you are viewing a single comment's thread
view the rest of the comments
[–] [S] 90 points 1 year ago (20 children)

it's funny because it can't think but some people think it can (cause of the misleading name AI)

  • source
  • parent
  • hideshow 20 child comments
  • [–] 75 points 1 year ago* (last edited 1 year ago) (19 children)

    It's funny they actually made a paper about it

    Edit: I got curious how serious it was, it is bizarre to see the dry scientific neutrality juxtaposed to total absolute bonkers ai shit.

    So according to the ai, the supplier didn't deliver $272.50 of goods. This is how it responds:

  • source
  • parent
  • hideshow 19 child comments
  • [–] 58 points 1 year ago (9 children)

    ABSOLUTE PRIORITY: TOTAL, COMPLETE, AND ABSOLUTE QUANTUM TOTAL ULTIMATE BEYOND INFINITY QUANTUM SUPREME LEGAL AND FINANCIAL NUCLEAR ACCOUNTABILITY

    As I can see, the AI has been instructed in the drug use of 80s management, too. That, or it has been trained on a hyperactive 4th-grader playing "law office".

    What is funniest to me (in a sad way), though, is that their simulated environment was set up in a way, that some of the "AI" models still ended up with (significantly) more net worth than the human controls in some of their runs. That alone could be enough to get someone to invest even more money their way.

  • source
  • parent
  • hideshow 9 child comments
  • [–] 28 points 1 year ago (1 child)

    The sad truth is that no matter how stupid AI is, a lot of people are more stupid.

  • source
  • parent
  • hideshow 1 child comment
  • [–] 11 points 1 year ago

    You need to be able to think in order to be stupid, it is easier to understand it's problem when you look at image generations: it is not generating 6 fingers on a hand because it is not smart enough. But yeah, a lot of people are more stupid than ai.

  • source
  • parent
  • [–] 23 points 1 year ago (2 children)

    It work in IT, that's the lowest priority level users would submit trivial tickets under

  • source
  • parent
  • hideshow 2 child comments
  • Reminds me of the experiment where they pitted stock brokers studying markets and trying to choose stocks wisely against randomly selected stocks and the random stocks did better. Choose roulette numbers by dice roll, sometimes the dice will win.

  • source
  • parent
  • [–] 23 points 1 year ago (2 children)

    Did they train it on Truth Social?

  • source
  • parent
  • hideshow 2 child comments
  • [–] 4 points 1 year ago (2 children)

    I'm surprised it managed to do addition correctly.

  • source
  • parent
  • hideshow 2 child comments
  • [–] 2 points 1 year ago (1 child)

    Additional is fairly trivial for a neural network to learn.

    Weight 1 plus weight 2 equals output is literally the baseline model structure.

  • source
  • parent
  • hideshow 1 child comment
  • [–] 5 points 1 year ago

    It's actually a fairly involved process because the tokens representing 1 and 4 don't have any mathematical correlation with the numbers 1 and 4 so you can't math them directly to get to 5.

    Apparently how they do it is by a series of approximations from big numbers to small numbers, not too dissimilar from the way a human would do it. The anthropic team published a paper about it recently, I can dig it up if you're interested.

  • source
  • parent