AI made your team faster and trusted themselves less: the NNG number is 3.57 out of 5

Your team shipped faster this year. The numbers probably say so. But walk the floor and the mood does not match the dashboard. People are using AI, saving time, and feeling worse about it. That gap is the real story of this cycle, and it is sitting in your next performance review whether you planned for it or not. Let me catch you up.

The deep cut

• Speed is not the same as morale. NNG found AI saves time on nearly every task, yet UX sentiment landed at 3.57 out of 5.
• You cannot grade work you cannot see inside. Over half of managers in Monday.com's survey say polished AI output makes evaluation harder.
• A tool for freedom becomes a tool for control. The Co-op's OpenAI system scores 50-plus aspects of every call and watches workers for hours.

The dashboard says yes, the room says maybe

NNG surveyed 521 UX practitioners and the time savings are real. AI made every task faster, rated 4.2 out of 5 for speed. Almost everyone has access. So you would expect people to be thrilled.

They are not. Sentiment came in at 3.57 out of 5, barely above neutral. And of those who left an open comment, 69% carried some negative feeling. Researchers were the sourest. The complaints were not about speed. They were about mandates from above, output they could not trust, and jobs they feared for.

One line from a respondent should stop you cold: "It's a strange feeling to be forced to use something that I do not believe in." Faster work, lower confidence. That is the trade your team is living right now.

The polished inbox problem

Here is the catch nobody budgeted for. When work shows up clean, you cannot tell who did it. A Monday.com survey of 1,700 managers found more than half now find it harder to evaluate their teams. Output arrives finished, so the line between person and machine blurs.

So managers do more reviewing, not less. And 78% changed how they judge people, weighing prompt skill and task delegation. Fair enough. But 39% are now drafting reviews with AI, which bakes bias into a process that was already shaky. Monday.com's Michal Miller Levi admitted companies have not figured out how to handle that.

The honest read: your evaluation system was built for a world where effort was visible. That world is gone, and no one has replaced it yet.

When the tool for freedom becomes the leash

This is the trap, and it is uglier than slow reviews. The same tech that could free people is being pointed back at them. The Co-op in the UK runs an OpenAI model that scores more than 50 aspects of every customer call and watches some workers for hours a day. One employee called it "oppressive and dystopian."

Craig Unsworth names the pattern well. Give an org a liberating tool and its "immune system" wraps it in approval steps and reporting until it looks like the old way with a bigger bill. He draws a clean line: accountability means explaining what you decided and why. Surveillance means saving every hesitation to use against someone later. Measure output and you can end up with a perfect record of meetings where nobody said anything interesting.

The part that is quietly eating your talent pipeline

Push hard enough and you lose the people you most want to keep. Elena Verna, who works at Lovable, watches the "996" badge of honor spread, 9 to 9, six days a week, and refuses it. Her point is sharp: you are borrowing from employees' future health to book output today, and that debt never shows on a balance sheet. Normalize it and parents, caregivers, and experienced operators quietly stop applying.

There is a better model than grind. Joko's CEO Xavier Starkloff scrapped the idea of a single "AI transformation" hire because it lets everyone else off the hook, and wrote AI into performance reviews instead. Every Thursday afternoon the company stops: no email, no Slack, no meetings. He puts the cost at 10% of payroll and defends it.

And Will Larson warns that the rules changed under managers who did everything right. His advice to middle managers is blunt: re-skill or retire. People feel betrayed, and he says they are not wrong, but the work shifted anyway.

Three questions for your team

• If we cannot tell what came from the person and what came from AI, what are we actually grading in the next review cycle, and does that measure outcomes or output?
• Where have we pointed AI at our own people as a watch-tool, and would we call that accountability or surveillance if a reporter asked?
• What boundary can a junior employee hold on this team without hurting their career, and can we name one, not just the execs?