You can bet your sweet bippy that AI boosters will still be using AlphaFold as one of their go-to examples of how useful we're supposed to believe AI is, though...
-
RE: https://scholar.social/@jrhawley/117009347431672491
You can bet your sweet bippy that AI boosters will still be using AlphaFold as one of their go-to examples of how useful we're supposed to believe AI is, though...
"But the tech really *is* useful, this is just a business model problem!" - well, if you believe those are separate things, I don't think I can help you. Wake up & smell the tech bro era.
-
RE: https://scholar.social/@jrhawley/117009347431672491
You can bet your sweet bippy that AI boosters will still be using AlphaFold as one of their go-to examples of how useful we're supposed to believe AI is, though...
"But the tech really *is* useful, this is just a business model problem!" - well, if you believe those are separate things, I don't think I can help you. Wake up & smell the tech bro era.
@jwcph
I think there are a few reasonable interpretations, but @m_correa_m has a good and reasonable take here
https://fediscience.org/@m_correa_m/117008634177947668For the arguments regarding alpha fold, it's important for me to point at the data that's used to train the model.
It's governed by nature with clear relations between DNA, coevolution effects and biophysics of the resulting molecules.To me that's much different from the models trained on texts, which are the ones we generally talk about when mentioning AI.
-
@jwcph
I think there are a few reasonable interpretations, but @m_correa_m has a good and reasonable take here
https://fediscience.org/@m_correa_m/117008634177947668For the arguments regarding alpha fold, it's important for me to point at the data that's used to train the model.
It's governed by nature with clear relations between DNA, coevolution effects and biophysics of the resulting molecules.To me that's much different from the models trained on texts, which are the ones we generally talk about when mentioning AI.
-
@m_correa_m @folfdk Exactly - it's also kind-of my point, that the hype boosters keep using examples from related tech fields, even though they have little if anything to do with the hyped versions ("If AI can fold proteins it can also write research papers") & will never survive the business models of the companies behind them...
-
J jwcph@helvede.net shared this topic
-
@m_correa_m @folfdk @jwcph I try to be precise about "LLMs" vs other AI/ML, but I did use the term "AI industry" intentionally. I see lots transformer-style models published all the time for DNA interactions, protein folding, etc. I'm not worried about the lack of work in these areas, scientifically.
What I wanted to highlight was the labour aspect of it all. The industrial corporate players in this industry are anti-labour, and that should inform people how these tools can and will be used.
-
