The Confidence Gap
-
770
professional surveyed in healthcare, finance, and manufacturing.
-
6 of 10
governance measures show a significant confidence gap.
-
+23 pts
widest gap, on whether AI training actually changed behavior.
-
28%
independently verified their AI vendor’s safety claims.
Leadership believes AI oversight is working. The people actually using it disagree — by up to 30 points.
We asked 770 professionals — split between individual contributors and managers & executives — the same ten questions about AI oversight at their organization. On six of them, the two groups gave meaningfully different answers, and the pattern wasn't random.
The gaps cluster around one theme: whether something can be proven, not whether it exists on paper. Could the organization produce documentation of human review? Could someone explain how the AI reached its output? Did training actually change behavior — not just get completed?
Managers said yes at far higher rates than the people doing the work. On every one of six governance measures, leadership is more confident than the people doing the work. Full breakdown, including healthcare, finance, and manufacturing sector detail, in the report.
What’s inside the report.
-
The Seniority Test
Is the gap really about hierarchy, or just familiarity with AI jargon? We tested it directly.
-
Does Experience Help?
Confidence rises with years of AI use — except for one surprising group.
-
Healthcare, In Full
The complete ten-statement breakdown for clinical AI oversight.
-
Finance & Insurance, In Full
Where compliance-driven adoption creates its own blind spot.
-
Manufacturing & Operations, In Full
The widest single gap in the entire survey lives here.
-
Who Owns AI Oversight?
It's not who you'd expect — and 1 in 3 organizations can't say for sure.
-
What Closes the Gap
Four concrete moves, backed by the data, not generic advice.
-
Get the Full Report
The full report includes the complete ten-statement breakdown, sector-by-sector deep dives for healthcare, finance, and manufacturing, the test for whether the gap is really about seniority or just AI-jargon familiarity, and what actually closes it. Free. No strings beyond your email.
-
Who's Most Confident? Who's Least?
The gap isn't uniform. When we combined sector and seniority into a single ranking, the spread between the most and least confident professionals in the entire survey was larger than any single factor on its own — it took both dimensions together to find the real extremes. The most confident segment and the least confident segment differ by more than half a point on a 5-point scale. See where every sector and seniority level falls in the full report.
-
This Isn't a Morale Problem — It's an Audit Exposure Problem
If your managers are confident your organization could produce documentation of human review, that confidence needs to be tested against an actual case — not assumed from the policy binder. The gap this report documents is exactly the kind of blind spot a regulator, an auditor, or an incident review will find first.