候变化视为“两个巨大的生存风险”，与人类此前面临过的任何挑战都不同。鲁宾还指出，既要尝试为人工智能设立 safeguards，又要关注与中国的竞争，这是一项双重挑战。他提到，如果北京方面不实施类似的限制，美国可能会落后。
He said he sees AI safety and climate change as “two massive existential risks” unlike anything humanity has previously faced. Rubin also pointed to the dual challenge of attempting to put safeguards on AI but also keeping an eye on competition with China. If Beijing doesn’t impose comparable constraints, the US could be left behind, he noted.
鲁宾还表示：“美国仍然是最佳投资地。”但他认为，要维持这一优势，取决于美国的政治体系能否变得足够有效，以应对其财政、技术及其他长期挑战。
“The US is still the best place to invest,” Rubin also said. But he argued that maintaining that advantage depends on the country’s political system becoming effective enough to address its fiscal, technological and other long-term challenges.
鲁宾说：“我们的政治体系必须变得足够有效——而目前它尚未做到。它不必完美无缺，但需要足够有效，以应对我们所面临的问题。”
“Our political system has to become effective enough, again, — which it’s not now,” Rubin said. “It doesn’t have to be terrific. It wants to be effective enough to deal with our issues.”