The only thing that seems to live up to the hype is AlphaFold, which predicts protein folding based on amino acid sequences, and of which people say that it actually makes their work significantly easier.
But, disclaimer, this is only from second-hand knowledge, I'm not working in the field.
This is another dimension of the problem - what's even considered AI ? AlphaFold is a very specialized model - and I feel the AI boom is driven by hypothesis that general models eventually outperform specialized ones given enough size/data/whatever.
While I hate the apparent renaming of everything ML to "AI", things like AlphaFold would be "narrow AI".
As to the common idea of having to wait for general AI (AGI) to bring the gains, I have been quite sure since the start of the recent AI hype cycle that narrow AI will have silently transformed much of the world before AGI even hits the town.
In my head, I just substitute "AI" with "machine learning" or "statistics".
> and I feel the AI boom is driven by hypothesis that general models eventually outperform specialized ones given enough size/data/whatever.
I think in the sciences, I'd generally put my money on the specialized models.
I hope that the hype around AI makes it easier (by providing tooling, platforms, better algorithms, educational materials etc.) to train specialized models.
The only thing that seems to live up to the hype is AlphaFold, which predicts protein folding based on amino acid sequences, and of which people say that it actually makes their work significantly easier.
But, disclaimer, this is only from second-hand knowledge, I'm not working in the field.