Hacker News Reader

Comment by denton-scratch | Hacker News Reader

As far as I can see, this isn't news; medical insurers (and really any kind of insurers) use rules (algorithms) to decide which claims they should pay out on. Even if they could evaluate each claim completely individually and personally, they'd get sued for not treating all claimants equally. They have to use rules.

And saying it's "AI" when it's just an algorithm (or even just a checklist) is getting annoying.

I started getting annoyed when game developers called the algorithms used to control NPCs and so on "AI", when it's actually hand-made rules, finely tuned. Nowadays, if it's been anywhere near automation, then it's OK to call it "AI". A calculator could be called "AI" (if they still made them).

View on HN

> They have to use rules.

There are rules and then there are rules. It is the details of the particular rule that is of interest. The use of "AI" in that article seems like clickbait to me.

The article seems to suggest that predictions based on aggregate statistics are being applied to implement strict coverage approval for individuals. That seems like a complete misunderstanding of predictive value of those statistics for a particular individual. Not sure if that is gross incompetance or malfeasance. Either way, a problem.

Whenever money is involved, invert Hanlon's razor. High prior on malice, low prior on ineptutide.

Some well-meaning team of data scientists might have produced this model originally. But its employment in this manner cannot be assumed to be anything other than malfeasance, except in the face of significant evidence to the contrary.

I think it's just a simple profit motive. If the insurers can minimize the amount of care while also minimizing the risk of getting sued, why wouldn't they do that? The algorithm/rules can be tweaked to minimize costs while protecting themselves from lawsuits when the algorithm "accidentally" drops coverage for someone who still needs it, as described by the examples in the article.

That sounds like malfeasance to me!

The news here is that insurers have a new tool for excusing themselves of responsibility, and might start to wield it in order to deny even more claims than before.

Is there any system on the whole planet, where the entities responsible for the payment are willing to just write a blank check no matter what? I'm pretty sure that there's always going to be some kind of cost-benefit balancing, probably based on QALYs and such.

I think the difference is in who makes the decision about how and when to set limits. Insurance companies' purpose is to maximize profits. A public owned service might be willing to sacrifice some costs for better overall health outcomes.

Reply on news.ycombinator.com