AlphaFold is great, but is it a revolution?
We are on the verge of a new scientific revolution driven by AI, according to an article in the Gradient and another recent article in the Guardian. AlphaFold, Deep-minds hugely impressive solution to the protein folding problem, has solved an important problem in biology. There are also some other (less notable) examples. Put all these examples together and you have a revolution!
I don't buy it. ML is a useful tool for solving certain types of problems that are amendable to brute force number crunching and pattern matching. AlphaFold is an impressive achievement, but scientific progress will still come from forming hypotheses, making predictions and conducting experiments.