Registration
Past Event
Welcome! To join the event, please register below.
About Event

​Before a frog embryo builds its face, a rough map of that face appears as a voltage pattern across its cells. Scramble a tadpole's face and the parts move back into place by routes they have never taken. Block the electrical links between a cut flatworm's cells for 48 hours and it can regrow with two heads. Cut it again weeks later in plain water and it grows two heads again, with no change to its DNA.

​Michael Levin's lab at Tufts has produced results like these for two decades. Simulators exist, but the field has no agreed formal language. Lately Levin has argued that bodies draw on non-physical "Platonic" patterns, which has set off a loud debate.

​Joel Dietz, a historian of mathematics formerly at the California Institute for Machine Consciousness, is in SF for one week. He turned the seven claims in Levin's latest paper into runnable code, Lean proofs and predictions locked before the data was opened. His score: the math holds up at 88%. The non-physical extra earned 0%, because ordinary local rules already explain every result. He is careful to say that "not needed" is not the same as "false".

​The Talk

  • ​What a predictive language linking voltage patterns to genes and body shape could look like

  • ​The test any "extra ingredient" must pass: beat the best physics-only model on data it hasn't seen

  • ​What failed, including a fish-shape law that held across 10,000+ species and then went 0 for 7 on hornet wings

  • ​What an engineer needs before tissue becomes programmable

​Levin says LLMs are "trapped in the same attractors of thinking" as the fields they learned from. Joel's audit was mostly run by AI agents, refereed blind by models from three labs.

​His answer to Levin is a method: pin every claim to a locked prediction about a real organism, so the failures come from nature and not from the training data.

​Read more: https://metameditations.substack.com/p/auditing-michael-levins-platonism

​About Joel

​Joel Dietz is a systems architect and entrepreneur. He studied at Brown and the University of Pennsylvania. Today he works on Bioelectricity.net, an open hub for developmental bioelectricity that indexes 400+ papers and 80+ researchers.

​Doors at 6. One talk with a long Q&A. food and lively discussion to follow.

​

​Bioelectricity.net | Telegram channel for organizing future collaboration

Location
AGI House SF: 170 St. Germain Ave. San Francisco CA 94114