The statement seems absurd to me too.  And even though I might not be able to predict their *exact words*, I think how AI will respond to a question is actually very predictable.

SimpleStacker

we cannot understand how AI agents think, even as they understand (at least to the point of being able to predict our actions better than we do theirs) how we think.

Thus seems like an absurd statement to me. If it was the case, I doubt we'd find AI tools very useful. Instead, we have all grown quite comfortable with relying on AI tooling because we do understand how it works.

At the very least, we understand AI as much as AI can be said to understand human minds or society. I see no asymmetry.

econ

The Commentariat Muses on AI and Financial Market Risk (FT, Martin Sandbu)

denlillaapan

> we cannot understand how AI agents think, even as they understand (at least to the point of being able to predict our actions better than we do theirs) how we think.

Thus seems like an absurd statement to me. If it was the case, I doubt we'd find AI tools very useful. Instead, we have all grown quite comfortable with relying on AI tooling because we do understand how it works. 

At the very least, we understand AI as much as AI can be said to understand human minds or society. I see no asymmetry.