You ask AI to help with a simple email.

The first draft is useful, but not quite right. So you ask it to make the tone more natural. Then you correct a few facts. Then one paragraph sounds too generic, so you ask it to try again. The next version is better in one place and worse in another.

Before you know it, you have spent far more time polishing the email than it probably deserved.

You are not doing the whole thing yourself. But you are not fully free of it either. You are reading, checking, nudging, editing and deciding whether the output is good enough to use.

Instead of feeling more productive, you feel oddly drained.

Welcome to AI brain fry, the hidden cost of AI-enabled productivity.

AI arrived with a utopian promise: increased efficiency, reduced drudgery and more headspace for creative, meaningful work. Yet across workplaces, the opposite feeling is emerging. Employees report mental exhaustion after hours of reviewing AI-generated outputs. Managers are navigating technology that evolves faster than humans can keep up. Teams are spending less time solving problems and more time monitoring systems. What was supposed to lighten workloads seems to be intensifying them.

If you feel a strange fatigue after wrestling with ChatGPT, Claude and Copilot, you’re not alone. New research from BCG shows that 14% of employees are now experiencing AI brain fry – a state of acute mental exhaustion marked by brain fog, a buzzing feeling, lack of focus and headaches. Participants describe this strain in different ways, such as:

“I had been back and forth with AI reframing ideas, synthesizing data, forming and organizing…I couldn’t even comprehend if what I had created even made sense.”

“Instead of moving faster, my brain just started to feel cluttered… I was working harder to manage the tools than to actually solve the problem.”

Researchers found that the excessive use or oversight of AI tools carries a sharp cognitive penalty. Employees with AI brain fry showed:

• 33% greater decision fatigue
• 11% more minor errors
• 39% more major errors
• 39% higher intent to quit

In business terms, that translates into millions lost through rework, subpar decision-making and attrition. Given that many of today’s most intensive AI users are rising talents and high performers, this cost is set to become even more pronounced.

And what about you, as an individual, now turning to your friendly AI chatbot for everything from writing emails and crafting visuals to analysing data and uncovering insights?

The implications of AI brain fry aren’t restricted to work performance – they’re far more profound. The way we use AI tools is fundamentally reshaping our ability to think.

Marking a quantum leap beyond calculators and search engines, generative AI has given us the option to outsource core cognitive tasks such as ideating, analysing, reasoning, synthesising, and summarising. This is an unprecedented moment in our collective cognitive history – and with emerging research, we now have a clearer idea of how this technology is changing our brains.

This week, let’s explore AI brain fry at the workplace. Why are AI tools increasing mental overload and compromising performance? And how can we protect our cognitive endurance – as individuals, teams and organisations?

AI brain fry is rooted in a simple, undeniable fact: our cognitive capacity is limited. Watching machines process vast amounts of data in seconds, it’s easy to forget that the human brain is wired very differently. We can only push our attention, working memory and executive control so far before they collapse.

The way AI is currently being deployed isn’t aligned with this biological reality. In too many organisations, the focus is on quantity above all. Employees are instructed to work with several AI tools and oversee swarms of agents. They’re assessed on how many prompts they generate or how many lines of code AI tools produce for them.

The BCG study highlights the fatigue of supervising these complex, high-volume systems – juggling, evaluating, tweaking, and validating. Because AI outputs can appear very convincing even when flawed, employees remain in a state of constant vigilance. The process demands relentless back-and-forth, with every context switch exacting a mental toll.

The issue isn’t AI itself. It’s the way we’re using it. For instance, when employees start working with two AI tools simultaneously, productivity jumps. Add a third, and you still see some gains. Beyond this, however, productivity dips sharply. Another example: using AI for repetitive tasks (without continuous oversight) can curb burnout by 15% and bolster engagement. As seen in both cases, it all comes down to how thoughtfully AI is embedded into workflows.

Research from UC Berkeley points to another paradox. In theory, AI promises more freedom. As the burden of routine tasks is offloaded, people should have more time for high-value tasks such as deep thought, strategy, and innovation. But in practice, AI is intensifying workloads rather than reducing them.

As production speeds up, the baseline expectation rises with it. Employees produce more outputs, stack more tasks, and multitask all day long – because the tools make it possible. Their scope of work also expands. Researchers jump into engineering. Product managers start coding. With AI, anything seems possible – until the initial high fades and the inevitable exhaustion creeps in.

Taken together, these findings shed light on a clear mismatch between technological efficiency and human mental endurance. The challenge isn’t whether we should use AI. It’s whether we can integrate it while protecting the cognitive capacities that make us genuinely effective.

For individuals: how to use AI without frying your brain

1. Keep one layer of thinking manual.

Don’t offload every stage of a task. If AI generates a first draft, rework and refine it yourself. Resist endless prompting. Move from perfecting your instructions to solving the problem.

2. Augment, don’t replace.

Use AI to enhance your mental capacity rather than bypass critical thinking. As Mark Cuban says, “There are generally 2 types of LLM users, those that use it to learn everything, and those that use it so they don’t have to learn anything.”

3. Batch AI usage.

Instead of keeping your chatbot open all day and constantly checking outputs, timebox AI-assisted tasks. This helps to lower the cognitive drain associated with oversight.

4. Carve out AI-free windows.

Set aside dedicated time for unassisted thinking. Go back to pen and paper. Settle into reflection. The best ideas and connections emerge when the brain is quiet, not cluttered.

5. Limit tool-switching.

Using two AI tools may be helpful. Using five can quickly become exhausting. Each additional tool creates another interface, another style of output, another set of comparisons and another decision to make. What feels like optionality can become clutter.

For leaders: how to deploy AI without frying your team’s brains

1. Reward outcomes, not activity.

Measure quality and impact instead of tracking how much AI employees use. Incentivising token volume is counterproductive, resulting in AI brain fry that leads to poor decisions and avoidable mistakes.

2. Redesign work for human + AI.

Don’t layer AI supervision endlessly onto already stretched teams. Define clear limits for AI oversight, just as you do for managing humans. Audit workflows for duplication and over-complexity. Curb the use of multiple AI tools. Build realistic timelines instead of assuming AI-generated work is usable as-is.

3. Clarify new boundaries.

Don’t use AI tools to quietly expand output expectations and role scope. Be explicit about evolving workloads. BCG researchers recommend devoting 70% of AI transformation efforts to people and processes.

4. Treat attention as a strategic resource.

The most valuable human capabilities depend on sustained focus. Track mental load as carefully as efficiency metrics. Carve out protected time for deep thinking and cognitive recovery.

5. Train employees to partner with AI.

Build skills in problem framing, prioritisation and critical evaluation. Clear thinking helps employees augment their capacity with AI, rather than multiplying work.

Framing AI as either a miracle or a menace misses the more important reality: it is already reshaping how we work, think, and pay attention. The challenge is whether we can leverage the potential of this powerful technology without compromising our irreplaceable cognitive capacity.

Instead of simply asking what machines can do, we must refocus the conversation on what humans need in order to work alongside them sustainably. Without thoughtful, evidence-backed norms for usage, we risk paying the steep price of AI brain fry.