Only AI Can Build the Next AI | Richard Socher artwork

Only AI Can Build the Next AI | Richard Socher

MTS

October 26, 2026

Recursive co-founder and CEO Richard Socher breaks down the launch of Flint, their open-source local coding model built largely by an autonomous community of AI researchers.

0:00

Richard Socher

Right now, we're focused a lot on coding, because AI is code, and now that this year, really, only AI can code, you really want to focus on that capability, because that's very crucial to self-improvement.

0:15

SPEAKER_2

All right, we are back. We are live with Richard Socher, who is the co-founder and CEO of Recursive, company building self-improving super intelligence to automate knowledge discovery.

They just released their first model, Flint. So congratulations on the launch, Richard, and welcome to MTS.

0:32

Richard Socher

Thank you, and thanks for having me.

0:34

SPEAKER_2

Absolutely. So tell us about Flint. What is it, and who is it for, and how does it differ from other models?

0:40

Richard Socher

Flint is the best and fastest local coding model that you can actually download. It's fully open source. It's better than any of the other models in the category, pretty much all of which are Chinese. It's also one of the first American-made open source models that's on the predo frontier and beyond.

We think for us, it's a great first step, get into the habit of shipping. These things usually take companies a lot longer as they get started, and we used AI and a very small but very strong team of AI researchers to actually develop this model. So the AI did a lot of the work for us.

1:26

SPEAKER_2

Interesting. Can you go into a little more detail about that? How is the AI helping improve the AI?

1:31

Richard Socher

Yeah. So a big part of Recursive is building this Eureka machine. We call it the Eureka machine, somewhat aligned with my book, but focused first exclusively on AI research. The idea is that inside this Eureka machine, you have a community of AI scientists that essentially ideate, implement and validate ideas in an open-ended fashion. Open-endedness is this idea that's inspired by biological, cultural, and technological evolution to combine interestingly different ideas, allow some branches of evolution to go on for a while, and then recombine when they come up with interestingly different ideas.

That is what we've done here. Pretty much every aspect of this model, AI helped us come up with new ideas, implement them, and then validate them. That's something that we think no other company has been able to automate to that degree.

2:33

SPEAKER_2

You said coming up with ideas, implementing them, and then validating them. Which of these steps right now is easiest for AIs to do and which is hardest?

2:44

Richard Socher

I think almost every company when they talk about Recursive Self-Improvement, they say, oh, we're doing a lot of it because our developers are using AI to implement ideas. So that one, I think, is most consensus. I think how good you can make AI at coming up with freely novel ideas is probably the least consensus. But then also truly validating things is non-trivial. So there's like, it's sort of people don't think about it as much to make it either consensus or non-consensus. But there's also a lot of complexity there. People know that as these models get more intelligent, they often try to solve problems based on the letter of your prompt, not the meaning and the underlying intent of it.

So we think we have a better handle on some of those problems as well.

3:38

SPEAKER_2

What will it take for AI to actually become as good at radiation as the best human AI researchers?

3:46

Richard Socher

So yeah, there's a lot to unpack there. I think at a high level, people underestimate the crazy, complex, combinatorial space of already existing ideas. If you just combine all the concepts and all the ideas of physics, chemistry, biology, computer science and theoretical computer science, and complexity theory and systems and operating systems, all of these different things, there's actually a lot of optimizations. There's a lot there that you can already play around with. And similar to chess, you have to not just try them all out because they're combinatorially too many, but you have to build an intuition of which of these idea combinations is the most viable. And so, yeah, I think this combinatorial space is really underexplored. I think the next level is to have even completely new concepts. And to be honest, even there, not many human researchers are able and capable of doing that. And in many cases, the next invention is an incremental step across existing ideas. Even let's look at the transformer.

The title of the paper is Attention is All You Need. Why?

It's not because attention was invented in the transformer paper. My papers, lots of other people had used various levels of attention before. They learned, and now the big insight was you don't need any recurrent neural network layer. So sometimes even deleting an idea is novelty in and of itself, right? To simplify things, it can be very beautiful and then create more generality. So there's a lot to unpack there, but I have no doubt, and we're seeing already, that AI is getting better and better and better at ideating.

Read the other 13 minutes of this episode here.

$15 gets you this episode and 99 more. Credits never expire. After checkout, one link brings you back here to read it.

Not what you expected? Full refund within 14 days.

Thousands of transcripts fetched by people building searchable podcast archives

Fetch transcripts with code

The same credits work from a script or an AI agent. The demo key returns a sample episode in full, no card needed:

request
curl -H "x-api-key: pt_demo" \
  https://spoken.md/transcripts/1000651996090

Markdown with the speakers named, for your notes, your knowledge base, or anything that makes HTTP calls.

From $0.10 per transcript. No subscription. Credits never expire. Prices exclude VAT, added at checkout for EU customers. Not what you expected? Email us within 14 days with 20 or fewer credits used and we refund the pack in full.

Using your own key:

request
curl -H "x-api-key: YOUR_KEY" \
  https://spoken.md/transcripts/1000798243533