For once I agree with Sam Altman (it feels weird...).

I don't feel the hype for coding assistant tools using LLMs (sorry for HN nerds here).

However I am incredibly excited with all the new research and tools produced by DeepMind since years now, like "AlphaFold 2". "AlphaFold 2" has more potential to save lives than every other LLMs produced in the last decade.

This kind of AI tools and research need to attract news (unfortunately it does not... yet?).

I have a friend at a startup for "autonomous agents for science" and they have seen enough to also think this will be a big deal. It will be interesting to see whether or not there's a bunch of low-hanging fruit of insights to be gained from combining existing literature in a way that AI is able to do at scale (compared to human scientists who have only so much time/ability to read papers).