o3 is Smarter Than My Friends
And my friends are pretty smart
The Bluesky tweet above struck me this morning - the way I’ve been using AI has been changing even over the last several weeks.
OpenAI’s recent model, o3, is the first model that I find myself constantly using for advice. I mean constantly. From vetting a business decision, to deciding on a moving company, to deciding where to travel over the next holiday break, it consistently delivers advice that's as good as (or better than) the sharpest minds I know on that given topic.
Normally, accessing this level of expertise means scheduling coffee chats or Zoom calls days in advance. And that’s assuming I know a relevant person to begin with. And if I push too many follow-up questions, the goodwill balance gets dicey (though I’m always happy to reciprocate).
o3 doesn’t have those constraints. It’s always on, immediately reachable, and endlessly patient.
This model is built for reasoning, which means there’s a brief, strangely satisfying, pause of about 10 seconds before it responds. The pause gives the impression of a thoughtful person pondering before dispensing their wisdom.
The combo of consistently good, immediately available advice on practically any topic has profound downstream implications. Right now, it feels like most people haven’t quite grasped the power of using AI this way, but I think they will soon.
The value of social networks (real ones, not facebook dumpster fires) isn’t going away, but it’s going to change. Especially, for the cases where you don’t “know a guy”.
