Sometimes I get a really good quote in.
“Pointing out that AI is prone to giving ICE bad information is missing the entire point of ICE,” EFF’s Eva Galperin told Rolling Stone. “They don’t care if the information they have is good.”
https://www.rollingstone.com/culture/culture-commentary/ice-ai-mistakes-1235507183/
@evacide@hachyderm.io Yo! Appreciate the love—right back atcha. I’m moving my updates to text, so hit me up here: 🖕 202 374-3427. Talk soon!
@evacide So much of the treatment of this issue is framed around how the otherwise noble mission of ICE is tainted by “poor training” or “increasingly violent resistance” or “poor information.”
It’s great to see some pushback against the frame. I’d love to see a major media outlet reject it explicitly, in the lede
@evacide ICE don't need or want reasons
their boot lickers only need rationalizations
@evacide reasonable people tend to miss the point about these fascist behaviors. They look for reasonable motives for the crimes, like profit or genuine hatred of a particular group. And while those motives do exist, the REAL motivation is to flex. I can do these things and no one can stop me because I'm powerful. That is always the real reason. It's why Donnie put his hands on his daughter's hips in public. To show that he could do awful things with impunity.
You are accused of the grave transgression of looking like someone!
Defense is futile, the machine has spoken.
You are awaited by the camps!
@evacide However, the inherent unreliability of the LLMs (yes, the LLMs - there's no such thing as AI) will enable ICE to say "it's all the computer's fault!" which is the actual point.
They want to keep murdering people and then shift the blame.
@evacide
ICE not caring if the information they get from their Ai apps is any good reminds me of the W Bush gang's nonchalance about the poor quality information generated by their torture techniques. The cruelty was, and still is, intentional.
They never learn because they don't want to.
#ICE #SurveillanceApps #Ai #Torture
@evacide Maybe, now hear me out on this, what if, someone writes an AI program thing that takes selfies of all ICE Holes and then send that image to all OTHER Ice Holes telling them it's a wanted individual who is armed and dangerous? Just sayin'
Adding more evidence to the hypothesis that the administration wants an ICE goon killed for propaganda purposes and keeps giving them worse and worse ideas until they get that death. "You can kick down whatever door you want without a warrant, don't worry about it" and now "here just use autocomplete to tell you who to target"
@evacide they only care about one thing, comply or die. Ghosts of the confederacy.
@evacide excellently put, and absolutely correct.
@evacide been saying this for years. Ai is only good for military and police who don’t care about veracity and won’t be openly evaluated
@evacide People keep holding ICE to old standards, and they're stunned when ICE doesn't stick to the script.
@evacide true 100 percent
@evacide a point I was hoping to see in this article is that randomness—maybe slightly skewed towards a target—is an effective instrument of terror. It's not just that they don't care and are happy to invent pretexts after the fact, but systems like what they are building, which may result in targeting anyone, serve their purpose *better* than accurate systems that don't make mistakes.
AI as a pretext for doing fascism is basically seems to be most of the point of AI 😟
