本文检视推动加速 AI 采用的政治压力，并将其置于大众对这项技术扩散速度日益不安的背景下。Sam Altman 曾抱怨 AI 扩散感觉太慢，而英国部长如 Rachel Reeves 也呼应在 G7 中实现最快 AI 采用的呼声，但作者认为，这种说法与选民情绪以及许多人对 AI 所联想到的风险相冲突。文中引用的一项美国民调发现，只有 6% 的登记选民认为 AI 前进得太慢，30% 认为速度刚好，60% 则认为它前进得太快。
支持加速的理由是，更快的扩散可能提升生产力，并帮助像英国这样的经济体复苏，然而文章指出，经济层面的情况更为暧昧，因为社会对技术变迁的适应并不均衡。历史证据显示，长期调整可能幅度很大，例如英国劳动力中从事农业与采矿的比重，已从 1920 年的超过 14% 降到 2016 年的将近 1%。但突如其来的转变也可能造成持久伤害，MIT 教授 David Autor 对中国冲击的研究，以及 Eduardo Levy Yeyati 的论点都显示，即使技术最终到达同一终点，若采用速度快于劳动市场重新配置，就可能造成永久性的社会损害。
文章也强调，公众的担忧不只限于工作：Anthropic 对 159 个国家共 80,000 名 Claude 使用者的调查发现，人们担心自主性、认知退化、错误资讯、隐私、福祉与依赖。作者援引 Karl Polanyi 认为，快速变化只有在不是无方向的、且社会能够承受其速度时才是可接受的。更好的安全网与主动的技能政策可以提升这种承受能力，但若缺乏这些，许多人会觉得自己身处一辆超速行驶、没有安全带也无法控制的车里，这也有助于解释，为什么尽管存在潜在的经济收益，他们仍会抗拒加速 AI 采用的呼声。
The article examines the political push to accelerate AI adoption, set against growing public unease about how fast the technology is spreading. Sam Altman has complained that AI diffusion feels slow, and UK ministers such as Rachel Reeves have echoed calls for the fastest AI adoption in the G7, but the author argues this rhetoric clashes with voter sentiment and with the risks many people associate with AI. A US poll cited in the piece found that only 6% of registered voters thought AI was moving too slowly, while 30% said the pace was about right and 60% believed it was moving too quickly.
The case for speed is that faster diffusion could lift productivity and help economies such as the UK recover, yet the article says the economics are more ambiguous because societies adapt unevenly to technological change. Historical evidence shows long-run adjustment can be large, such as the UK workforce share in agriculture and mining falling from more than 14% in 1920 to almost 1% by 2016. But abrupt transitions can inflict lasting harm, as shown by MIT professor David Autor’s work on the China shock and by Eduardo Levy Yeyati’s argument that even with the same technological end point, an adoption pace that outruns labor-market reallocation can create permanent social damage.
The piece also stresses that public concerns extend beyond jobs: Anthropic’s survey of 80,000 Claude users across 159 countries found worries about autonomy, cognitive atrophy, misinformation, privacy, wellbeing, and dependency. The author invokes Karl Polanyi to argue that rapid change is acceptable only when it is not undirected and when society can cope with the pace. Better safety nets and proactive skills policies could improve that capacity, but without them many people feel they are in a speeding car with no seatbelt and no control, which helps explain why they resist calls to accelerate AI adoption despite the potential economic gains.