The more problems an agent can solve, the higher the probability of natural selection.
Isn't this the process by which humans have become general-purpose intelligence by solving many problems in nature?
Isn't this article I wrote similar to this study?
they said "The agent playing a variety of test tasks. The agent was trained across a vast variety of games and as a result is able to generalise to test games never seen before in training."
I think this is similar to my argument