The video is a year old but it just showed up on my feed. I’m so amazed thinking of the crazy amount of possibilities this opens up for! Imagine making protein simulations available to anyone interested in testing out ideas!

  • Dusty@l.dusty-radio.com
    link
    fedilink
    English
    arrow-up
    5
    ·
    1 year ago

    If how AI is currently going (and currently way overhyped) is anything to go by, it’s just going to make everything up in a plausible sounding way and not actually accomplish anything.

    But it seems like every man and his dog has an opinion on why it’s going to be Next Big Thing tm

    • nottheengineer@feddit.de
      link
      fedilink
      English
      arrow-up
      1
      ·
      1 year ago

      A lot of the stuff LLMs spit out works, so give them a way to check what does and what doesn’t. Just like with any other form of machine learning: If you can build a loss function, you’re halfway there.