70
Google shuts down its Nobel-prize winning AlphaFold project as it focuses on Gemini
(www.engadget.com)
Wow AI! Such transformation! Such innovation!
this post was submitted on 29 Jul 2026
70 points (100.0% liked)
Technology
43303 readers
260 users here now
A nice place to discuss rumors, happenings, innovations, and challenges in the technology sphere. We also welcome discussions on the intersections of technology and society. If it’s technological news or discussion of technology, it probably belongs here.
Remember the overriding ethos on Beehaw: Be(e) Nice. Each user you encounter here is a person, and should be treated with kindness (even if they’re wrong, or use a Linux distro you don’t like). Personal attacks will not be tolerated.
Subcommunities on Beehaw:
This community's icon was made by Aaron Schneider, under the CC-BY-NC-SA 4.0 license.
founded 4 years ago
MODERATORS
Keep the slop, kill the science. Got it.
Arguably, approaches to optimisation problems such as AlphaFold are inherently biased and not really suited for real science. While other approaches take longer, their biases can at least be described. It doesn't help if AlphaFold is super fast, but then you don't really know if you can trust it's output. It's just like ChatGPT which is often correct, but that doesn't help if you never know when it is correct. However, with protein folding at least you can compare solutions and easily see which is better, because computing the potential energy of an existing solution is relatively simple. The hard part is to find a good solution.
Anyways, AlphaFold surely has its uses, and surely works well for proteins that fold in some sort of "standard" way, similar to the proteins in the training data set.
Mind you, theyre killing it in favor of an LLM, which is arguably further biased (you know, the giant "predict the most likely next work" algorithm, which doesn't predict anything new).
This is like saying computational biology is not suited for real science, as AlphaFold beat existing docking search algorithms on accuracy. Neither means you just get an answer spit out and your science is done, but it doesn't mean they're not productive tools in the process.
And all of these predictive methods, ML and not, are biased to existing data, because that's all the information we have to guide their development and gauge their success.
Arguably. Maybe.
My understanding is the predictions of alpha fold are always verified with the slower, more accurate process - if the predictions from alpha fold showed something promising.
Its main use was to help identify directions to further investigate. Not to blindly take the protein structure it predicted.
So there might be initial bias in terms of the how possibilities are narrowed, but you can certainly use it to do real science.