AI Use at Work Causing ‘Brain Fry’ in 14% of Workers, Study Finds

Artificial intelligence tools designed to boost workplace productivity are instead causing mental fatigue and cognitive strain in a significant portion of workers, according to new research from Boston Consulting Group and the University of California.

A study of nearly 1,500 full-time US workers found that 14% reported experiencing “AI brain fry,” defined as mental fatigue resulting from excessive use of, interaction with, or oversight of AI tools beyond one’s cognitive capacity. The findings were published in the Harvard Business Review on Friday.

Respondents described symptoms including a “mental hangover” accompanied by a “fog” or “buzzing” sensation, along with an inability to think clearly. Workers also reported headaches, slower decision-making capabilities, and difficulty maintaining focus on tasks.

The research challenges AI companies’ marketing messages that position their products as productivity boosters capable of allowing workers to offload portions of their workloads. Some companies have embraced this narrative to the extent of measuring AI usage as a performance metric.

Cryptocurrency exchange Coinbase provides a notable example, with CEO Brian Armstrong stating he fired engineers who refused to adopt AI tools. Armstrong set a goal late last year to have AI generate half of the platform’s code.

“As enterprises use more multi-agent systems, employees find themselves toggling between more tools,” the researchers wrote. “Contrary to the promise of having more time to focus on meaningful work, juggling and multitasking can become the definitive features of working with AI.”

Marketing and human resources workers reported the highest levels of AI-induced brain fry among all professional categories surveyed, according to the study’s findings.

The mental strain carries high organizational costs, including increased employee errors, decision fatigue, and higher turnover intentions. Study respondents experiencing AI brain fry reported 33% more decision fatigue compared to those without such symptoms, which researchers estimated could cost large companies millions of dollars annually.

Workers suffering from AI brain fry were approximately 40% more likely to have active intentions to quit their positions. They also self-reported making nearly 40% more major errors than colleagues without brain fry symptoms. The study defined major errors as those with serious consequences affecting safety, outcomes, or important decisions.

However, the research identified positive outcomes when AI tools were deployed for specific purposes. Workers who used AI to reduce time spent on routine and repetitive tasks reported burnout levels 15% lower than those who didn’t utilize AI in such ways.

Burnout, characterized as a state of chronic workplace stress leading to negative feelings about one’s job and decreased effectiveness, appeared to decrease when AI successfully eliminated mundane work.

The researchers offered recommendations for company leaders seeking to reduce AI brain fry among their workforce. Organizations should “clearly define AI’s purpose” and explain how workloads will change with the introduction of these tools.

Companies should also focus on “measurable outcomes” rather than usage metrics. “Incentivizing quantity of use will lead to waste, low-quality work, and unnecessary mental strain,” the researchers cautioned.

The findings suggest that while AI tools hold promise for reducing certain types of workplace stress, their implementation requires careful consideration of cognitive load and clear strategic alignment. It also indicates that the technology’s benefits materialize primarily when AI replaces genuinely repetitive tasks rather than adding another layer of tools requiring management.

As artificial intelligence becomes increasingly integrated into workplace environments across industries, including the cryptocurrency sector, the research highlights the importance of thoughtful deployment strategies that account for human cognitive limitations rather than simply maximizing tool adoption rates.