From Isiah Andrews, via Emily Oster and the excellent Samir Varma.  A good piece, though I think it needs to more explicitly consider the most likely case, namely that the models are better at all intellectual tasks, including “taste,” or whatever else might be knockin’ around in your noggin…I am still seeing massive copium.  But the models still are not able to “operate in the actual world as a being.”  Those are the complementarities you need to be looking for, namely how you as a physical entity can enhance the superpowers of your model, or should I express that the other way around?