This scares the hell out of me. Not only are they going to use our photos for a purpose we never consented to, but I would expect the misidentification rate for Black people to be far higher than whatever is claimed. As someone who has been stopped by the police before due to a ridiculous case of mistaken identity, I can only imagine trying to talk them out of an arrest when they say "computer says this is you".
@airadam the idea that computers never make mistakes is deeply engrained in UK law. However, any objective world, the facts show they frequently make mistakes. The Bridgend nurses case is probably the best documented example, prior to the Horizon scandal at the Post Office, though in the latter no-one was charged with homicide.
See "Computer Bugs in Hospitals: a New Killer" Gresham College lecture by Martyn Thomas & Harold Thimbleby.
@epistatacadam@airadam Computers are programmed by people. People make mistakes. Even AI they are trained by people. The smallest unit of code is a line of code with a bug in it.
@ariaflame@epistatacadam absolutely, very much my specialist subject! Even in the first year of my degree, the Computer Ethics module illustrated several horrible outcomes from computers "going wrong" (requirement/spec/programming errors, combined with carelessness) which are well-known in the field, but less so to the public at large. The average citizen or police officer won't believe that something as seemingly sophisticated as auto facial recognition could be wrong...
@ariaflame@epistatacadam ...especially if the person flagged is from an unpopular group. The breathless propaganda from the AI industry around things like this won't help!
I agree, you only need to look at the UK post office scandle to understand computer systems make errrors or rather have faults in the system that produce errors, which lead to serious consequences.a\
Add comment