文章主张，与其设计 AI 专属利润税或 robot taxes，不如改革既有资本利得税，因其直接指向 AI 繁荣累积收益的位置。在美国及多数国家，最富裕者的投资收益税率低于劳动所得；美国 AI 投资者出售股票获利时，税率约为同等薪资或奖金的一半，这一差距已存在近 40 年，虽曾由 Ronald Reagan 在 1980 年代缩小，但很快恢复。
另一项漏洞是继承时资本利得税的成本基础会重设为当前价格；Yale University Budget Lab 估计，把继承视为应税事件可在 10 年创造超过 1,000 亿美元收入。Wharton School 2024 年估算，将投资利润按普通所得税率课税并取消继承漏洞，30 年可带来 1.8 兆美元；在美国联邦债务超过 GDP 100% 且 AI 将造成再培训成本时，税制改革比政府直接持股更可行。
The White House proposal to place shares in large AI companies such as OpenAI into a kind of sovereign wealth fund recognizes that AI will create windfalls for some and harm others, and that the gains are concentrated in rising equity valuations; however, public ownership ideas advanced by Donald Trump, Bernie Sanders, and OpenAI face definitional problems over whether participation is mandatory or voluntary, what counts as an AI company, and whether Nvidia, Blackstone, or AI-using banks that lay off workers should qualify.
The article argues that instead of creating AI-specific profit taxes or robot taxes, governments should reform existing capital gains taxation, because it targets where AI boom benefits have accumulated. In the US and most countries, investment gains for the wealthiest are taxed below labor income; an American AI investor selling shares for profit pays roughly half the tax rate that would apply to the same amount earned as salary or bonus, a gap that has existed for nearly 40 years after Ronald Reagan briefly closed it in the 1980s.
A second loophole resets the capital gains tax basis to current market value when wealth is inherited; Yale University’s Budget Lab estimates that treating inheritance as a taxable event would raise more than $100bn over 10 years. Wharton School estimated in 2024 that taxing investment profits at ordinary income rates and removing the inheritance loophole could yield $1.8tn over 30 years; with US federal debt above 100 per cent of GDP and AI creating reskilling costs, tax reform is more practical than direct state ownership.