OpenAI Studied How Real Workers Actually Use AI — It's Not Just Chatting

OpenAI's economic research team has published new findings on how people actually use AI at work — and the picture is broader than the usual story about coders and copywriters. Instead of only doing the tasks AI was "designed" for, workers are bending these tools to fit their own jobs. Someone in HR uses it to rewrite a policy memo. Someone in sales uses it to summarize a messy call. A small-business owner uses it to answer customer emails at 11pm. The interesting part isn't the technology — it's that people are finding these uses themselves, without anyone handing them a manual.

The research also looks at which of these uses stick. Lots of AI experiments are one-and-done: you try it, shrug, and go back to your old way. But some tasks become recurring parts of a job — built into the weekly routine the way email or a spreadsheet is. Those repeated uses are the ones that actually change how work gets done and how much time a role takes. That matters for anyone wondering whether AI will touch their job, and for employers deciding where to spend training budgets.

For you, the practical question is simple: which parts of your week are repetitive, text-heavy, or involve sorting information? Those are the spots where AI habits tend to form fastest. You don't need technical skills. You need to notice the boring 20 minutes you repeat every day.

The honest caveat: this is OpenAI researching its own products, using its own data. It's a useful signal, not a neutral, independent audit. It also can't tell you whether the work people are doing with AI is better — just that they're doing it, and coming back. Treat it as evidence that the shift is real and already underway in ordinary offices, not as proof of how big the payoff is.