For the record if anyone didn’t know, they ran the software against the state politicians. It misidentified 26 (I think) state politicians as existing criminals. Which I see irony in there.
First, the threshold used was the default suggested for one-to-one matching (lower false reject rate, higher false accept rate). It doesn't make sense to use this threshold for search applications, and when the study was reproduced with a more appropriate (higher) threshold there were 0 mis-identifications.
Second, law enforcement use of face recognition doesn't even involve the algorithm making a lights-out identification decision. Instead, the most similar faces are presented to the user in ranked order by similarity (like a search engine). It's a tool for generating investigative leads, often preferable to publishing the face image of a wanted perpetrator on local news.
What makes you think that a LEO would change the defaults? They're not computer scientists, they're people who bought a product to perform a specific function.
It's a tool for generating investigative leads
So were photo books, lie detectors, surveillance cameras, and other tools at first. They became a close enough proxy for "guilt" in the eyes of the users for people to automatically be treated as criminals, even if they'd done nothing wrong.
The analogy would be blaming an anchor manufacturer when the "default anchor" for rowboats doesn't hold an oil tanker.
It's both funny and sad at the same time on multiple levels.
What do you mean "mis"? It correctly identified politicians as criminals. In that, this AI showed itself to be lightyears ahead of the electorate in terms of understanding how politics works.
Either the politicians are criminals, and this tool was only able to correctly categorize 26 of them.
Or the politicians are not criminals, and this tool erroneously categorized 26 of them as such.
And that's based only on what we know. The experiment may have given rise in state leaders to more fundamental objections.
Consider, what if the 26 misidentified politicians, were largely black or hispanic males?
We just need to know more about what's been happening in all the testing to say whether these systems are useful or not. That's all I'm saying.
The hard part is figuring out which ones are the criminals and which ones are politicians doing their thing.