Will future value chains miss links?

There's an infathomably large body of work out there about all the things AI is going to do for us in the future. On one thing there is fairly broad consensus: AI will help us automate the simple tasks so we can focus on higher value ones.

On the one hand, sure, I can see how this will go. But the fact is that there's a function to the simpler tasks in terms of letting us reflect on what we are doing to internalize lessons, abstract up and make the higher order realizations that allow us to make improvements higher up the value chain. I think most specifically about the passage in Tolstoi's journal where he describes the reflective state he gets into when sweeping the floor. Victor Shklovskii grounded his concept of "defamiliarization" in this passage from Tolstoi about sweeping.

One can easily argue that it has ever been thus. Technology and mechanization have taken lower-value add tasks off of our plates and we've subsequently been able to step into a given value chain somewhere much higher along it and begin our work there. Again, I think back to Samuel Pepys walking or waiting around London in 1660 to distribute cash to people, in particular soldiers. It's important for soldiers to get paid, of course, lest they get pissed and go out looting. We can just direct deposit to their bank accounts now.

But might their come a time when so much has been automated that too much of the original skill is lost. As I have written before, my utter dependance on Google Maps to get around disturbs me. I used to get to know the places I lived and would be able to easily navigate. I can still do it in Chapel Hill, but is that only because I lived there earlier in life? Could a lack of analogous knowledge-base formation in the future entirely dislocate important value chains in ways we can't anticipate?