Margin note · 22 November 2026 · Leading through change
Why do people like AI less the more they use it?
Translated from Italian. Read the original.
The fact
Indeed Hiring Lab and Glassdoor, 18 November. Among Glassdoor reviews (United States) that mention AI, 45% are positive in 2026, against 82% in 2019. Regular use of AI at work rises from 36% to 42%, while the time regular users say it saves falls from 2.3 to 2.1 hours a day.
Source: Indeed Hiring Lab & Glassdoor Economic Research, 18 novembre 2026
The easy reading is that people are afraid of losing their jobs. There is some truth in it: those who criticise AI in reviews mention layoffs 7.1 times as often as the base rate.
But another figure struck me more. More people use AI at work, yet those who use it say they save slightly less time than a year ago. And among those who criticise it, only 37% approve of their CEO; among those who speak well of it, 89%.
I suspect the problem is not so much the tool as the way it arrives. Kim and Mauborgne, studying hard decisions, suggested that people accept even an unfavourable outcome if the process was fair: engagement, explanation, clear expectations. And they reject even a good outcome when the process is not. A recent study published in HBR points the same way: people who feel forced to use AI tend to produce more low-quality work and think more often about leaving.
These are American data, gathered by a job-listings platform, and should be read with caution. Still, I think the question is worth asking here too: when we introduce AI, do we explain what we will do with the time it saves?
Perhaps, if we companies don't say it, people imagine it for themselves. And they usually imagine the worst.