Meta just stopped scoring engineers on AI usage.

Meta told staff it will not use AI adoption dashboards or token counts to evaluate impact. Managers should judge the quality and speed of the output, along with problem complexity and the scope somebody takes on.

Engineers had been competing on a token leaderboard. Then the company found people were inventing extra work to climb it. Amazon deleted its own leaderboard after the same thing happened.

If your board report counts licences activated and prompts sent, it relies on measures Meta just stopped trusting.

What would you put in front of the board instead?