Less experience means you don't have to pay that person as much. You don't get the same quality of work out of them—and crucially, you never will. Unless you're using an LLM to help you learn as a specific goal, you won't learn much from prompting it to solve your problems for you.
So basically you have a workforce that never gets better, no matter how many hours they put into the work, so you can continue to pay them poorly FOREVER. They're just prompt-generating meat-sacks. I've argued for years that there's no such thing as 'unskilled' labour; fast-food workers, farm workers, manual laborers all learn skills and are meaningfully better at their jobs as time goes on. But AI workers? It's getting pretty close to being unskilled. If there's any differentiation between the results one person gets vs. another, AI companies will roll those into the model to homogenize the results.
That said, I don't think this future is going to happen. I don't think we can underestimate the value of human work just yet.