• 2 Posts
  • 679 Comments
Joined 2 years ago
cake
Cake day: February 27th, 2024

help-circle

  • Information doesn’t hold energy themselves, so yes the CPU produces 80W of heat, no matter what computation they do.

    (Also it’s pretty hard to define what’s information, people generally talk about information entropy, which is different from thermodynamic entropy. This can be confusing).

    But your instinct is correct, one cannot reduce entropy without spending energy. And the kind of computation computers do today does change entropy, so some energy must be dissipated as heat to do computation. This is the Landauer’s principle: https://en.wikipedia.org/wiki/Landauer’s_principle

    Also you need to remember that energy can not be destroyed or created. Expending energy to do something doesn’t mean there will be less energy in the end.

    and that’s a no-no.

    Not sure why you think that? If information holds no energy then creating it without converting energy isn’t against any physical laws.














  • nialv7@lemmy.worldtoScience Memes@mander.xyzhow things become science
    link
    fedilink
    English
    arrow-up
    7
    arrow-down
    11
    ·
    edit-2
    5 days ago

    Edit: so the bluesky post wasn’t accurate. The fake papers were never published, they were pre-prints, and were written to be obviously fake if you paid any attention to read it. LLM didn’t have the ability to tell, some researchers took LLM’s output and didn’t check the source material.

    Don’t know what to say. Also kinda ironic this bluesky user apparently didn’t know the difference between a preprint and a published paper either.


    Original post:

    (I’ve only read the title. If turns out I am terribly mistaken I will come back and correct myself). More like scientists commit academic fraud and fooled a bunch of people. How did this get through the ethics board? Why would any publisher play along with this?