SDF Chatter
  • Communities
  • Create Post
  • Create Community
  • heart
    Support Lemmy
  • search
    Search
  • Login
  • Sign Up
Espiritdescali@futurology.todayM to Futurology@futurology.todayEnglish · 1 year ago

AI chatbots tend to choose violence and nuclear strikes in wargames

www.newscientist.com

external-link
message-square
9
fedilink
  • cross-posted to:
  • becomeme@sh.itjust.works
  • technology@lemmygrad.ml
  • technology@lemmy.world
  • nottheonion@lemmy.world
  • artificial_intel@lemmy.ml
35
external-link

AI chatbots tend to choose violence and nuclear strikes in wargames

www.newscientist.com

Espiritdescali@futurology.todayM to Futurology@futurology.todayEnglish · 1 year ago
message-square
9
fedilink
  • cross-posted to:
  • becomeme@sh.itjust.works
  • technology@lemmygrad.ml
  • technology@lemmy.world
  • nottheonion@lemmy.world
  • artificial_intel@lemmy.ml
As the US military begins integrating AI technology, simulated wargames show how chatbots behave unpredictably and risk nuclear escalation
alert-triangle
You must log in or register to comment.
  • Ottomateeverything@lemmy.world
    link
    fedilink
    English
    arrow-up
    18
    ·
    1 year ago

    Why even bother testing this? LLMs are not how you would want to run anything strategy related. And they’re trained of internet text data - pretty much all YouTube/Reddit comments about conflicts are full to the brim of people screaming about exterminating and nuking the “enemy” so of course that’s what LLMs would do.

    It’s scary how many people, even those in pretty high up positions of power, have no fucking clue how this shit works or what it can/will do. I used to think that AI wouldn’t be as dangerous as it is portrayed in books/movies because we would just keep it in line and have proper guard rails… But when people in the military etc seem to have no fucking clue as to how LLMs work… Yeah, we’re fucking doomed.

    • CanadaPlus@futurology.today
      link
      fedilink
      English
      arrow-up
      6
      ·
      1 year ago

      Exactly. I bet they escalate because they’re more instinctively drawn to a dramatic story than to pragmatic victory or desirable outcomes.

      Western militaries, at least, have made it doctrine to always keep a human in the loop, so that’s good.

  • Cyborganism@lemmy.ca
    link
    fedilink
    English
    arrow-up
    12
    ·
    1 year ago

    They WERE trained on human behaviour after all.

  • theodewere@kbin.social
    link
    fedilink
    arrow-up
    3
    arrow-down
    1
    ·
    1 year ago

    i really doubt you’re going to successfully train a chatbot to care about future generations

  • runner_g@lemmy.blahaj.zone
    link
    fedilink
    English
    arrow-up
    1
    ·
    edit-2
    1 year ago

    Must have been trained on Civilization’s Gandhi

  • NoLifeKing@ani.social
    link
    fedilink
    English
    arrow-up
    3
    arrow-down
    4
    ·
    edit-2
    1 year ago

    deleted by creator

    • Amerikan Pharaoh@lemmygrad.ml
      link
      fedilink
      English
      arrow-up
      4
      ·
      1 year ago

      The only winning move is not to play. Failed step one ig

      • NoLifeKing@ani.social
        link
        fedilink
        English
        arrow-up
        2
        arrow-down
        1
        ·
        edit-2
        1 year ago

        deleted by creator

        • iAmTheTot@kbin.social
          link
          fedilink
          arrow-up
          1
          ·
          1 year ago

          In case you weren’t aware, you’re responding to a reference from a popular 80s movie.

    • Semi-Hemi-Demigod@kbin.social
      link
      fedilink
      arrow-up
      3
      ·
      1 year ago

      How about a nice game of chess?

      • NoLifeKing@ani.social
        link
        fedilink
        English
        arrow-up
        1
        ·
        edit-2
        1 year ago

        deleted by creator

        • Semi-Hemi-Demigod@kbin.social
          link
          fedilink
          arrow-up
          1
          ·
          1 year ago

          Since you’re one of the lucky 10,000 today

Futurology@futurology.today

futurology@futurology.today

Subscribe from Remote Instance

Create a post
You are not logged in. However you can subscribe from another Fediverse account, for example Lemmy or Mastodon. To do this, paste the following into the search field of your instance: !futurology@futurology.today
Visibility: Public
globe

This community can be federated to other instances and be posted/commented in by their users.

  • 200 users / day
  • 371 users / week
  • 1.38K users / month
  • 6.27K users / 6 months
  • 24 local subscribers
  • 2.61K subscribers
  • 1.86K Posts
  • 11.7K Comments
  • Modlog
  • mods:
  • voidx@futurology.today
  • Lugh@futurology.today
  • Espiritdescali@futurology.today
  • AwesomeLowlander@futurology.today
  • BE: 0.19.8
  • Modlog
  • Instances
  • Docs
  • Code
  • join-lemmy.org