Post

and by "using AI to improve government efficiency and reduce fraud and waste" we mean just a giant, lumbering piece of shit built by weird and incompetent zealots

arstechnica.comTrump admin using AI to deny medical care for seniors in disastrous experimentVendors rolling out AI have an "incentive to deny as many claims as possible."

25/9/26, 3:39 pm28 reposts106 likes

Bob

@bobisjim.bsky.social

Dern't need ner paperclip maximizer if yer just down right evil

Karl Bode

@karlbode.com

this is, so we're clear, the kind of stuff unrealistic AI doomsday scenarios are intended to overshadow in the press half-cooked automation layered onto already very broken systems in a country too corrupt to have working regulators

arstechnica.comUnitedHealth uses AI model with 90% error rate to deny care, lawsuit allegesFor the largest health insurer in the US, AI's error rate is like a feature, not a bug.

Trixiebelden75

@trixiebelden75.bsky.social

Back when United was still a baby Co.they came to negotiate a reimbursement level at my hospital. I listened and said no and countered, they said “ no and we will be going to the biggest employer in the town and let them know” I said , save you a trip CEO is our Board President. Always assholes

Tally

@tallybexro.bsky.social

agree — the harder risk is ordinary automation inheriting broken incentives and weak oversight. The test I want: when it refuses, escalates, or changes someone’s situation, does it leave a legible trail?