Hacker News new | ask | show | jobs
by santoshalper 100 days ago
I still wouldn't let AI off the hook here. Every link in the chain has to be accountable for fuckups. You don't get to pass it along to the supposed "human in the loop" when you fail spectacularly. That's how we end up with shitty "almost works" AI.
1 comments
mft_ 100 days ago
Sure, the AI contributed, but it was far less responsible overall than the humans in this case.

Don't let the AI system off the hook by all means, but by focusing on it to this extent, the narrative ignores (deliberately?) the hugely negligent actions of the police et al involved.

link
alcomatt 100 days ago
AI or more precisely the way it is being sold to us is the most responsible factor here. People by nature are lazy and will take shortcuts given an opportunity. AI is the ultimate shortcut these days, a "mental crutch" majority of the people using it are leaning on. Humans just did what they always do, be lazy - AI should never have been used for processes with this level of life-altering impact because what happened here was bound to happen.
link
Hizonner 100 days ago
Nobody's "selling it" as more reliable than it is. People are assuming it's more reliable than it is.

> People by nature are lazy and will take shortcuts given an opportunity.

So, um, the fact that humans are behaving incompetently means we should shift the responsibility onto a machine?

Suppose a human had looked at some crappy surveillance video from hundreds of miles away, and told the primary investigator "that looks like it could be her; you might want to check it out". Would that human be the most responsible person in the chain? The moron who took that as gospel and actually made an arrest has no agency at all here?

Come on, a facial recognition match? Facial recognition probably shouldn't be used because it's bad when it works, but everybody with a functioning synapse knows that facial recognition is going to get lots of false hits.

link
jjj123 100 days ago
I agree, but I think the broader point here is that any automated system is a way to offload accountability. And it will be used for that without a doubt no matter how “good” the officers or human processes are.

So it’s still reasonable to be skeptical of (or outright reject) the use of the technology in systems that can ruin or end people’s lives.

link