Tim O'Reilly

I've been semi-retired for the last 10 years and AI has got me back in founder mode for two reasons. One, it's an existential threat to our business. You know, I do think that there's a lot of new capabilities and basically the AI companies have basically been pretty ruthless in stealing everybody's content and basically making

I've been semi-retired for the last 10 years and AI has got me back in founder mode for two reasons. One, it's an existential threat to our business. You know, I do think that there's a lot of new capabilities and basically the AI companies have basically been pretty ruthless in stealing everybody's content and basically making

products that effectively compete with the people whose content they trained on. That means we have to think about, well, what's our value beyond replacement? And we have a lot of ideas about that and we're doing some great new products, but it's also calls for kind of some of the things that I've always been It's been a big part of what I do in the past, which is storytelling.

products that effectively compete with the people whose content they trained on. That means we have to think about, well, what's our value beyond replacement? And we have a lot of ideas about that and we're doing some great new products, but it's also calls for kind of some of the things that I've always been It's been a big part of what I do in the past, which is storytelling.

So one of the things that's been a threat to our business is there's been a whole narrative out there that we're not going to need programmers anymore. O'Reilly is primarily an enterprise learning service that we sell to corporations. I mean, we do still sell books and so on, but the biggest part of our business is a subscription.

So one of the things that's been a threat to our business is there's been a whole narrative out there that we're not going to need programmers anymore. O'Reilly is primarily an enterprise learning service that we sell to corporations. I mean, we do still sell books and so on, but the biggest part of our business is a subscription.

So these corporate learning buyers are like, hey, everybody's telling us we're not going to need developers anymore. And so I go, well, A, it's wrong. And B, I want to tell you why. I just organized a four-hour virtual conference. We had something like 22,000 people sign up.

So these corporate learning buyers are like, hey, everybody's telling us we're not going to need developers anymore. And so I go, well, A, it's wrong. And B, I want to tell you why. I just organized a four-hour virtual conference. We had something like 22,000 people sign up.

And we told the story of why AI actually still requires programmers to change their job, which, of course, for our business is a good thing because they actually have to learn new skills or double down on skills that we're undervaluing. Chelsea Troy, for example, is one of our computer science professor and one of our experts who does stuff for us on the platform, gave this fabulous talk.

And we told the story of why AI actually still requires programmers to change their job, which, of course, for our business is a good thing because they actually have to learn new skills or double down on skills that we're undervaluing. Chelsea Troy, for example, is one of our computer science professor and one of our experts who does stuff for us on the platform, gave this fabulous talk.

And he said, look, we have this incredible mismatch between the idea of what a programmer does and what they actually do and how we teach them. We try to teach people to write code from scratch. And then once they get out in the real world, most of their work is actually fixing code that was written by someone else. And now, you know, with AI, maybe it's gone from 80% to 96%, you know.

And he said, look, we have this incredible mismatch between the idea of what a programmer does and what they actually do and how we teach them. We try to teach people to write code from scratch. And then once they get out in the real world, most of their work is actually fixing code that was written by someone else. And now, you know, with AI, maybe it's gone from 80% to 96%, you know.

But it's not as big a gap as you think. And we've always had to teach people the skills of how do you actually think through a problem? How do you actually figure out whether the code is doing what you thought it should do? Again, one wonderful quote from a friend in the past, a guy named Andrew Singer, told me this wonderful line once. I was working on a manual.

But it's not as big a gap as you think. And we've always had to teach people the skills of how do you actually think through a problem? How do you actually figure out whether the code is doing what you thought it should do? Again, one wonderful quote from a friend in the past, a guy named Andrew Singer, told me this wonderful line once. I was working on a manual.

It was the first C compiler for the Mac, ThinkC. It was called LightspeedC originally. And Andrew... Drop this line, which I've treasured. He said, the skill of debugging is figuring out what you really told your program to do instead of what you thought you told it to do. When you think about that, what did you really tell your LLM to do? It's still true. You've got to figure out.

It was the first C compiler for the Mac, ThinkC. It was called LightspeedC originally. And Andrew... Drop this line, which I've treasured. He said, the skill of debugging is figuring out what you really told your program to do instead of what you thought you told it to do. When you think about that, what did you really tell your LLM to do? It's still true. You've got to figure out.

We've got this wonderful piece recently by Philip Carter at Honeycomb who wrote about LLMs are weird computers. We now have two computers really, two full computers that have to be brought together. One of them can write a sonnet but can't do math. And the other one can do math, but couldn't write a song no matter what you did to it.

We've got this wonderful piece recently by Philip Carter at Honeycomb who wrote about LLMs are weird computers. We now have two computers really, two full computers that have to be brought together. One of them can write a sonnet but can't do math. And the other one can do math, but couldn't write a song no matter what you did to it.

Programming is now getting the best out of these two systems, not just one. So it's like this amazing time to be helping to figure out what does programming look like today? What can we do with these machines? And I'm super excited about it, and we're figuring it out and teaching it. Yeah.

Programming is now getting the best out of these two systems, not just one. So it's like this amazing time to be helping to figure out what does programming look like today? What can we do with these machines? And I'm super excited about it, and we're figuring it out and teaching it. Yeah.