You say you can have increasingly nuanced discussions with stronger models.
What I say is, when I asked Claude why he applied a certain change I didn't understand, and boy, it was a small change, he said he "reasoned from first principles" based on the code paths. But it didn't work, and when I asked, "Okay, describe the steps of your reasoning from first principles," it literally answered that it had just made it up.
So, nuanced discussions with models, I don't buy it.
What I say is, when I asked Claude why he applied a certain change I didn't understand, and boy, it was a small change, he said he "reasoned from first principles" based on the code paths. But it didn't work, and when I asked, "Okay, describe the steps of your reasoning from first principles," it literally answered that it had just made it up.
So, nuanced discussions with models, I don't buy it.