When we warn the real threat of AI is how it’s used against people in the present, not the fantasies that some day computers might think for themselves, this is exactly the kind of thing we’re talking about: health insurers using AI to deny care.
arstechnica.com/health/2023/...
I do this for a living and when people ask me if I’m afraid of the robots I say, no, I’m afraid of the people who own the robots. Same as it ever fucking was.
I somehow managed to escape Silicon Valley without developing a case of the eschatological brainworms that apparently have taken up residence in ~all of my peers’ heads. Technology isn’t teleology, y’all! It’s human activity subject to human political control!
I'm with you on that one. I've worked in AI research, and this is precisely the things that are worrisome. It's the people.
And now we have people trying to turn it into a pseudo-religion. It's deeply strange.
Hopefully we won’t have radroaches and murder robots for a while, but drone warfare is concerning. Wait til people realize how deadly swarms of tiny drones can be and we all have to live underground in separate bunkers, communicating via styrofoam cups and string.
Oh we're already on the verge of cults appearing.
There was a famous study on using a super basic "AI" that would reword what people said into it as a question. It had nothing on what we have now. The head researcher was horrified their secretary was secretly using it as their psychologist.
Exactly right. We need legislation drafted by people who actually understand the dangers that blindly trusting (racist, classist, etc) algorithms can pose, and aren't motivated by profit.
It shocks me that one of my friends, who has also worked on machine learning, is absolutely terrified of evil AI world domination... It's not magic! And I've had them explain it step-by-step. There's so many assumptions, like that AI would decide that bankrupting and murdering people will help it.
I'm lucky that most of my peers were informed that ML leans on data/human biases *along with* making errors. But wouldn't that also apply to an AI's morality? And in a scenario that a kill-directed intelligent AI is even possible, wouldn't we also have technology for benevolent AI antiviruses?
like i'm fond of saying, the AI apocalypse won't be SkyNet waking up and siccing T-800s on us, it'll be some asshole exec/politician fiddling with the settings on the CopBots and setting their enforcement protocols to "fire indiscriminately on anyone with a complexion darker than bleached chalk"
I guess this is another version of the "Guns don't kill people, people kill people!" defense that gets thrown around by sociopaths when their technology inevitably murders a human. Not looking forward to this new breed of murder apologists.
I think it's (currently) more analogous to nitrogen uses than guns because of low access + high skill needs. Something where you have to be negligent or malicious for catastrophy, but accidents happen. ML has promising uses in the medical and engineering fields, and for highlighting biases (ex. GIS)
Seeing United Health doing their thing, it's time again. My mother had terminal pancreatic cancer, was in hospice, and would be dead within 2 weeks when United Healthcare sent her a letter threatening to cancel her policy if she didn't get her yearly mammogram. Fuxk everyone who works for them.
I personally feel it's time for widespread John Q (or similar) events, and demand healthcare in the USA be fixed for the betterment of all (except the insurance companies)
I think it's helpful to remember people have been banging the drum about algorithmic reoffense assessment in criminal sentencing for a while. Always a good place for test balloons for stuff like this. With an algorithm people can wash their hands and point at that for their shittiness.
It's not even about speed, it's about deflecting blame by having a black-box make the decision. No one decided to deny care, it's just *the machine* says we can't help you.
“The machine said so” makes it harder to appeal. Instinctively people know they can question a person’s judgment, even if the person is an expert, but we give (unearned) deference to computers. Also helps that without a person involved there’s no one for the beleaguered account rep to email about it
As someone UHC has tried a few times to convince that I should replace an opthalmalogic immunosuppressive with just an overpriced generic Visine.... no, no shit was ever given by UHC execs about how shitty they are
The "existential threat" - LOL - is what my colleague @dcbrock.bsky.social calls a "wishful worry...something that only distracts us from the true villainy and dumbfuckery that is afoot.
Most state insurance regulators are prepared to tolerate quite a bit of bad faith claim denials but 90% is probably uh abusing the courtesy. Remember insurers’ main customers are other businesses, who have actual sway with state governments
Yeah, well...
Doesn't require an "AI" in any form to blanket refuse claims...
That's just a company policy.
The "AI" part of this gives the company something to blame it on and a pretense of plausible deniability of any liability.
