EFF sued the government back in March for information about the Wasteful and Inappropriate Service Reduction (WISeR) model, a new Medicare program that uses AI to evaluate prior authorization requests for certain medical services… The documents (available here) show that WISeR has resulted in widespread delays and denials of care, operational chaos, and reports of patient harm.
Just going to leave this here.
Testing in prod, where prod may be life and death medical decisions, seems… unwise.
I don’t know about hating AI for this. Prior authorization is pretty straight-forward and rules-based. It’s actually not a bad use case in theory. In practice, it’s probably more like, “oh, great, we can deny all these claims and dodge accountability by blaming AI, while we pay Trump’s buddies for the service.”
If it’s rule based, you can use a deterministic program that only let a human intervene in controversial cases, instead of a random one that fires whatever it wants without supervision
The whole authorization process is a boondoggle.
Listening to Corey Doctorow on Adam Conover talk about the difference between AI working for humans (good) and humans working for AI that is working for captial (bad). The example he gave was AI helping an individual radiologists (good) and AI replacing nine radiologist so the tenth one could sign off on an AI diagnsosis every two minutes and that doctor only being in the loop as a liability sponge, so they could be held legally liable for mistakes the AI made.
The problems we have with AI largely boil down to problems we have with capitalism, particularly the oligopolic landscape we are currently inhabiting.
There is the background issue of how most of the AI we’re using got and is being developed, but your point does fall back down to the idea that it’s often not the tool, but how we’re using it. Or another way to say it, the human is always the problem.
And it would be politically a little more savvy to understand who our real enemies are, and it’s not math professors using AI to solve problems or a programmer using AI to dig through code to find a flaw they know how to fix… it’s profit-maxxing CEOs–and those people are a threat to life itself even without AI.
