A group of OpenAI researchers ensuring that a future AGI system does not pose catastrophic or even existential danger has been decimated.
  • huginn@feddit.itEnglish
    253·
    3 months ago

    Well the good news is: it’s not agi, will never be agi and agi is a long way off.

    The bad news is: Open AI doesn’t publicly say this: they’re just firing the safety team with no explanation.

    • FaceDeer@fedia.io
      106·
      3 months ago

      will never be agi and agi is a long way off.

      That’s contradictory.

      Given the unexpected advances in AI over the past few years I wouldn’t be so sure that AGI is a long way off, and I certainly see no reason to expect that there will never be AGI. Whenever there’s an example of something already existing in nature it’s a good bet that we’ll be able to copy how it works using technology at some point.

      • Ephera@lemmy.mlEnglish
        7·
        3 months ago

        I guess, it depends on what we’re referring to as “it”. If we’re talking about the LLM-based ChatGPT, that’s true pretty much by definition. But if we’re talking about “AI”, which is a word that has been used for everything from calculators to Skynet, then yeah, at some point AI will be AGI.

        Well, if humanity doesn’t obliterate itself in the coming decades, which isn’t looking too good.

        • FaceDeer@fedia.io
          3·
          3 months ago

          Nobody specified LLMs until this comment and OpenAI does more than just LLM research, so “it” should be assumed to be AI in general.

      • mutant_zz@lemmy.worldEnglish
        11·
        3 months ago

        People say humans firing lightning bolts out of their hands is a long way off, but look at the progress we’ve made on lasers in the last few decades. I reckon we’re 5 years away, and we should be enacting laws to stop someone doing a Palpatine.