If someone is using machine learning to advertise to me on the web and they come up with some crazy ad they think I might like, no harm done. Using the "probabi

If someone is using machine learning to advertise to me on the web and they come up with some crazy ad they think I might like, no harm done. Using the "probabilities" they may be making the ads marginally better. But if someone uses this for police actions, that is a different story. I expect a little more certainty for something like police actions.