I'm very bullish on autoresearch in AI for biology.
There are lots of niche domains in biology that require their own specific datasets and inductive biases, so the "AI for bio" talent naturally gets diluted across a number of problem areas. Contrast this with LLMs, where all of the researchers cluster around a small subset of hyper-generalizable problems
So I think bio ML is much more "intelligence constrained" than LLMs and will see huge gains from autoresearch
