If you ask me what was the best AI/ML result in last year, it was definitely AlphaFold! Forget OpenAI‘s GPT-3. AlphaFold is the real important result. In short, this is what research in AI/ML should aim for: to advance, or even solve, important scientific problems questions. I was very impressed with the results achieved with AlphaFold 2.0.
I must confess, when the first iteration came out, I was much more skeptical. Most due to how things were presented back then. But move forward a couple of years and you have an astonishing result! I’m not part of the field and never worked directly with protein folding ( I did a tiny bit of molecular docking more than 10 years ago) but reading from the multiple sources, it makes you dream. There is still much that AlphaFold can’t do but it opens a lot of doors.
I’m also happy to see a research lab like DeepMind to put this out. It makes their mission more realistic and it’s not just about games or something else. I hope this is just the first step towards more important results.
Jorge Tavares Notes 