Just ask GPT3 how many papers resulted in proving the negation of the hypothesis, and how many papers resulted in reasonably proving the hypothesis. And maybe inconclusive results for sake of completeness. It would be interesting to see the numbers, especially in the field of psychology.