With increased usage of AI, we can expect more and more automated decisions being taken, anywhere and across most industries.
The risk with this approach is that you need to make sure that your model of the world is close enough to reality, or else you might have some unexpected results!
I took some creative liberty here by retrofitting my story to this video, but I think you will get my point. Imagine you want ideal grass conditions on Wimbledon courts so you want to properly water the courts. Imagine you take measurements and collect data for a few months when the courts are not in use and observe how grass is growing... and then using that data you automate the watering. In this simple model you would not have included the impact of games on the watering system, right?
Now, this is a just a creative story, but imagine potential impacts in the real world by over simplistic modeling...