If we assume AI is a tool to solve problems and it's more prevalent as the years go by, we are not going to keep up with the whole new research, because specialists in many different industries will create models for specific purposes. We can try to be generalists, but just one human doesn't have enough time to consume the content generated by thousands. In the end, we can try to learn general models, have a superficial understanding of promising new techniques, and learn about a handful of specific problems to be a specialist in those.