文章接著指出，即使是「有利劳工」的 AI 课税设计，通常都想区分「补强人力」的 AI 与「取代人力」的 AI。但在实务上几乎做不到，因为大多数工作是由一组任务组成：自动化其中一项任务会改变整个职务，并可能产生其他任务需求。考虑到未来技能需求高度不确定，政策无法精准引导 AI 的采用。文章以19世纪电气化作比较，若美国在 AI 竞争中落后于他国，等同于在电气化初期刻意拖延，最终会削弱美国的竞争力。
证据显示企业 AI 采用在加速上升。政策不应抵制或扭转 AI 投资，而应协助劳工调适并分享收益：放宽职业执照规范、改革失业保险以免企业因聘用不确定人员而受罚、补助训练与再技能化、扩大可得性薪资税额抵免、并进一步将税基转向消费。这些以劳工为中心的改革本就有价值；若 AI 真的引发剧烈工业转型，则更显急迫。
The article argues AI is among the most transformative technologies ever, and although its macroeconomic trajectory is hard to predict, it could radically change millions of jobs and eventually eliminate most occupations. Proposals to tax AI services or investment aim to slow worker displacement, but the argument is that restraining AI adoption would undermine growth and living standards rather than protect workers.
The article then says that even “pro-worker” AI tax designs usually try to distinguish AI that augments labor from AI that automates labor away. In practice, this is almost impossible because most jobs are bundles of tasks: automating one task reshapes the whole role and can raise demand for other tasks. Since future skill demand is highly uncertain, policy cannot direct adoption precisely. By analogy, trying to delay AI while rivals advance is akin to stalling electrification in its infancy and would weaken U.S. competitiveness. (Key numbers: 19)
Evidence shows corporate AI adoption is surging. Instead of resisting or redirecting AI investment, policy should help workers adjust and share the gains: relax occupational licensing, reform unemployment insurance so hiring workers with uncertain tenure is not penalized, subsidize training and reskilling, expand the Earned Income Tax Credit, and shift the tax base more toward consumption. These worker-centered reforms are worthwhile anyway, and the urgency increases if AI represents a sharp industrial transition.