I see how the automation of those analytical scientist work a huge companies like google, where you probably save quite a few millions on having less employees, but even at our 20.000 employee municipality AI is an expense. Sure we could maybe cut the data/analytical staff down to half the size if we replaced them with data scientists, but data scientists are more expensive and you’d also need to pay a license for whatever ML platform you use, and none of our POCs with various consultants or academics have provided us with anything that was better than what we already have. In fact, In every test, it turned out our own human build models were a lot better at predicting and auditing than ML.

Heh.

AI is also struggling outside of big datas. We’ve run a poc on self driving cars for three years now, and they don’t work, at all. They certainly can work, if we replace and maintain every piece of road infrastructure to their exact needs, but that’s never going to happen.

I guess AI will eventually find a usage outside of advertising and privacy invasion, and don’t get me wrong, we certainly use it for things like trolling through millions of old case files as part of our quality control. So it’s not completely useless, but I don’t see how it’ll change the world anytime soon.