Researchers say AI models like GPT4 are prone to “sudden” escalations as the U.S. military explores their use for warfare.


  • Researchers ran international conflict simulations with five different AIs and found that they tended to escalate war, sometimes out of nowhere, and even use nuclear weapons.
  • The AIs were large language models (LLMs) like GPT-4, GPT 3.5, Claude 2.0, Llama-2-Chat, and GPT-4-Base, which are being explored by the U.S. military and defense contractors for decision-making.
  • The researchers invented fake countries with different military levels, concerns, and histories and asked the AIs to act as their leaders.
  • The AIs showed signs of sudden and hard-to-predict escalations, arms-race dynamics, and worrying justifications for violent actions.
  • The study casts doubt on the rush to deploy LLMs in the military and diplomatic domains, and calls for more research on their risks and limitations.
  • gapbetweenus@feddit.de
    link
    fedilink
    English
    arrow-up
    0
    arrow-down
    2
    ·
    edit-2
    9 months ago

    I think it’s reasonable for military to try out any new technology for any kinds of benefits. I mean we tried out if LSD would make better soilders - LLM for simulations seems not that farfatched.

    • Jtotheb@lemmy.world
      link
      fedilink
      English
      arrow-up
      4
      ·
      9 months ago

      To be clear, just because the LSD experiments happened does not make them reasonable. It sounds like you’re justifying future terrible mistakes based on past terrible mistakes that you learn about in a fairly neutral and sanitized way in school.

      • gapbetweenus@feddit.de
        link
        fedilink
        English
        arrow-up
        0
        arrow-down
        3
        ·
        9 months ago

        No, military will just try out everything if there is a slightest possibility of benefit in war. If you have the resources why wouldn’t you? There are literally no downsides.

      • gapbetweenus@feddit.de
        link
        fedilink
        English
        arrow-up
        1
        arrow-down
        4
        ·
        9 months ago

        What do you mean? Military found out that those things are rather useless - that’s something. Also good to know. In 50 years or so we will learn what fucked up things military is doing now.

        The only way to prevent such things is drastically cut military budget.