For the largest health insurer in the US, AI's error rate is like a feature, not a bug.
    • EmergMemeHologram@startrek.websiteEnglish
      0·
      11 个月前

      Nonsense.

      It’s most likely just a ~linear regression~ random forest model trained with the loss function of sum((y - y_hat)^2 * cost)

      I have that basic prediction tools are “AI” And they definitely were in 2019.

    • ellabee@sh.itjust.worksEnglish
      0·
      11 个月前

      I was in medical billing about 20 years ago, specifically working to get ambulance billing paid by United Healthcare, Blue Cross, whatever. at that time I hated united slightly more than the VA. the VA was a year behind on payment, and they sent a lump check with the list of what it covered separate. but at least they kept track and paid.

      we had to take United Healthcare to the insurance commissioner because their process was deny, then lose the claim, then deny for late billing.

      instead of responding to the insurance commissioner or providing the requested docs or anything, they waited it out, paid the fine, paid the specific claims, and continued as usual.

      so yeah. AI working the way they trained it.

  • Jaysyn@kbin.social
    0·
    11 个月前

    I guess we’re going to have a decade or so of these companies using AI as a scapegoat before a lawsuit finally makes them responsible for the AI’s output.

    • Rentlar@lemmy.caEnglish
      0·
      11 个月前

      Yeah. If there was anything I’d want to come from this lawsuit if not the company effectively being tried for murder, it would be the need for insurance companies to disclose the data and reasons behind denials which they claim are proprietary.

    • Chariotwheel@kbin.social
      0·
      11 个月前

      I can do it in one minute with PHP.

      <?php

denyClaim();

?>

      That might actually have a better change to be correct that the UnitedHealth AI,

  • ShaunaTheDead@kbin.social
    0·
    11 个月前

    I read about an early study into AI where they were using it to predict whether the pictured animal was a dog or a wolf. It got really good at detecting wolves and when they analyzed how it was determining whether it was a wolf or not, they found that it wasn’t looking at the animal at all but instead checking if there was a lot of snow on the ground. If there was, it would say it was a wolf, if there wasn’t it would say dog.

    The problem was with the data set used to train the AI. It was doing exactly what it was told. That’s the big problem with AI is that it does exactly what we tell it to do, but people are hilariously bad at describing exactly the result they want down to the absolute finest level of detail.

    • Cethin@lemmy.zipEnglish
      0·
      11 个月前

      I would describe it more as giving the results we’re asking for rather than doing what we tell it to, but that’s a little bit of too much semantics probably. We mostly don’t tell it what to do. We just give it data with some labels and it tries to generate reasons for those labels basically. It’s essentially the issue humans have of “correlation does not equal causation” except with no awareness of this and significantly worse.