One thing AI for biology can do TODAY is save domain experts from wasting time retreading paths already walked in other domains. A lab focused on extracellular matrix finds a signaling pathway... that has already gone into clinical trials by inflammation experts. Cardiologists start talking about senescent cells, using markers the aging field dropped years ago. The aging field makes "biomarkers" that have none of the traits pharma or FDA require for use in humans. The wisdom of ages is just a short prompt away, and we can speed up science before automated labs just by making sure each experiment matters.