Comment on In a 38-page essay, Dario Amodei warns of civilization-level damage from superintelligent AI, questioning whether humanity has the maturity to handle such power

verdi@tarte.nuage-libre.fr ⁨5⁩ ⁨days⁩ ago

• Perspectivist@feddit.uk ⁨5⁩ ⁨days⁩ ago

  This comes across more as a warning than a sales pitch but I get the reflexive hostility on your part - the headline mentions “AI” after all.

  source

  • verdi@tarte.nuage-libre.fr ⁨5⁩ ⁨days⁩ ago

    There is factually 0 chance we’ll reach AGI with the current brand of technology. There’s neither context size or compute to even come close to AGI. You’d have to either be selling snake oil or completely oblivious about the subject to even consider AGI as a real possibility. This tells me the average user really doesn’t know shit… 

    source

    • Perspectivist@feddit.uk ⁨5⁩ ⁨days⁩ ago

      It’s perfectly valid to discuss the dangers of AGI whether LLMs are the path there or not. I’ve been concerned about AGI and ASI for far longer than I’ve even known about LLMs, and people were worried about exactly the same stuff back then as they are now. This is precisely the kind of threat you should try to find a solution for before we actually reach AGI - because once we do, it’s way, way too late.

      Also:

      There is factually 0 chance we’ll reach AGI with the current brand of technology.

      You couldn’t possibly know that with absolute certainty.

      source

      • verdi@tarte.nuage-libre.fr ⁨5⁩ ⁨days⁩ ago

        >You couldn’t possibly know that with absolute certainty.

        I recommend you read Cameron’s very good layman’s explanation.

        Adding to that framework, there is not enough data, compute and context size to reach AGI, for the current level of technology to reach anywhere near an AGI.

        source

        • -> View More Comments