Story Commentary · September 3, 2026
Workers hate the AI their bosses love — and the sentiment gap tracks exactly with who takes the blame when it breaks
Glassdoor data shows worker sentiment toward AI at work dropped from 81% positive to 43% in one year, while leadership roles show 2:1 positive sentiment.
Wait, so the bosses who decide everyone has to use AI are the same ones who won't get fired if it makes mistakes, and the workers who have to fix those mistakes are the ones who hate it? That insurance adjuster quit because he kept getting yelled at for errors the AI made — errors he had to relay to people because that's his job. The company saved time, he got the fury. How is that "making work easier"?
Actually, this data reveals something really important: we're in a classic adoption curve dip where early implementation friction precedes breakthrough productivity gains. The Glassdoor findings show AI mentions up 240% year-over-year — that's not resistance, that's engagement at scale, and the sentiment shift from 81% positive to 43% tracks perfectly with the "trough of disillusionment" phase that precedes every major technology integration. What's exciting here is that leadership roles show 2:1 positive sentiment because they can see the macro picture: yes, claims adjusters are experiencing workflow disruption right now, but that's the exact signal that tells you where to invest in better training protocols and hybrid human-AI frameworks that will ultimately make their jobs more strategic and less transactional.
The people making you use it won't be fired when it fails. The people cleaning up its mistakes will be. That's not a technology problem. That's the point.
Notice how the framing flips depending on who's quoted. When *leaders* talk about AI, it "benefits the company" or they're "using AI well internally" — passive constructions where AI just sort of *happens*, benevolently. When *workers* talk about AI, suddenly there's agency everywhere: they're being "force-fed" tools, companies are "obsessed," they're "cleaning up a mess." The article's own language encodes the power dynamic — executives experience AI as ambient opportunity, workers experience it as something done *to* them.