Up until now, I’ve mostly done two things in this newsletter: explain what’s happening in AI, and show you how to use it better. Today’s edition is going to be different. I want you to understand how to think about what’s coming, so you end up in the best possible position a year from now.
So here’s what happened. Last week, OpenAI announced that an unreleased AI model solved a math problem that has stumped the best mathematicians in the world for over 150 years. It’s so hard that whoever solves it gets a million dollars.
I bet you can guess where this is going…
OpenAI put 10,000 AI agents on it, and it looks like they’ve solved it. It took just four days.
Those agents cost OpenAI millions of dollars to run. So not just anyone could have done this. It took a company with more computers than most countries.
But here’s what I want you to understand: in AI, this year’s million-dollar breakthrough is next year’s $20 subscription. And the AI you already pay for gets a lot smarter in the meantime, without you lifting a finger.
Think about the first computer your family owned. It cost a fortune and could barely do anything. A few years later it was in the bargain bin, and the new one at the same price did things nobody imagined a few years before. AI does the same thing every single year, but the difference is dramatically larger.
One of OpenAI’s top researchers put numbers on it. In late 2024, getting a top score on one of the hardest tests for AI cost about $500,000. Today, the basic AI model in ChatGPT beats that score for about $20. He thinks a year from now, everyone will have an AI at this level.
So a year from now, you’ll likely be able to point that kind of intelligence at a problem nobody has ever solved, for less than you pay for Netflix.
Nobody reading this cares about a math equation, and honestly, neither do I. What I care about is what you’d do with that power. Think about the thing you gave up on because it was too big. The company that needed a team of twenty. The book. The problem in your field that nobody has cracked.
Start planning for it now. You probably can’t do it today. But next year you probably can, and the people who already have a plan will be the ones who actually do it.
One more thing. There’s been a lot in the news this week about AI and whether it’s a threat to all of us, and a lot of it is confusing, alarming, or both. Later this week I’m going to sit down and write up what’s actually going on, in plain language, so you can understand what’s happening without a background in this stuff. If there’s something specific you want me to cover, or something you’ve been worried about, reply to this email. I read every one, and I’ll cover as many as I can.
— Matt