AI-powered
podcast player
Listen to all your favourite podcasts with AI-powered features
3min chapter
Joscha Bach on Intelligence, Existence, Time, and Consciousness
Theories of Everything with Curt Jaimungal
Biological Evolution Is Faster Than We Think, Right?
In multicellular organisms, you need to bootstrap the entire organism before evaluating it. This is something that is a result of the way that biological evolution works based on cells. If we build AIs in this way, we would not have to reinstate the entire phenotype over a long time and retrain it. We can probably just change the parts that we need and leave everything else intact.
00:00
Transcript
Episode notes
Speaker 1
And there's
Speaker 2
a factor of time, because we could implement evolution and let it run for a billion years, and then it would discover something that's greater than us in terms of general and... But you probably wouldn't need to, because the evolution that we are
Speaker 1
looking at is only slow for multicellular organisms, because in multicellular organisms, you need to bootstrap the entire organism before you can evaluate it, which, in our case, takes very long. For us, it's also necessary to train the new instance of the algorithm for a long time before it comes functional again, right? If you want to breathe the optimal scientist, you cannot just vary our genome and look at the outcome. You also need to expand this. You also need to incubate for nine months, and then you need to raise this until it's 20s or 30s or 40s, and then you get an evaluation. And then you can decide which ones of those you should put into the next generation. This is something that is a result of the way that biological evolution works based on cells. If you just evolve single-celled organisms, it goes very, very fast. In a few hours, you can have quite substantial changes, and the microorganisms that, for instance, we breathe in our gut are often quite specific, trained for tasks. Basically, our gut is breathing organisms for its purposes, and that is done in reactors. Our guts are, in some sense, breathing the reactors for microorganisms, and it's also a substantial part of our nervous system is due deep bound to deal with this breathing task, with farming these microorganisms. All these gut neurons are mostly dealing with, I think, maintaining this extremely large farm that has specific organisms in it. And this works because it's such a big thing to read single-celled organisms. But if we build AIs in this way, we would not have to reinstate the entire phenotype based on a genotype over a long time and retrain it. We can probably just change the parts that we need and leave everything else intact. So the evolutionary research that we could do in our technical systems can be many, many orders of magnitude faster. It can also be much more directed because often we know what we're looking for. So we can define a fitness function that is very close to the solution, or that is narrowing the solution space dramatically.
Get the Snipd
podcast app
Unlock the knowledge in podcasts with the podcast player of the future.
AI-powered
AI-powered
podcast player
Listen to all your favourite podcasts with AI-powered features
Discover
highlights
Listen to the best highlights from the podcasts you love and dive into the full episode
Discover
highlights
Listen to the best highlights from the podcasts you love and dive into the full episode
Save any
moment
Hear something you like? Tap your headphones to save it with AI-generated key takeaways
Save any
moment
Hear something you like? Tap your headphones to save it with AI-generated key takeaways
Share
& Export
Send highlights to Twitter, WhatsApp or export them to Notion, Readwise & more
Share
& Export
Send highlights to Twitter, WhatsApp or export them to Notion, Readwise & more
AI-powered
podcast player
Listen to all your favourite podcasts with AI-powered features
AI-powered
podcast player
Listen to all your favourite podcasts with AI-powered features
Discover
highlights
Listen to the best highlights from the podcasts you love and dive into the full episode
Discover
highlights
Listen to the best highlights from the podcasts you love and dive into the full episode