Nobody here agrees on what AI can actually do for us
Most AI arguments inside an organization are literacy problems wearing a strategy costume. Once decision-makers know what these tools genuinely do well, where they fail quietly, and what to verify, the disagreement usually resolves itself.
Why does every AI conversation stall?
Because the people in the room are working from different mental models, and none of them are written down. One person has seen a demo and extrapolated; another has seen a hallucination and generalized; a third is quietly using tools nobody approved. Nobody is wrong exactly, and nobody can settle it.
The gap is rarely technical depth. It is knowing which questions these systems answer reliably, which they answer confidently and wrongly, and what a sensible person checks before acting on an answer. That is teachable in days, and it is what turns an argument into a decision.
What usually goes wrong with AI training?
- Training aims at tools and prompts, so it expires the moment the tools change.
- The loudest opinion sets the direction, because there is no shared basis to argue from.
- Leaders delegate the judgment to whoever seems most technical, which is not the same as most accountable.
- Confidence in these tools and the habit of checking them pull in opposite directions — in a 2025 survey of 319 knowledge workers, Microsoft and Carnegie Mellon found that the more confidence people reported, the less critical thinking they reported applying.
How do we get a leadership team aligned on AI?
Literacy for deciders
What models do, what they cannot do, and how to read an answer critically — pitched at people who decide, not people who build.
The critical-thinking guardrail
The habit of verification, taught explicitly — because people who report more confidence in these tools also report applying less critical thinking.
A shared vocabulary
When a leadership team uses the same words for the same things, the strategy conversation gets much shorter.
Sound familiar?
Start with an assessment — people, process and technology, looked at together.