Scott Shimp, VP of Solutions at Data Society, argues the line between trust and blind trust with AI comes down to whether the person looking at your work decides to discount it, not how comfortable you personally are with the tool. The data on how AI use actually gets perceived at work says that discount is real, and it’s steep.
Shimp’s point is that this isn’t a personal, internal question. “It’s much less about how you feel about AI and where you draw the line between trust and blind trust,” he says. “It’s much more about other people, and whether they trust you. The line between trust and blind trust with respect to AI is the line where someone else is going to look at what you’re putting together and discount it, because they’re not sure if you used AI, or you used it too much, or there’s just something about what you put together that seems suspect.”
Atlassian’s research on AI disclosure at work put actual numbers on it. In a controlled study of 961 U.S. knowledge workers, those who disclosed using AI on a task were rated ten times lazier than peers who did identical work without disclosing, and were 24 percentage points less likely to be recommended for high-visibility projects, even though the work itself was the same. That’s despite 94% of U.S. knowledge workers using AI at work, with roughly three in four vocal about it to managers and peers and closer to four in five entirely transparent about it once asked. The penalty fell on the people who admitted it, not on AI use itself, which is close to exactly the line Shimp is describing: the discount comes from the reader, not from the tool.
That same research found the penalty nearly disappears inside companies that treat AI use as normal. In those cultures, people who disclose AI use are rated as more efficient than people who don’t, without anyone changing how much AI they actually used. The discount is a function of whether the person judging your work already trusts how you used it, not a fixed cost of using the tool.
That fear of being discounted is also why so many people hide their AI use altogether. A Slingshot survey fielded by Dynata found 45% of employees conceal their AI use at work. Thirty-four percent said it’s because they worry it will look like they’re cutting corners, and 27% said they’re worried about being judged by colleagues. Employers guessed differently: 47% assumed workers were mainly hiding AI use out of job security fears, but only 24% of employees actually cited that reason. The bigger driver, by the survey’s own numbers, is exactly the suspicion Shimp names: not fear of being replaced, but fear of being discounted.
If the line runs through the reader instead of the person doing the work, then managing it becomes less a question of personal ethics, how much AI feels acceptable to use, and more a question of what the room you’re handing work to already believes about AI, and what you do to keep your work legible inside that belief. The same output can read as competent or suspect depending entirely on that context, which is why the Atlassian data shows the stigma evaporating in AI-positive teams without anyone changing how much AI they actually use.
If you manage a team, the more useful question isn’t whether someone used AI. It’s whether your team already treats that as normal enough that disclosing it doesn’t cost anyone anything. Data Society’s AI upskilling programs build that kind of shared norm across a team, and AI Advisory helps leadership set the policy that makes disclosure safe in the first place.
