High Signal Podcasts Evidence ledger
Method
Browse
← Back to evidence

Evidence receipt / evaluation

Published · transcript-backed

Francois Chollet: evaluation

11 Jun 2024 Dwarkesh Podcast Francois Chollet — Why the biggest AI models can't solve simple puzzles

“The big difference is that you can never pre-train on everything that you might see at test time because the world changes all the time.”

— Francois Chollet

Source trail

Everything needed to verify it.

Speaker
Francois Chollet
Attribution
Verified speaker
Claim type
evaluation
Recorded
11 Jun 2024
Publisher
Dwarkesh Podcast

Transcript context

…The advantage they have is that they do get to see everything. Maybe I'll take issue with how much they are relying on that, but obviously they're relying on that more than humans do. They do have so much in distribution, to the extent that we have trouble distinguishing whether an example is in distribution or not. If they have everything in distribution, then they can do everything that we can do. Maybe it's not in distribution for us. Why is it so crucial that it has to be out of distribution for them? Why can't we just leverage the fact that they do get to see everything? Basically you’re asking what's the difference between actual intelligence — the ability to adapt to things you've not been prepared for — and pure memorization, like reciting what you've seen before. It's not just some semantic difference. The big difference is that you can never pre-train on everything that you might see at test time because the world changes all the time. It's not just the fact that the space of possible tasks is infinite. If you're trained on millions of them, you've only seen zero percent of the total space. It's also the fact that the world is changing every day. This is why we, the human species, have developed intelligence in the first place. If there was such a thing as a distribution for the world — for the universe, for our lives — then we would not need intelligence at all. In fact, many creatures, many insects for instance, do not have intelligence. Instead they have hardcoded programs in their connectomes, in their genes, behavioral programs that map some stimuli to appropriate responses. They can actually navigate their lives and their environment in a way that's very evolutionarily fit without needing to learn anything. If our environment were static and predictable enough, what would have happened is that evolution would have found the perfect behavioral program: a hard-coded, static behavioral program. It would have written it into our genes. We would have a hard-coded brain connectome. That's what we would be running on. But that's not what happened. Instead, we have general intelligence. We are born with extremely little knowledge about the world. We are born with the ability to learn very efficiently and to adapt in the face of things that we've never seen before. That's what makes us unique. That's what is really, really challenging to recreate in machines. Before we dive deeper into that, I'm going to overlay some examples of what an ARC-like challenge looks like for the YouTube audience. For people listening on audio, can you describe what a sample ARC challenge would look like?…

Stored transcript either side of the excerpt. The highlighted words are the published quote; the surrounding text is unedited source, never generated.

Search evidence