在2月最后一周股市震荡之际，Citrini Research的一份报告因描绘2028年AI使大量白领岗位和从American Express到DoorDash等公司业务过时而走红。其核心数字锚点是短期时间线（到2028年）与广泛行业冲击，但论证被批评为宏观经济机制薄弱。

该情景声称“幽灵GDP”会出现：由不休息的AI代理推动的产出继续增长，而就业与劳动收入下滑使消费同步萎缩。这个组合意味着实际经济中的支出循环断裂，即便国民账户中的产出数字上升，需求端却呈统计性空心化趋势。

文本将这一逻辑与19世纪初“普遍过剩”争论对照，指出萨伊定律下只能出现部门性过剩，而凯恩斯框架允许货币囤积导致全局需求不足。关键趋势判断是：若AI利润被囤积，结果更可能是通缩与GDP下降而非名义GDP激增；再加上企业仍在再投资数据中心且政策端有降息、购债、转移支付工具，AI引发“无消费经济”被认为概率较低。

As stocks wobbled in the last week of February, a Citrini Research note went viral by projecting that by 2028 AI could obsolete large swaths of white-collar work and business models from American Express to DoorDash. Its main numerical anchor is the short horizon to 2028 plus broad cross-sector disruption, but the macroeconomic logic is criticized as weak.

The scenario claims “Ghost GDP”: output keeps rising through always-on AI agents while employment and labor income fall enough to collapse consumer spending. That pairing implies a broken expenditure loop, so national-accounts output can rise statistically even as real-economy demand trends toward hollowing out.

The text maps this to the early-19th-century “general glut” debate, contrasting Say’s Law (sectoral gluts only) with Keynesian dynamics where cash hoarding can create economy-wide demand shortfalls. The key trend conclusion is that if AI profits are hoarded, deflation and falling GDP are more likely than a nominal-GDP surge; with firms still reinvesting in data centers and policymakers retaining rate cuts, bond buying, and transfers, an AI-driven no-spending economy is viewed as unlikely.

Source: A viral research note on AI gets its economics wrong

Subtitle: Too much of a good thing

Dateline: 2月 26, 2026 04:27 上午