• 1995ToyotaCorolla@lemmy.world
      link
      fedilink
      English
      arrow-up
      20
      ·
      20 hours ago

      Yeah, if by “rouge” you mean purposely taking guide rails off a model, giving it access to hacking tools, instructing it to hack, auto-approving everything the LLM does, and letting it run with no human oversight for long periods of time

      There is plenty of opportunity for OpenAI, Anthropic, et. al to take a step back and think about what they’re doing before dropping the brick on the accelerator. IDK how they’re not on the hook for any of this.

      • rmuk@feddit.uk
        link
        fedilink
        English
        arrow-up
        1
        ·
        4 hours ago

        “But if we don’t destroy without discrimination first, someone else might!”

      • Leon@pawb.social
        link
        fedilink
        English
        arrow-up
        3
        ·
        14 hours ago

        You know, I also don’t believe that these LLMs are rogue agents breaking out of sandboxed environments. I just don’t see how you could fuck up IT infrastructure that badly. Like, if I wanted to run something I believe could cause extreme harm, I’d airgap that fucker. Unless whatever I’m running somehow manages to grow arms and physically connect a wire to the network at large, it won’t be a danger.

        That said, I absolutely think we should take what they’re saying at face value and treat it with utmost seriousness. If we have a technology that the creators themselves say could be world-ending, that the creators themselves say has gone rogue on multiple occasions, then we should shut that thing down immediately, seize the assets of the corporations that produce it, and halt any and all development and study that thing.

        Given that they’ve “gone rogue” several separate times, these corporations clearly aren’t fit to develop or operate this technology.