Robert M

Description.

There are some other positive slopes, but frankly they look like noise to me when 3 on both MMLU and GPQA.

Anyways, notice that on 4.

Of the 5 groups of questions, Gemma 3's incoherence drops with increasing model size.

Only on the hardest group of questions does it trend slightly upward.

I think that particular headline claim is basically false.

But even if it were true, it would be uninteresting because they define incoherence as the fraction of model error caused by variance.

Okay, now let's consider a model with variance of 1 times 10 to the power of negative 3 and bias of 1 times 10 to the power of negative 6.

Huge incoherence!

Am I supposed to be reassured that this model will therefore not coherently pursue goals contrary to my interests?

Whence this conclusion?

Similarly, an extremely dumb, broken model which always outputs the same answer regardless of input is extremely coherent.

A rock is also extremely coherent, by this definition.

A couple other random complaints.

The paper basically assumes away the possibility of deceptive schemas.

The paper is a spiritual successor of the 2023 blog post, The Hot Mess Theory of AI Misalignment.

More intelligent agents behave less coherently, LW discussion.

I think Guern's comment is a sufficient refutation of the arguments in that blog post.

This paper also reports the survey results presented in that blog post alongside the ML experiments as a separate line of evidence.

This is unserious.