It’s better at spotting patterns, but without knowing the why and how the machine arrived at its conclusions, I think it’s going to be fairly limited in application.

Your detective example is just as likely to reveal our biases as it’s going to find a clue. It’s all in the training, and we can’t eliminate our own cultural biases. What’s the data set for suspicious behavior consist of? Who chooses it?

There’s a ton of AI hype and it’s barely had any impact in actual real life despite billions of dollars invested. I’m still not convinced. It was very cool for a few weeks until the limitations became very obvious to me.