AI tools were supposed to make work easier. For many employees, they're creating a new kind of exhaustion instead.

A January 2026 study of 1,488 full-time U.S. workers conducted by Boston Consulting Group put a name to what many workers have been feeling: “AI brain fry.” Defined as mental fatigue from excessive use or oversight of AI tools beyond one’s cognitive capacity, it’s distinct from burnout. It shows up as a mental fog, difficulty focusing, slower decision-making, and a kind of cognitive noise that makes careful judgment harder.

Fourteen percent of workers using AI in the study reported experiencing AI brain fry. Among HR and people operations professionals specifically, that figure rose to 19.3%.

‍

What BCG Researchers Found

The BCG researchers examined a broad range of AI engagement patterns: the number of tools used simultaneously, whether AI replaced or augmented tasks, the degree of oversight required, and whether AI had increased or decreased overall workload. Several findings stand out.

Oversight is the primary driver of fatigue. Workers who reported high rather than low degrees of AI oversight expended 14% more mental effort on the job, experienced 12% more mental fatigue, and reported 19% greater information overload. The act of monitoring, verifying, and second-guessing AI outputs is what depletes cognitive resources.

The tipping point is three tools. The study found that productivity increased as workers moved from one AI tool to two or three tools. After three tools used simultaneously, productivity scores dropped. This has direct implications for how HR tech stacks are designed and how employees are trained to use them.

AI brain fry is not the same as burnout. Burnout is an emotionally driven, chronic condition. AI brain fry is an acute cognitive strain caused by marshaling attention, working memory, and executive control beyond their limits. The study found that using AI to replace repetitive tasks predicted a 15% decrease in burnout scores. The same was not true for mental fatigue, which was driven by oversight intensity, not task volume.

The business costs are significant. Workers who experienced AI brain fry reported 33% more decision fatigue, made major errors 39% more frequently, and were 39% more likely to show active intent to leave. The researchers noted that the workers most vulnerable to AI brain fry are often the organization’s heaviest AI users.

‍

Why HR Is Vulnerable

Several features of HR work create compounding risk.

• High-stakes outputs. HR decisions affect real people’s careers, compensation, and development. There is less margin for error, which means more verification of AI outputs and more cognitive overhead with each review.
• Ambiguous quality signals. A subtly off-tone AI summary or a slightly biased candidate assessment may not be obviously wrong. Detecting it requires judgment, not just attention.
• Volume amplification. AI tools often increase the volume of outputs HR teams must process without proportionally reducing the review burden. As the BCG study notes, when AI increases perceived workload, mental fatigue scores rise in tandem.

‍

What Helps: Manager, Team, and Org-Level Practices

AI brain fry is not purely an individual problem. It is shaped significantly by the environment in which people work. Their findings point to concrete levers at multiple levels.

At the manager level

Workers whose managers took time to answer questions about AI had 15% lower mental fatigue scores than those whose managers did not. By contrast, employees left to figure out AI on their own showed 5% higher fatigue. Manager engagement with AI questions has a measurable protective effect.

At the team level

Teams where technology is treated as a shared capability rather than an individual differentiator showed less mental strain. By contrast, team pressure to use AI and wide variation in AI use across team members were both associated with higher fatigue. The way a team collectively relates to AI tools matters as much as how any individual uses them.

At the organizational level

Two organizational signals proved especially potent. When employees felt their organization expected them to accomplish more work because of AI, mental fatigue scores were 12% higher. When employees felt their organization genuinely valued work-life balance, mental fatigue scores were 28% lower.

‍

What HR Leaders Can Do Right Now

The difference between AI that depletes people and AI that supports them is how thoughtfully it’s implemented. Several practical steps follow from the research.

1. Audit your AI oversight burden

Map the AI tools your HR team uses and estimate how much time is spent reviewing, correcting, or second-guessing outputs for each one. This audit often reveals that a handful of tools are generating disproportionate cognitive load. The BCG data suggests that stacking more than three tools at once is where diminishing returns become negative returns.

2. Prioritize tools designed for low oversight

When evaluating AI tools, weight output accuracy and interpretability as heavily as feature sets. Support focused work rather than demanding constant monitoring. Tools that produce immediately clear, actionable outputs with minimal verification overhead reduce the oversight burden that drives fatigue.

3. Deploy AI where it reduces toil, not where it creates surveillance

The BCG study found that using AI to replace repetitive tasks predicted a 15% decrease in burnout. The key is identifying tasks that are genuinely routine and unenjoyable, not redirecting AI toward tasks that require human judgment and then monitoring the results.

4. Build skills for managing AI workload

The BCG researchers identified problem framing, analysis planning, and strategic prioritization as critical skills for workers in AI-intensive roles. Building this capability is an emerging priority for L&D functions and a concrete one: scenario-based practice, structured reflection on AI-assisted decisions, and explicit coaching on when not to use AI are all delivery mechanisms that exist today. Workers who develop these skills are better positioned to decide when to use AI, when to stop iterating, and when to trust their own judgment.

5. Communicate clearly about workload expectations

The BCG study found that when employees expected AI to intensify their workload, fatigue scores rose. When they felt the organization valued their wellbeing, fatigue scores fell sharply. HR is uniquely positioned to ensure that organizational messaging about AI efficiency is matched by explicit signals that cognitive health is also a priority.

‍

What Good AI Design Looks Like

The BCG researchers closed with a specific call to action for tool designers:
Build AI that supports cognitive thriving rather than demanding cognitive sacrifice.

For HR leaders evaluating tools against this standard, the gap between principle and practice is often where decisions get made.

This principle shapes how Colleva approaches AI-powered training and employee insights. Sessions are designed to deliver personalized, substantive feedback that reaches learners immediately, without requiring HR or L&D teams to review, edit, or verify it first. Human oversight is preserved for meaningful decisions.

To see how Colleva puts this into practice, including how session feedback reaches learners without HR or L&D review, visit colleva.com.

Source

Bedard, Julie, Matthew Kropp, Megan Hsu, Olivia T. Karaman, Jason Hawes, and Gabriella Rosen Kellerman. “When Using AI Leads to ‘Brain Fry.’” Harvard Business Review, March 5, 2026. https://hbr.org/2026/03/when-using-ai-leads-to-brain-fry