Hot take: some of the most unexpectedly valuable data for AI may be all the academic and scientific studies that either failed or were submitted to journals and rejected. AI can be applied to reevaluate those studies to see if an alternative approaches or data sets could’ve been