True that the software might well differ. This thread is interesting in that ‘facial recognition AI’ is discussed (and treated) as if it’s just one facial recognition software. Whereas as mentioned there’s a multitude or plethora of different AI systems being developed including B2B vs B2C.
As training data is expensive and training an AI model vastly so, different systems use different individual image training data sets. For example, one of the most significant training sets in the history of AI was ImageNet, but which was highly criticised for its sexist, homophobia, and racist labelling of images.
The bottom line might be for Jenny or others to ask the software provider which image data training set was used to develop their software as many have subsequently been discredited.
@SugarandSpiceandAllThingsNaice many many police forces are relying on AI though that might not being reported as much as it should.
@AmaryllisNightAndDay I hadn’t noticed that about Joy differentiating ‘gender’ from sex, but she is in academia in the US… I like Kate Crawford’s work but she repeatedly uses the word ‘gender’ when I think (hope that) she means sex?