Something strange is happening inside companies that have gone all in on AI. It is not the employees worried about being replaced who are struggling most. It is the ones every retention bonus was built around.
New survey data shows moderate or worse burnout among tech workers climbed to 56 percent this year, up from 45 percent the year before. Across all industries, roughly two out of three workers say they have felt burned out. Yet only about one in five people are actually worried AI will take their job.
The thing wearing people down is not fear of losing a job. It is the pace. Nearly all heavy AI users report their job has grown bigger than what they originally signed up for, not smaller.
That matches separate research from Atlassian, which found responsibilities expanding for most knowledge workers over the past year, with the biggest jumps among people using AI the most. AI was supposed to remove work. Instead it keeps adding to it, because once a task gets faster, leaders fill the extra time with more tasks rather than less.
Part of the pressure starts at the very top. Recent polling of hundreds of CEOs found that roughly three out of four believe they could lose their own job within two years if they fail to show a return on AI spending. That fear travels downward fast.
Boards push CEOs, CEOs push managers, and managers end up handing junior staff leaderboards that count how much AI output someone produces instead of how good the work actually is. Workers call this chasing token counts, and it rewards motion over value.
The cost of that setup is measurable. Research from BCG found that workers pushed into heavy, unsupported AI use show sharply higher decision fatigue and make more errors. They are also far more likely to say they want to quit.
Executives rarely see this cost because they only touch AI at the surface, for a quick draft or a rough prototype. They miss the cleanup work that someone else has to do afterward, and the training that stops happening because a manager fixed the junior's mistake instead of teaching them.
There is a clear counterexample, though. Gensler's workplace research found that heavy AI users who get autonomy and real time to learn new tools report stronger relationships with coworkers, not weaker ones. Same technology, completely different outcome.
The variable is not the tool. It is whether leadership sets a pace a human being can actually sustain. That is a choice, not a side effect of the technology.
For any company watching its best people quietly disengage, the fix is not a wellness program. It is rewriting what gets measured and rewarded. Track outcomes instead of AI usage, and put someone senior on the job of finding out what building with AI actually costs the people doing it every day.
Companies that skip this step will keep losing exactly the employees they can least afford to lose.