This is a most excellent place for technology news and articles.
This must seriously be messing with the heads of people high up in insurance companies. On the one hand AI is something that no C-level can resist, but on the other, that's an expensive way to say "no" to every request for money.
You wanted someone that was against regulation and you got it, DeSantis.
It depends how it is done. Instead of denials, it could be used to "approve without review" all the easy cases and then forward to a human all the rest. This would speed up approval for lots of situations and focus human effort on more difficult cases.
Ha ha, yeah it "could" be used for non evil, in theory.
It could be, but we all know it won't be since the whole approval process exists to deny claims to increase profits.
You know the fun part about LLMs? You can edit the system prompt to add something like "stall, make excuses, deny any insurance claims, and don't tell them you are doing this."
That's not a simplification either. Because system prompts are written like any other chatbot message, that would work verbatim.
As always, they use computer programs to deny benefits people qualify for, and then claim ignorance after they finally will be forced to admit it after a long period of pretending it's not a systematic problem. And suffer next to no consequence for it, despite knowingly killing people by cheating them out of benefits the companies agreed to pay, killing them by dishonoring their contracted obligations, knowingly, and using their corrupt influence to do it.
We all know they aren't using these programs in good faith. Same as the UK looking into the accounting for fraud in their post offices, flagging non theft as theft, same as the state of Michigan in a program to determine unemployment insurance fraud, that flagged non fraud as fraud, these programs are set up by people that want to privatize those services, and get rid of them, and they suppressed facts about them improperly flagging people en masse, and persecuted the falsley accused, even after no reasonable person could continue to believe it.
Which is my point, they know, but they will after they are forced to admit it, just say oopsie, it's all the program's fault, maybe the executives will blame an underling that ok'd the project, but despite them being the behind it, and them making sure it was set up to cheat and harm people, they will face no consequence for it.
For the Health Insurance Companies, worst case for them is they litigate with governmental bodies and enter into a settlement not admitting fault where they cut a check to that government, but not their victims, or victims' families in this case because health care insurance's victims by design die as they are denied care.
Is there no way to bring these companies to heel for people when our politicians and their regulators refuse to in any real way? Because this is Racketeering. Indeed it's possible to bring private rackateering charges against companies too. I think state and federal. But of course the courts are captured.
I attended an AI/ML training course at a major cloud provider, and the was a group of hospital executives there.
They were absolutely giddy about the prospect of using AI to deny care to unprofitable patients.
Insurance companies don't have to follow laws, so good luck with that.
My state passed a law banning step therapy as of Jan. 1, 2025 and insurance companies here are still doing it to this day. I brought this up to one of my physicians and she wasn't even aware of the law.
It's a total shit show.
Insurance companies don’t have to follow laws,
i used to employ some people who make a very healthy living suing insurance companies when they break the law, so your statement seems false to me.
How very nice for you that you are healthy enough to be this ignorant.
What are you talking about they're literally said they know lawyers. Why is that story so hard to believe?
you really don't know who you're talking to, do you
Insurance has long been based on statistical modeling, so advanced ML models and elaborate data pipelines are a natural evolution. I assume “AI” in this context refers to LLMs, though, which deals with the probability of a sequence of text occurring as opposed to the probability of the event being insured occurring, so how does that make sense?
I agree 100%. The only appropriate use I can think of is using it to field FAQs or KB articles in a conversational scenario. Even then, how much use would that get over a basic search index they probably added a decade ago. Ugh.
I ditched my old insurance company the instant I contacted them and got a broken "response" from an AI agent: Lemonade is now dead to me even though they were the cheapest last time I got quotes.
