At the edge of every system enhanced with Artificial Intelligence there is a person who:
- Understands the inputs that the AI is getting;
- Understands the information produced by the AI;
- Can associate the results with other information; and
- Makes decisions & takes action that the AI cannot yet take on its own.
Even as the Artificial Intelligence improves, there's still a person using it. Certainly one AI could use another AI, but still, at the end of that chain there's a person receiving the result.
At this edge, the AI levels up and so does the person using it. Put another way, the skill level of the person using the AI goes up as the AI improves. Consider the example of IBM's Deep Blue beating Kasparov. While training the AI, it didn't need to play the world's greatest chess player. It could at first just watch and learn how the game is played, then ingest the history of all the famous games played and all the potential moves possible, then play better and better people until it beats them all.
Eventually, all knowledge work will traverse this path in a way.