Dwarkesh Patel and others discuss what AI can do in the next 18 months. They focus on how models will handle longer tasks, from coding to day-long work, and where they'll still hit walls like video editing and booking flights. The economic impact could be significant within three years.
like how long can an AI do a task? Is it like 7 seconds and it loses its train of thought? Is it 14 seconds? And it seems to be doubling every 7 months. So before, they can maybe write you a little function. Now they can write you a whole package of code. Now they can maybe write you many different lines of code and keep engaging with you over a session as you say, no, I want to add this feature or something. So if that keeps doubling, going from 30 minutes to an hour is a big deal. Going from an hour to 2 hours is a big deal. And going from a day to 2 days is an incredibly big deal, right? The amount of increased scope you can have as a result will increase exponentially. I am still confused confused about why these models have no traction whatsoever on things which are like short horizon tasks. Like, it doesn't take you that long to, let's say, edit a video or book a flight. So how's it coming?
Good. This one looks pretty exotic.
Or organize the logistics for like a happy hour or something, right? So the next question is like, okay, look, they've conclusively shown that this gets you a ton of traction on math and code. What are the other kinds of RL environments you can spin up where now you're doing tasks that are like a human's day-long worth of input? I think in 3 months we will know if, um, if— or do we have models which can reliably do computer use? If not, we're not going to have some huge economic impact from AI that we haven't seen so far in the next 18 months, because fundamentally, like, there's not that much in the economy that's just like chat-based. But if you do have that, then I think you can start seeing tens of billions of dollars of revenue per AI company and other kinds of big impacts.
And if we don't have that, you're saying we've hit some kind of wall? There needs to be another breakthrough. That's right. To get somewhere new.
That's right. Yeah.
