Capitalism is a good way to solve the allocation of ressources to increase production output.

But unstructured problems where answers can both be equally wrong or right and where the meaning is inherently difficult to rate without human input, can easily be solved BY ADDING human input, which is what OpenAI is doing now.

Humans alone are often not very creative, but humans as a collective of individuals able to learn from one another are immensly creative see as was seen during the demo phase. I alone would never had guessed that it could be used to emulate a terminal or a linux program or that it can create different answers by assuming different persona.