One of the most robust findings in labor economics is that labor and capital are long run complements, not substitutes. I would be shocked if AI is an exception to that rule, for software engineers the sheer flood of code that will be generated in the coming years will demand more and more labor to manage.