Post #1782301
2026-04-21 09:56 UTC
RE: https://eigenmagic.net/@NewtonMark/116441914455373558
There is nothing inherently wrong in using AI (or any unreliable system) to help flag potential offenders.
There is everything inherently wrong with having the unreliable system send out fines without a review process.
Note: this would also be true if the photos were reviewed by a person and not double-checked by another.
(Also: the fact that only 4% of fines are challenged is _not_ an indication that they are 96% accurate... and if they were, they'd still need a review process)
See also: Robodebt.
Replies (1)
-
@bananamangodog@aus.social 2026-04-21 11:38
@twasink@aus.social "There is nothing inherently wrong in using AI (or any unreliable system) to help flag potential offenders." I'm not so sure about that. Having worked in places where people have received no or reduced or substandard treatment because of unproven rumours I think there comes a point where even flagging someone introduces biases in people affecting their decisions.