The US Air Force wants $5.8 billion to build 1,000 AI-driven unmanned combat aircraft, possibly more, as part of its next generation air dominance initiative - eviltoast

The US Air Force wants $5.8 billion to build 1,000 AI-driven unmanned combat aircraft, possibly more, as part of its next generation air dominance initiative::The unmanned aircraft are ideal for suicide missions, the Air Force says. Human rights advocates call the autonomous lethal weapons “slaughterbots.”

  • mo_ztt ✅@lemmy.world
    link
    fedilink
    English
    arrow-up
    14
    arrow-down
    2
    ·
    1 year ago
    1. We all know why they put “AI-driven” in the headline… I mean, it worked on me; I clicked on it.
    2. That doesn’t mean they’ll be “autonomous” in the sense that people think of when they see the headline and click on it.
    3. Having a human in the loop does make a difference. Snowden talked about watching on his desktop people getting killed by drone strikes in real time, as part of his motivating factor for why he turned against the NSA and its mission. The Nazis had a lot of “morale problems” with Nazi soldiers who were assigned to holocaust-adjacent operations and had to find other solutions. Etc. Every human you take out of the equation is one less person who can rotate home and tell people, “Yo what they’re telling us to do is really fucked up, let me tell you…”
    4. I see the air force’s point. I honestly don’t blame them for feeling that there’s no future in an air warfare system that has to have a squishy slow-thinking meatbag in the middle of it putting limits on its performance. This kind of thing was already part of the plan for the US’s next generation fighter (with the pilot as the “commander” of a little network of drones) and has been for a while.
    5. If you haven’t seen Slaughterbots it’s well worth a watch.
    • nBodyProblem@lemmy.world
      link
      fedilink
      English
      arrow-up
      3
      ·
      1 year ago

      You’re right on all counts here.

      Computer algorithms (such as AI) can’t replace organic judgement-based decision making, but they vastly outperform humans when there is a well defined cost function to optimize against, such as, “hit this target in the minimum possible time”.

      I think you can compare it to autonomous cars. They can drive from point to point while avoiding hazards along the way, but they still need the passenger to tell them where their destination is.

        • nBodyProblem@lemmy.world
          link
          fedilink
          English
          arrow-up
          1
          ·
          1 year ago

          You don’t think that was implied when I said they vastly outperform human pilots?

          There are numerous advantages to letting a flight computer do the piloting. Higher allowable G limits is one of them, albeit far from the most important.