Max Levchin

I decided two years ago that I was just going to read the attention paper and become an expert in LLMs, you know, just read all the papers after that. And I'm still in the process of getting anywhere near expertly status. I try to keep up. It's a lot of really interesting stuff happening in the AI.

I decided two years ago that I was just going to read the attention paper and become an expert in LLMs, you know, just read all the papers after that. And I'm still in the process of getting anywhere near expertly status. I try to keep up. It's a lot of really interesting stuff happening in the AI.

Agree. But I don't think we are going to stay with LLMs forever. I think we're going to expand the systems that use LLMs. Reasoning models already are not pure LLMs. They have some iterative processes within them. I think we'll invent more really interesting things with that. Like what?

Agree. But I don't think we are going to stay with LLMs forever. I think we're going to expand the systems that use LLMs. Reasoning models already are not pure LLMs. They have some iterative processes within them. I think we'll invent more really interesting things with that. Like what?

Agree. But I don't think we are going to stay with LLMs forever. I think we're going to expand the systems that use LLMs. Reasoning models already are not pure LLMs. They have some iterative processes within them. I think we'll invent more really interesting things with that. Like what?

An obvious idea, which I know has been explored and will be explored more, is a notion of multiple competing LLMs basically debating each other for the right answer. Like that's not a pure LLM model. You can amplify intelligence of humans by having three smart people in a room instead of just one person.

An obvious idea, which I know has been explored and will be explored more, is a notion of multiple competing LLMs basically debating each other for the right answer. Like that's not a pure LLM model. You can amplify intelligence of humans by having three smart people in a room instead of just one person.

An obvious idea, which I know has been explored and will be explored more, is a notion of multiple competing LLMs basically debating each other for the right answer. Like that's not a pure LLM model. You can amplify intelligence of humans by having three smart people in a room instead of just one person.

I generally really dislike stories that have no substance. So I think you have to ask yourself, are we taking full advantage of the best available tools? And the most interesting available tools right now are AI. So if you sort of look through your closet and say, hmm, we're not really using any of this stuff. I got to have a story for the market. You're doing two things wrong.

I generally really dislike stories that have no substance. So I think you have to ask yourself, are we taking full advantage of the best available tools? And the most interesting available tools right now are AI. So if you sort of look through your closet and say, hmm, we're not really using any of this stuff. I got to have a story for the market. You're doing two things wrong.

I generally really dislike stories that have no substance. So I think you have to ask yourself, are we taking full advantage of the best available tools? And the most interesting available tools right now are AI. So if you sort of look through your closet and say, hmm, we're not really using any of this stuff. I got to have a story for the market. You're doing two things wrong.

You're telling a story that isn't true, and you're not using the most interesting available tools. If you're using the most interesting available tools, your investors might be happy if you told the story. But if you didn't, I don't think they should really care that much. As you go public, a lot of this becomes, are you growing at the right scale? Are you taking appropriate risks?

You're telling a story that isn't true, and you're not using the most interesting available tools. If you're using the most interesting available tools, your investors might be happy if you told the story. But if you didn't, I don't think they should really care that much. As you go public, a lot of this becomes, are you growing at the right scale? Are you taking appropriate risks?

You're telling a story that isn't true, and you're not using the most interesting available tools. If you're using the most interesting available tools, your investors might be happy if you told the story. But if you didn't, I don't think they should really care that much. As you go public, a lot of this becomes, are you growing at the right scale? Are you taking appropriate risks?

Are you managing to profitability? The storytelling hour of like, oh, by the way, we have this shiny item that we're so excited about that feeds the voting machine. And I'm sure some of it is important, but I spent a lot more time feeding the weighing machine.

Are you managing to profitability? The storytelling hour of like, oh, by the way, we have this shiny item that we're so excited about that feeds the voting machine. And I'm sure some of it is important, but I spent a lot more time feeding the weighing machine.

Are you managing to profitability? The storytelling hour of like, oh, by the way, we have this shiny item that we're so excited about that feeds the voting machine. And I'm sure some of it is important, but I spent a lot more time feeding the weighing machine.

Basically do exactly the same thing I do now. So before Affirm was a thing, I made a list of difficult problems that I thought required solving. I decided that financial services or money at the core is one that I am well suited to solve, just given my past experience that I bring. But if I knew I couldn't fail at anything, I would revisit my list of hard problems.

Basically do exactly the same thing I do now. So before Affirm was a thing, I made a list of difficult problems that I thought required solving. I decided that financial services or money at the core is one that I am well suited to solve, just given my past experience that I bring. But if I knew I couldn't fail at anything, I would revisit my list of hard problems.

Basically do exactly the same thing I do now. So before Affirm was a thing, I made a list of difficult problems that I thought required solving. I decided that financial services or money at the core is one that I am well suited to solve, just given my past experience that I bring. But if I knew I couldn't fail at anything, I would revisit my list of hard problems.