Synthèse
Distrust in AI outputs isn’t a fringe concern; it’s becoming a shared consensus across teams trying to separate useful intelligence from confident noise. Kurtzman adds, “If you point a large language model at nothing, or at very shallow data, you mostly get noise. It’s garbage in, garbage out.” The State of Teams research quantifies this trust gap: only 22% of knowledge workers fully trust AI’s accuracy, and 69% say their data and knowledge foundations are not optimized for AI. Just like people
Distrust in AI outputs isn’t a fringe concern; it’s becoming a shared consensus across teams trying to separate useful intelligence from confident noise. Kurtzman adds, “If you point a large language model at nothing, or at very shallow data, you mostly get noise. It’s garbage in, garbage out. ” The State of Teams research quantifies this trust gap: only 22% of knowledge workers fully trust AI’s accuracy, and 69% say their data and knowledge foundations are not optimized for AI. Just like people need the right context to do their jobs well, AI needs the same foundation to produce work you can trust.
Without it, teams are more likely to get hallucinations, incomplete answers, or inaccurate information dressed up with confidence. “Where we see the value is when the AI has real context to work with: the history of decisions, the goals you are driving toward, the project outlines, and the teams actually building for the company, all in a place the AI can safely access. Once you have that, the quality of the intelligence gets much higher,” says Kurtzman.