“Doge believes that existing large language models are incapable of making novel scientific discoveries because they cannot iteratively conduct experiments.”