like darwin selection between the subsystem approaches? Put in a lot of different kind of LLMs and let them play the same game inside with each other.. whosoeve

like darwin selection between the subsystem approaches? Put in a lot of different kind of LLMs and let them play the same game inside with each other.. whosoever wins the simulation is allowed to externally present the approach... something like that?