OpenAI's own data: using AI more doesn't make companies earn more
Uprovd Take When OpenAI's own numbers show heavy AI use doesn't lift revenue per employee, it proves Uprovd's point: measure outcomes, not tokens sent.
Read the original at FortuneFortune reports that OpenAI quietly folded a striking result into a 69-page study of how enterprises use ChatGPT, released this week. Across the companies it examined, there was no meaningful statistical link between how heavily employees leaned on AI - counted in messages and tokens - and the revenue each employee generated.
The report’s headline story is a widening gap between adopters and holdouts. But the revenue-per-employee finding, buried deeper, cuts the other way: pouring more prompts into the tools does not, on its own, show up on the income statement.
This is the distinction Uprovd is built on. Message counts and token totals measure activity, not return, and treating them as a stand-in for value is exactly how companies end up busy and no richer. The only honest test is whether an AI initiative moved a business outcome against its baseline - and that is what we measure.
This is Uprovd's analysis of third-party reporting. Original article linked above.