I think any new technology just opens up space for new problems to be solved. Two examples:
We made computers because hand-calculating missile trajectories wasn't the best way to win wars. Now that we had computers, we needed to solve a large number of problems - the human-computer interaction problem, new programming languages at different levels of abstractions, new chips to enable even faster computations, make compute cheaper to democratize the computing power - to name a few, and each of these problems required geniuses and visionaries and creation of entire industries on it's own.
We made web to efficiently transfer and view documents stored on different computers. Now that we had web we needed to solve a large number of problems - web browsers, search engines, servers infrastructures, a billion frameworks, a framework to make the previous framework faster, another framework to make that different framework easy to use. Again each of these problems required geniuses , visionaries and creation of entire industries.
Now there's AI. It'll create even more problems to be solved. We will never run out of problems to solve, and therefore I think problems solvers will never be out of demand. Or is AGI different, because it'll be a problem solver which'll itself solve the problems it creates? How will it do so, because those problems can't possibly be in it's training data lol.