The keyword here isn't "understand". It's "novel". You "truly...
François Chollet@fchollet
115 views
Sep 10, 2025
~1 min read
Advertisement
1
The keyword here isn't "understand". It's "novel".
You "truly understand" a thing if your model of it lets you make sense of every possible instance of the thing, including those that are very far from what you've seen before (extreme generalization).
You "somewhat understand" the thing if you can approach at least some new instances of it (local generalization).
If you can only handle what you have seen before, you are merely memorizing / retrieving.
You "truly understand" a thing if your model of it lets you make sense of every possible instance of the thing, including those that are very far from what you've seen before (extreme generalization).
You "somewhat understand" the thing if you can approach at least some new instances of it (local generalization).
If you can only handle what you have seen before, you are merely memorizing / retrieving.
François Chollet@fchollet
A student who truly understands F=ma can solve more novel problems than a Transformer that has memorized every physics textbook ever written.
2
"Understanding" isn't some magical, ineffable concept. It's a very concrete and practical property of an agent, reflected in what it can *do*.
We know LLMs don't "truly understand" because we can see what they can't do. It's as simple as that.
We know LLMs don't "truly understand" because we can see what they can't do. It's as simple as that.