Authoritarian AI’s Last-Mile Problem

Counts differ because each database indexes a different set of publications. We treat OpenAlex as the canonical count; Google Scholar is not shown (no API, and crawling it violates its ToS).

Is artificial intelligence the autocrats’ silver bullet? No. AI-empowered autocrats are constrained by the “last-mile problem”: AI can help regimes to see, sort, and predict, but human police officers, soldiers, and local officials remain indispensable for physical coercion. In the long run, human agents will stay relevant: They not only guarantee an effective repressive system but also signal the autocrats’ popularity. For prodemocracy actors, this suggests that the conventional wisdom on combating autocracies, such as encouraging defections among security forces, still holds. The bottom line: Repression in the AI era remains a human affair.