Alex Komoroske (Common Tools)
Episode publishedPodcast: The Slow HunchThe Slow Hunch
Transcript preview
Preview only
Transformers were not some magical thing that we discovered that we just happened to find. It is inevitable. And it is inevitable with enough data, with a loopback thing where it's generating the next token based on the prior tokens, and then the next token it generates now has token generated. And so it must, to some degree, have an internal model of itself to be able to steer itself in a direction. So yes, we've witnessed the first prokaryote to eukaryote transition. It was going to happen no matter what with enough time and space. And like the transformer architecture merely is one of the t
I'd never thought about the transformer architecture in terms of kind of emergence, almost a biological process.
This process of of software engineering to build LLM, you grow an LLM. You don't build it. It is not normal software. The process to produce them and the things that they produce are wildly unlike.
Welcome back to The Slow Hunch. This week, I'm excited to welcome Alex Komorowski to the conversation. Alex is maybe the perfect guest for The Slow Hunch, just an incredible thinker and one of my favorite people to talk to. For a long time, he was at Google leading large product organizations. He was recently head of corporate strategy at Stripe. Now he's the CEO of Common Tools, which is a startup building some interesting stuff in the land of AI, which USB has backed. Wide ranging conversation, you know, spanning the potential and perils of, what's possible in the world of AI, including some
Yeah. So I'm Alex. Thank you. I'm Alex Komorowski. I am the CEO and co founder of a company called Common Tools that works with USB. And we have a before that I was at Stripe as a head of corporate strategy. Before that I was at Google for thirteen years, ending up as a director of product management. I worked on ambient computing, augmented reality, Chrome web platform, search, and DoubleClick over that time. At the very beginning of my kind of intellectual career, I wrote my undergraduate thesis on the emergent power dynamics of Wikipedia's user community. Like under what conditions does thi
It's funny this podcast, This Little Hunch was based on Stephen Johnson's book, Where Good Ideas Come From. And I've read a bunch of his books and the other one of his books that I quite enjoyed was called Emergence. And, you know, I hadn't known about that term until I read the book about fifteen years ago. And maybe you could describe what you think that term means. Yeah.
So fun fact, that book actually is the thing that kicked me onto this, which I forgot until I was talking with Steven a couple of years ago, because I did the slime mold deck about like organizations of slime molds as an emoji flip book and about why companies get slow as an emergent process, as opposed to a specific like, and it was viral outside of Google and it was viral kind of outside of it. Was catching up with Steven Johnson. And in that conversation, I was like, wait a second, did you write that book on like complex symptoms? He's like, yeah. And I was like, wait a second, does it ment
That's amazing. And I've also, just in my head, reduced emergence also to the idea of kind of unexpected outcomes or behaviors that weren't planned or determined that are kind of and maybe that's what you mean by more than the sum of the parts, but it's this the fact that there are kind of unexpected properties that emerge over time by by way of all these
The second order effects, the third order effects, like multi order effects of stuff of, like, often you take one if you think just about I think this one act
Keep exploring with a free account.
A free account preserves the short preview. Upgrade to Pro for the full transcript and episode chat.