> Better, it seems, to do what Jim Simons did, and try to imagine how we can contort our problem—understanding the world, so we can make better decisions about it—to fit what the technology is good at.
Reminds me of how early films still use ideas from plays on stages for their camera angles
Wide screen and no edits
It’s only when people embrace the new tech as a new medium and invent a new cinematic language do we have films flourishing as entertainment
A lot of the early internet seemed like that too. It was printed pages on a screen, and took a while for the internet to turn into something that was built for the medium.
I don't think this is exactly analogous, because I don't think every company will need to be built for the "AI medium." But it seems like the ones that are should operate very differently than those that aren't.
I wrote a comment before properly reading your piece, so … I apologise!
Broadly, yes, I agree with what you’re saying here! AI isn’t going to make a big impact on becoming data driven, because that’s a human problem, not a technical problem.
And the problem for which it’s going to be an ideal solution for is probably not going to be like anything we’ve seen before.
(It’s actually a little crazy to make a prediction on this, but for as long as businesses remain a human enterprise, the whole analytical thing is going to bottleneck on a human / org design problem, not a technical one. Bezos talks about finding out about things that don’t change, and how that is massively valuable. This insight might be one of those)
It will be quite fun to see what AI is suitable for; like you I’m looking forward to see what folks figure out.
Yeah, that's sort of the weird thing to me - so many people's reactions to AI stuff has been, "Ah, look, we will soon all be brilliant analytically minded companies," but never quite explain the mechanic for how that happens? Even in the best case - the AI bots are essentially identical to super smart generalists - that doesn't really make a business any closer to operating like Amazon. That's one of the reasons I really appreciate your articles: They make it really clear that the problem is this very messy, very hard organizational and cultural thing, not "we need more smart people with better tools."
If everyone company has access to a million lawyers and the rational choice is to pivot into becoming a law firm, then which newly-minted law firms win? The ones with the most to spend on a cloud bill? Best tv ads? Most persistent ambulance chasers?
I don't know anything about the history of trading, but I can't help think there had to be a lot of things about the way Jim Simons was running Renaissance pre-deep blue that made him and them well positioned to win on algorithmic trading. Surely others caught wind of what he was doing and tried to compete. At the very least he knew how to hire the right physicists and computer scientists, which isn't something everyone in finance could do.
Shoehorning LLM's into a better way of doing the status quo stems from an understandable desire to lean into moats or competitive advantages, but there's absolutely a balance there. Very excited to see what folks figure out here in the next few years. Maybe it turns out that the most caffeinated lawyers win?
On "everyone becomes lawyers," yeah, I thought about that a bit, and I think it's sorta true. In a world where there literally are a million lawyers available to anyone, I'm with you that it might not actually change anything. If your competitive advantage was in making coffee yesterday, it still would be tomorrow. But that's probably where the analogy falls down somewhat. AI is only kinda sorta like a bunch of lawyers, and most people won't become a law firm because it takes a lot of work, you still need some legal expertise, etc etc. So I think the actual version of this story is something like, "AI is this incredibly powerful thing that dramatically changes the resources you have. Given that, you probably should be thinking a lot more broadly about what sort of business you should build, and it's quite possible that the right thing to do is very different than what you're currently doing." But that was a less fun thing to say, I guess.
Which is probably closer the story about Renaissance. They were somewhat early to the quant game, and because of where they came from (MIT et al), they probably had access better talent (though that would equal out pretty quickly as soon as you start making huge piles of money that you can pay people). But from my understanding of their history, the most differentiated thing they did was fully lean into the quantitative piece. Most people tried to be traders with better calculators, whereas they tried to just make trading a problem a calculator could solve. Most people took half steps towards what they were doing; Renaissance went all in.
> Better, it seems, to do what Jim Simons did, and try to imagine how we can contort our problem—understanding the world, so we can make better decisions about it—to fit what the technology is good at.
Reminds me of how early films still use ideas from plays on stages for their camera angles
Wide screen and no edits
It’s only when people embrace the new tech as a new medium and invent a new cinematic language do we have films flourishing as entertainment
A lot of the early internet seemed like that too. It was printed pages on a screen, and took a while for the internet to turn into something that was built for the medium.
I don't think this is exactly analogous, because I don't think every company will need to be built for the "AI medium." But it seems like the ones that are should operate very differently than those that aren't.
I wrote a comment before properly reading your piece, so … I apologise!
Broadly, yes, I agree with what you’re saying here! AI isn’t going to make a big impact on becoming data driven, because that’s a human problem, not a technical problem.
And the problem for which it’s going to be an ideal solution for is probably not going to be like anything we’ve seen before.
(It’s actually a little crazy to make a prediction on this, but for as long as businesses remain a human enterprise, the whole analytical thing is going to bottleneck on a human / org design problem, not a technical one. Bezos talks about finding out about things that don’t change, and how that is massively valuable. This insight might be one of those)
It will be quite fun to see what AI is suitable for; like you I’m looking forward to see what folks figure out.
Yeah, that's sort of the weird thing to me - so many people's reactions to AI stuff has been, "Ah, look, we will soon all be brilliant analytically minded companies," but never quite explain the mechanic for how that happens? Even in the best case - the AI bots are essentially identical to super smart generalists - that doesn't really make a business any closer to operating like Amazon. That's one of the reasons I really appreciate your articles: They make it really clear that the problem is this very messy, very hard organizational and cultural thing, not "we need more smart people with better tools."
this is why it's all so interesting... we are all missing an important idea and yet it feels so very close.
If everyone company has access to a million lawyers and the rational choice is to pivot into becoming a law firm, then which newly-minted law firms win? The ones with the most to spend on a cloud bill? Best tv ads? Most persistent ambulance chasers?
I don't know anything about the history of trading, but I can't help think there had to be a lot of things about the way Jim Simons was running Renaissance pre-deep blue that made him and them well positioned to win on algorithmic trading. Surely others caught wind of what he was doing and tried to compete. At the very least he knew how to hire the right physicists and computer scientists, which isn't something everyone in finance could do.
Shoehorning LLM's into a better way of doing the status quo stems from an understandable desire to lean into moats or competitive advantages, but there's absolutely a balance there. Very excited to see what folks figure out here in the next few years. Maybe it turns out that the most caffeinated lawyers win?
On "everyone becomes lawyers," yeah, I thought about that a bit, and I think it's sorta true. In a world where there literally are a million lawyers available to anyone, I'm with you that it might not actually change anything. If your competitive advantage was in making coffee yesterday, it still would be tomorrow. But that's probably where the analogy falls down somewhat. AI is only kinda sorta like a bunch of lawyers, and most people won't become a law firm because it takes a lot of work, you still need some legal expertise, etc etc. So I think the actual version of this story is something like, "AI is this incredibly powerful thing that dramatically changes the resources you have. Given that, you probably should be thinking a lot more broadly about what sort of business you should build, and it's quite possible that the right thing to do is very different than what you're currently doing." But that was a less fun thing to say, I guess.
Which is probably closer the story about Renaissance. They were somewhat early to the quant game, and because of where they came from (MIT et al), they probably had access better talent (though that would equal out pretty quickly as soon as you start making huge piles of money that you can pay people). But from my understanding of their history, the most differentiated thing they did was fully lean into the quantitative piece. Most people tried to be traders with better calculators, whereas they tried to just make trading a problem a calculator could solve. Most people took half steps towards what they were doing; Renaissance went all in.
Gold 🙌
Thank you 🙏