[sobkas]

> Before that happens, the whole world would need to be under continual observation. Then we would have to realize that it always takes more people to observe the people than there are people to observe, and turn the eternal watch over to AI.

> Then the AI wouldn't be saying, "I think it's likely you did this", it would be saying, "Here's the video of you doing this".

Reality of what this video shows will be is much closer to: " we have some video(we are not sure if AI properly reconstructed low quality recording and then matched person accused of crime, example: changing letters in photocopies), does video shows real crime or it only looks like a crime from certain angle, it search data for crimes not for proofs of innocence. After all what stops people behind AI from over representing as criminals people with evil mustaches (consciously or not, doesn't matter), so at start it assigns them a higher score?