LLMs Proposing Scientific Method Tweaks

In this conversation with Alexander Unzicker, physicist and author of Bankrupting Physics, we dig into why physical constants like the fine structure constant 137 have the values they do, whether mathematics is being confused for physical explanation, and whether artificial intelligence and large language models can make the conceptual leaps that have historically driven breakthroughs in theoretical physics and cosmology. Unzicker argues these constants represent genuine unsolved mysteries, while we push back: are these numbers arbitrary, or do they fall out of the geometry of nature the way pi falls out of a circle? We also tackle AI consciousness, embodied cognition, the limits of machine learning, and whether AGI would simply inherit a century of elegant and deeply confused mathematics dressed up as physical truth.Alexander's book, A Physicists Guide to the Age of Superintelligence: https://www.amazon.com/dp/B0FLNP13GNPATREON https://www.patreon.com/c/demystifysciPARADOX LOST PRE-SALE: https://buy.stripe.com/7sY7sKdoN5d29eUdYddEs0bHOMEBREW MUSIC - Check out our new album!Hard Copies (Vinyl): FREE SHIPPING https://demystifysci-shop.fourthwall.com/products/vinyl-lp-secretary-of-nature-everything-is-so-good-hereStreaming:https://secretaryofnature.bandcamp.com/album/everything-is-so-good-herePARADIGM DRIFThttps://demystifysci.com/paradigm-drift-show00:00 Go! 00:06:16 AI lacks embodied understanding of reality00:08:30 Why superintelligence would care about physics00:11:12 Physical constants: the unsolved problem00:17:10 What would a real explanation look like?00:20:06 Maxwell as a model for scientific progress00:23:17 Constants only make sense materially00:30:01 Physics demands models, not just equations00:36:03 Civilizational survival needs deeper physics00:39:58 Toddlers outperform AI in spatial tasks00:44:56 Neuroscience over philosophy for understanding intelligence00:48:13 Human creativity versus machine novelty00:55:19 LLMs assemble knowledge, not new ideas01:03:05 AI doesn't learn from experience01:06:17 AI replicates style, can't shift paradigms01:11:37 Human genius: defining unseen problems01:15:04 AI trapped in its training worldview01:19:24 Physics requires initiative, not just data01:23:18 Unverifiable theories become epistemic loops01:27:33 Embodied experience drives human innovation01:35:34 Animal cognition challenges human uniqueness01:39:09 Will plus intelligence equals life01:45:45 Human survival extends into culture and ideas01:53:03 Earth as an emerging superorganism02:01:15 Institutional drift and ethical relativity02:04:44 Hope for human and AI coexistence #Physics #ArtificialIntelligence #Consciousness #ChatGPT #Science #Philosophy #QuantumPhysics #AGI #FutureOfAI #MachineLearning, #physicspodcast, #philosophypodcast, #quantum , #quantumphysics, #quantummechanics MERCH: Rock some DemystifySci gear : https://demystifysci-shop.fourthwall.com/AMAZON: Do your shopping through this link: https://amzn.to/3YyoT98DONATE: https://bit.ly/3wkPqaDSUBSTACK: https://substack.com/@UCqV4_7i9h1_V7hY48eZZSLw@demystifysci RSS: https://anchor.fm/s/2be66934/podcast/rssMAILING LIST: https://bit.ly/3v3kz2S SOCIAL: - Discord: https://discord.gg/MJzKT8CQub- Facebook: https://www.facebook.com/groups/DemystifySci- Instagram: https://www.instagram.com/DemystifySci/- Twitter: https://twitter.com/DemystifySciMUSIC: -Shilo Delay: https://g.co/kgs/oty671