AIエージェントの監視で労働者が認知限界に達し「AI脳疲労」の警告が発せられる

A BCG study of nearly 1,500 workers shows that simultaneously overseeing too many AI tools triggers cognitive exhaustion. The consequences are measurable, from higher error rates to increased intent to quit.

AI is supposed to make work easier, but a new study by Boston Consulting Group paints a more nuanced picture. The researchers surveyed 1,488 full-time U.S. workers about their AI usage patterns and the cognitive effects that come with them. The results are sobering. Intensive oversight of AI tools leads to a phenomenon the authors call "AI brain fry," defined as mental fatigue from excessive use or oversight of AI tools beyond one's cognitive capacity.

Those affected describe a "buzzing" feeling, mental fog, slower decision-making, and headaches. One senior engineering manager put it this way, according to the study: "I was working harder to manage the tools than to actually solve the problem."

Three simultaneous AI tools is the limit

The study identifies direct oversight of AI agents as the strongest driver of cognitive strain. Workers with high oversight demands expended 14 percent more mental effort, reported 12 percent more mental fatigue, and experienced 19 percent greater information overload.

The productivity curve is particularly revealing. Going from one to two simultaneous AI tools, perceived productivity rises significantly. At three tools, it peaks. Beyond four, it drops again. Multitasking with AI agents appears to hit the same cognitive ceiling as conventional multitasking.

Fourteen percent of AI-using respondents reported experiencing "AI brain fry." In marketing departments, the figure was 26 percent, while in legal departments it was just 6 percent.

Measurable costs for businesses

The business consequences are significant, according to the study. Those affected report 33 percent more decision fatigue and 39 percent more major errors. Intent to quit rose from 25 to 34 percent. Companies that use AI adoption as a performance metric could amplify this dynamic further, according to the researchers.

At the same time, the study shows that AI use is not inherently harmful. Those who use AI specifically to replace repetitive tasks report 15 percent lower burnout scores. The crucial difference lies in the type of strain. Burnout is emotional exhaustion, while "AI brain fry" is cognitive. AI can alleviate the former and trigger the latter.

The researchers recommend setting upper limits on the simultaneous use of AI agents and decoupling productivity metrics from the intensity of AI usage. Workers whose organizations value work-life balance showed 28 percent lower fatigue scores, according to the study.

All data is based on self-reports, and whether "AI brain fry" is a lasting phenomenon or reflects a transitional phase of adapting to new tools remains an open question. The study also comes from BCG, a consulting firm that itself advises on AI transformation projects.

AI News Without the Hype – Curated by Humans

Subscribe to THE DECODER for ad-free reading, a weekly AI newsletter, our exclusive "AI Radar" frontier report six times a year, full archive access, and access to our comment section.

Subscribe now