AI Doesn't Have to Be Stupid to Do Stupid Things

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).

PREPRINT — Not peer reviewed. Approved author version 0.2, released October 5, 2026. This essay applies Carlo M. Cipolla’s human gain-and-loss framework to AI-mediated institutional decisions. It distinguishes computational capability from human judgment and examines how mechanistically coherent AI outputs can cause losses to patients, workers, institutions and society. Illustrative scenarios are hypothetical. Submitted to The New Republic October 4, 2026; editorial acceptance and publication have not been established.