Can ChatGPT be used to generate scientific hypotheses?
This addresses the problem of automating creative scientific tasks for researchers, though it is incremental as it builds on existing AI capabilities.
The study explored whether large language models like ChatGPT can generate scientific hypotheses, finding that while error rates are high, they can effectively structure scientific knowledge to produce testable hypotheses.
We investigate whether large language models can perform the creative hypothesis generation that human researchers regularly do. While the error rate is high, generative AI seems to be able to effectively structure vast amounts of scientific knowledge and provide interesting and testable hypotheses. The future scientific enterprise may include synergistic efforts with a swarm of "hypothesis machines", challenged by automated experimentation and adversarial peer reviews.