Maybe that's a sign that incentive structures are the most important (meta) problem to work on.
Maybe the most important problem of all is to make it trivially easy for people to solve the problem of this thread. (And not only easy, but lucrative.) Once that mechanism is in action, you'd have done 100x more good for the world than you'd have done choosing to work, say, in the energy sector instead of AI.
