Alex argues that since almost all species communicate via chemistry, giving AI a sense of smell trains it on a genuinely alien form of intelligence, and invokes Terry Tao's call for a 'Copernican' view that decenters human intelligence in AI. ✦ AI generated
Alex Wiltschko · The TWIML AI Podcast · 2026-07-08 · original ↗
starts at this moment · 46:56
“So we've talked a lot about kind of where this is all going and what's possible, but any any additional thoughts on that?”
99% of species on this planet can only speak with chemistry, right? Thinking of bacteria and fungi and and plants and insects, like they just only can talk with molecules. And I think that it's really worth adding other kind of alien forms of intelligence to our AI models. And the way to do that is to train it on the intellectual output of those other intellects, which is that's chemistry.
verbatim transcript · starts at 46:56
46:56That's the sense that's in the air. They're produced by living things for reasons to talk to each other. And uh you know, Terry Tao, who is this famous mathematician, he has this one paragraph in a paper he just wrote where, you know, he's he's you know, reflecting on these new forms of AI and the kind of weird mathematics that they can do. And he says, you know, I think increasingly
47:16we need to have a capernac view of intelligence. Meaning for a long time, you know, in astronomy, we had the earth at the center, but then we had to dislodge it to actually face facts. Um, and earth's a great place to be, but it's not the center of the universe, right? Um, and similarly, we've placed human intelligence at the center of the debate about artificial intelligence. And like pretty cool to be human. It's
47:38it's awesome. I love it. Um, but like why should we be at the center? Like what other forms of intellect are there? And so when he when he when I read what he wrote, I've kind of instantly like it instantly resonated with me and I've very much taken on this view that like we should have a Capernac view of intelligence. There's so many forms of it and there's so much uh richness out
47:58there um of things that are intelligence, some of which we might not even recognize as such. Uh and I think if we're building [clears throat] AI uh for our species and for the planet going forward, like we should have those forms of intelligence built in. thinking about all the progress that we've been making in you know broadly physical AI you know robots autonomous vehicles all that um
48:23and thinking as well about like how good our sensors are for you know this approximation of visual stimuli like do you have a sense for I'm I'm imagining you do like where are we with like the you know oldactory camos or whatever that uh you a nose on a chip. >> I would say the devices that we use to detect chemistry are truly phenomenal, but they're they've been very
48:53specialized to live in a laboratory. So, think about computation in like the 1970s, big mainframes, right? Stuck in a back office or like a data center, right? So, we're still in the data center era of chemical sensing and of all factory intelligence. Um, so like our scent printer is the size of a school bus. It's just not We're not going to That's not going to leave. That's bolted into the concrete, right?