5 pointsby 1vuio0pswjnm76 hours ago2 comments
  • miladyincontrol6 hours ago
    Negativity bias imo. We all hear about the failures associated or attributed to AI. I say attributed as I'm curious how many whether LLM generated output or not get prescribed away from being any one individual or team's fault.

    Of course we see larp success stories, marketing fluff promising the sky, but genuine realistic success stories in devops relating to AI are rarely exciting to read or write.

    • jqpabc1234 hours ago
      Devops bias imo.

      Devops are somewhat unique in that testing is an expected/normal part of the process. No one (that I am aware of --- yet) is promoting the idea that AI produces fewer bugs or superior design or greater "certainity" in outcomes --- subjective claims that are not easily testable.

      These sort of claims are being made (without any real proof) in other, unsuspecting areas.

  • jqpabc1235 hours ago
    Splunk calls it the reliability paradox

    The real paradox here is that reasonable people continue to drink the AI marketing kool-aid/hype.

    AI is fundamentally built around probability.

    When relying on probability, it should be expected that an element of randomness will be instilled into your operation.

    The only real certainity is that if you continue to roll the dice, you will eventually lose. Probability (aka AI) makes it so. Failure to grasp this points to an educational failure imo.