One of the more striking findings in this year's GRIT data is that the people driving AI adoption in insights are also the least confident it is being managed well.

Less than half of brand-side researchers (44%) and analytics professionals (42%) say they're confident their organization minimizes the risks of AI misuse. Part of the reason may be that most of them aren't in the room where governance decisions get made.

AI governance tends to get treated as a legal or procurement exercise, something that happens upstream of research teams. The data suggests that's exactly the problem.

Confidence rises sharply when expectations are clear. Among brand-side researchers who said their organization's expectations around AI use were completely or mostly clear, 68% were confident risks were being minimized. Among those who found expectations unclear, that dropped to 32%. The same pattern held across analytics teams and supplier segments. Formal policies help, but writing rules down without communicating them isn't enough. People need to understand what AI is allowed to do, what it shouldn't do, and how outputs get validated.

That matters because AI in research is no longer just summarizing documents or drafting emails. It's in the analysis workflows that shape business decisions.

Trust in market research has always come from traceability. Researchers can inspect a crosstab, review weighting logic, check significance testing, and challenge an interpretation. AI changes the speed of work dramatically, but it doesn't remove the need for that chain of evidence. When insights teams lose visibility into how AI is being used, or governed, that chain becomes harder to defend.

The performance data makes the stakes concrete. Among brand-side researchers confident in how AI risks were being managed, 57% exceeded their goals, compared with 34% among those who were less confident. The least confident analytics professionals were more than three times as likely to fall short.

The future of AI in insights won't be determined by who adopts it fastest. It'll be determined by who can scale it while keeping the answers defensible. And that requires insights professionals to have a say in how AI is governed, not just how it's used.