10% chance AI could ‘kill all humans within a decade’ Anthropic alignment lead warns

Posted by tohya-san

8 Comments

  1. As AI is making new jumps and advancements this week, the alignment lead at Anthropic says something very strange and scares the hoes

    The real question is, if you sincerely believe your work could kill all of humanity in the next ten years, a non-trivial chance at that, why aren’t you sabotaging that work? To me his excuse if “We can be the only ones to build it safely, if we don’t, someone else might kill everyone!” doesn’t add up with the level of risk i’d be *truly* believes what he’s saying

    Or maybe I’m more averse to killing humanity, if such a thing could be real

  2. Even-Promotion-4024 on

    When people wonder why AI and tech in general has a bad image with the public, it might be smart to pause and think about just how misanthropic and downright weird a lot of the sector has become recently. It’s hard to buy into data center YIMBYism when they’re being used by an industry constantly touting the fact its product is a literal existential risk

    Totally expecting the “graph go up” crowd to downvote me, but I’ll personally take a guarantee of being alive in 10 years over the potential for 20% greater GDP growth 11/10 times on this one

  3. National-Return9494 on

    I am peripheral to the specific communities. But P doom has been a thing for ages, (well it used to be a way more common convo topic a few years ago). Anyway, I think most are above 5 below 25%

Leave A Reply