Tenobrus, a pseudonymous San Francisco-based AI commentator, says parallel AI instances can advance math and science by exploring thousands of hypotheses simultaneously without needing uniform high intelligence

— Academic Daniel Litt predicts an early surge of results then slowdown.