No. of Recommendations: 15
I considered putting this on the STEM board, but the thought started with Charlie Munger's words:
Munger: “You must know the big ideas in the big disciplines and use them routinely — all of them, not just a few.” (1)
Munger, again: “You've got to have models in your head, and you've got to array your experience — both vicarious and direct — on this latticework of models.” (2)
The journal Science, today (bolding mine): “A physics researcher wouldn’t spend 6 months reading linguistics research in order to figure out how to repurpose linguistics methods...For AI, however, hopping between fields can be a breeze. Researchers highlighted that ability earlier this month with the release of 36 preprints produced with a suite of AI tools called BootLoops. The so-called harness can be used with any large language model (LLM), such as Anthropic’s Claude or OpenAI’s ChatGPT. And in just 3 months, researchers used Claude outfitted with BootLoops to crosspollinate at a massive scale, spotting— and solving—problems in seemingly unrelated fields that share the same mathematical structures. The resulting papers range widely, from figuring out the age of sunspots to attempting to decipher a medieval manuscript.
In essence, BootLoops “kept noticing the same pattern: many fields have problems that a technique from mathematics, physics, or computer science would solve outright if anyone knew it existed,” Matthew Schwartz, a physicist at Harvard University and visiting researcher at Anthropic who developed the tools, wrote in a 1 October blog post unveiling the preprints. (3)
In fairness, today's article goes to some length that this is an immature technology, frequently producing conclusions that are dubious and/or irrelevant. But, still.
--sutton
(now, they just need to teach the LLMs to "invert, always invert")
(1) Stanford Law School, 1996 (Poor Charlie's Almanack) · 1996
(2) A Lesson on Elementary, Worldly Wisdom (1994 USC Marshall talk), Poor Charlie's Almanack · 1994
(3) Celina Zhao Science 22 Oct 2026: 366-367 doi: 10.1126/science.zwu3e54
