• wuffah
    link
    fedilink
    English
    arrow-up
    14
    arrow-down
    1
    ·
    17 hours ago

    I wonder if AI agents are vibe-coding exploits into production code so that other AI agents can find them in a circular-success loop:

    Agent 1 writes flawed, but working code

    Agent 2 exploits flawed code

    Agent 1 writes more flawed code to fix exploit

    Agent 2 exploits flawed fixed code

    …etc. etc. etc. It’s just Agents jerking each other off in a digital dopamine loop until the bubble pops.

    • marcos
      link
      fedilink
      arrow-up
      4
      ·
      14 hours ago

      The intention is not there, LLMs can’t do it.

      The facts on the ground, yeah, that’s probably happening.

    • Boomer Humor Doomergod
      link
      fedilink
      English
      arrow-up
      5
      ·
      17 hours ago

      In the future humanity is two warring factions fighting over a nuclear hellscape, ruled subconsciously through memes by one or the other agents endlessly hacking and patching.