Everybody knows, nobody says it
Much of my work over the past year and a half has been about AI, especially how leadership teams make sense of all the noise.
My work with leadership teams has long been based on the concepts by Patrick Lencioni. Things like The Five Dysfunctions of a Team and Working Genius. I still genuinely think those concepts and the thoughts behind them are the best that the market can offer in helping leadership teams get more functional.
Helping leadership teams with AI lately has been both fulfilling and worrying. What do I mean by worrying? Let me explain.
The very cornerstones of a healthy leadership team (trust that allows everybody to speak their mind, constructive debate, and peer-to-peer accountability) have all been challenged by the topic of AI. In so many companies, leadership choices and AI policies have brought an elephant right into the middle of the boardroom. Honestly, I have not often seen execs this unsure of themselves, and letting things happen in the room that they would normally stop.
Let me give you a concrete example that represents one of the most typical cases I see almost every day:
Medium-sized company X has been using Microsoft tools for years and made a seemingly logical decision to standardize on one AI tool for the whole company, Microsoft Copilot, with a policy that bans the use of any other AI tool for work use.
The only person in the company who was even remotely happy with this decision was the CIO (or the CISO if that role existed). Why? Because it made the CIO’s life easier, and to be fair, he had a real problem to solve. One tool means one place where company data sits, one contract, one set of controls. Everyone else’s life just got harder. Prior to the Copilot decision, everybody had been using ChatGPT, Claude, Gemini, Perplexity, etc. Their first experience with Copilot did not go well. Most went right back to the tools they had been using, and used Copilot only to make their work look like it follows the policy.
Everyone uses the tool that makes them most productive, myself included. Some are advanced enough to do that while keeping information security risk to a minimum, but many are not. When people have to choose between the policy and getting the work done, the policy loses every time.
And here is the problem. Everybody in that leadership team knows this is happening. They know it about themselves, and they know it about each other. Nobody says it out loud in the meeting. That is the first and fundamental dysfunction of any leadership team: absence of trust that leads to avoidance of difficult conversations. The real risk here is a leadership team getting into the habit of knowing about a problem and saying nothing about it. That is a habit that can sink companies.
I wrote about the policy side of this, and what works better, in Why Copilot-only AI approaches backfire. This article first appeared in my LinkedIn newsletter, Team Direct.