Why is Meta destroying its engineering organization?

Pressing engineers to show heavy AI usage can reward activity instead of better engineering. Gergely Orosz makes that argument in an essay about Meta's engineering organization. He reports that managers at Meta would inspect AI token counts (the units of text a model processes) during performance reviews, and that engineers responded by using AI tools to generate more tokens.

The essay ties this to an existing review culture in which, in Orosz's account, engineers learn to keep every measured number above their peers'. A new metric then becomes one more number to push up. The claim rests on Orosz's conversations with Meta engineers and figures he cites from press reports. It is an argument about how management metrics shape behavior, not a measurement of AI's effect on productivity.

“You have zero purpose in life all of a sudden, you barely interact with anyone, you just have these tasks every week.”

“It is like what the fuck.”

Chris Cox