Isn't the point that now Andrej's published this, it will be in-distribution soon?

That's a very provocative last point, but I suppose we are living in an unwitting experiment.

Assuming this observation applies generally, give one point to the embodiment crowd.

My prior is that if Daron Acemoglu thinks the impact of Gen AI on productivity is muted, then it's probably a good unbiased gauge.

This seems like an extreme of example of "all models are wrong, but some are useful".

Yes why are these not published in a nice book? I would love to read these too!

Need more data.

from the article "A pattern we see in some interpretability and interpretability-adjacent ML papers is defining some metric which is claimed to correspond to some property of interest, and then very rigorously measuring this metric. We see this as a kind of Cargo-Cult Science."

This is a good overview https://www.akingump.com/en/insights/alerts/commerce-issues-...

I think people will continue to value human-made paintings much like we still value human-played chess games.