Melanie Mitchell is a professor at the Santa Fe Institute and a leading thinker on artificial intelligence, analogy, and abstraction. She reflects on how analogy quietly drives creativity and scientific discovery even in the most rigorous fields. Analogies often emerge during moments of mental rest and don’t need to be accurate to nudge you into new avenues of thinking. We discuss how many core scientific concepts began as metaphors, how analogies can both illuminate and mislead, and whether large language models truly grasp abstraction. The conversation ranges from the role of analogies in Einstein’s thought process to evolutionary “landscapes” and the balance between generative night science and critical day science.

The Night Science Podcast is produced by the Night Science Institute. For more information on Night Science, visit night-science.org .