AI x physics is endless rabbit hole: You can study AI using methods from physics, you can study physics using AI models, you can try to make AI systems model physics as accurately as possible through physics biases, you can design better AI architectures using physics, many AI architectures are applied physics, etc. https://x.com/sirui_lu97/status/2072685366453141961 Our book is out this month. Generative AI and Stochastic Thermodynamics: A Tale of Free Energies wondering how good this connection is more broad ai x physics connections: principles of deep learning theory (https://deeplearningtheory.com/) "learning mechanics https://arxiv.org/abs/2604.21691 Lenka Zdeborová - Statistical Physics of Machine Learning https://www.youtube.com/watch?v=TLHYwbrhGJc Physics has a broad toolkit for modeling systems. Artificial neural networks can be analyzed as dynamical or statistical physics like systems, where many of those tools apply. In training dynamics, how representations form, loss landscape geometry, generalization, etc. And lots of architectures are based on physics, like Hopfield networks. https://youtu.be/gT07OoBOPNo?is=YR688j6BKoYXCeXL https://youtu.be/dRkehLL19Wo?is=F5gaiPSEv9mzv6Kr https://youtu.be/m2bXL5Z5CBM?is=7n8qdfxM4AGWg354 Or also for example diffusion models use concepts from non equilibrium thermodynamics And flow matching is learning vector fields more broadly, which are everywhere in physics Entropy in information theory has a certain correspondence to entropy in statistical mechanics Schrodinger bridges are even more general than flow matching https://arxiv.org/abs/2603.18992 "At the core of modern generative modeling frameworks, including diffusion models, score-based models, and flow matching, is the task of transforming a simple prior distribution into a complex target distribution through stochastic paths in probability space. Schrödinger bridges provide a unifying principle underlying these approaches, framing the problem as determining an optimal stochastic bridge between marginal distribution constraints with minimal-entropy deviations from a pre-defined reference process." Schrödinger Bridges are called Schrödinger Bridges because Erwin Schrödinger himself, one of the fathers of quantum physics, posed, and to a large extent solved in 1931/32 the problem of finding the most likely random evolution between two continuous probability distributions. https://arxiv.org/abs/1806.01364 The lovely math of diffusion models and flow matching visualized https://youtu.be/iv-5mZ_9CPY?is=gVK-wajGAqHShpO- A lot of AI is standing on the shoulders of physics giants But not just physics, that is one influence, but a big one I want to do AI for physics, and I'm not sure LLMs, now reasoning large multimodal models (RLMM), trained on the internet, is the way to go. " https://schema-harness.github.io/ "Today, we’re introducing [schema]: a harness reaching 99% RHAE with Opus 4.8 + Fable 5 and 95.35% with GPT-5.6 Sol on ARC-AGI-3 Public set. ︀︀ [schema] makes an LLM think like a physicist." Observe, code it up, try to find counterexamples in the real world, update simulation, code a solver for the simulation, try it and loop until it works. The model observes the world, makes a simulation of it, then tries beating it, if the simulation mispredicts it tries correcting it or gathering evidence that could falsify it and tries again. Attempts at replicating parts of scientific methods works! Also I wonder to what degree can you use the trajectories (data) in this harness (that uses llms to generate symbolic programs) for fine-tuning the agents. And how can you turn this into scientific method RL env more properly. Or design more deeply neurosymbolic architectures. I'm thinking of extending it with more possible aspects of the scientific method. In empirical sciences, I believe automated approximations of parts of scientific methods will require loops in this style more: "define domain -> define question in the domain -> gather existing information, data, models related to it -> formulate a hypothesis based on it -> gather more empirical data by experiments to test the hypothesis -> analyze, interpret data, draw conclusions -> refine hypothesis according to that -> test with experiment -> refine with theory -> test -> refine -> replicate ->..." Also since some scientists like Galileo basically broke the scientific method, and still found novel predictive and explanatory models, then I do wonder if some of that can also be put into an algoritm, and get useful results in scientific discovery [https://youtu.be/v7a65AvELdU](https://youtu.be/v7a65AvELdU?is=2xplwF2mnfx__an3) Damn this makes me excited. Automating science is probably the biggest thing for me. I think actual open problems solved in formal and empirical sciences is the real benchmark. The amount of them, and their difficulty. From simplest Erdos problems to stuff like solving Riemann hypothesis to somehow solving quantum gravity or just stuff like incremental better differential equations that model some system. The difficulty of open problems are on a spectrum. But on the top of the mountain, I'm gonna be the most impressed when an AI system builds novel more out of distribution mathematical mountain/s of languages and abstractions to solve a problem in a completely different domain. Analogous to modern algebraic geometry by Grotenderick, or modularity theorem for fermat's last theorem, or coming up with quantum mechanics/relativity. For all of this you need to more properly automate science. But a lot of it is also hard to verify, so RL isn't sufficient. And Goodheart's law is creeping up everywhere. Big part of the puzzle is probably also discovery in domains without clear goals, without clear objectives, without clear metrics, that are good for hillclimbing with reinforcement learning. I like approaches trying to attack this problem space with evolutionary methods, divergent search, novelty search, etc. that Stanley et al propose. But there's still a lot to be done here. I do think some future automated scientist architecture will merge Cholletian neurosymbolic program synthesis in Popperian/Kuhnian/Quineian/Feyerabendian scientific method loop, with Stanleyian/Clunian divergent open-ended search that never stops accumulating information, and Suttonian reinforcement learning. And mixing these approaches, depending on the task. Sometimes you want to climb clear mountains of verification signals or mountains of abstractions. Sometimes you want to divergently explore the novel unknown without any concrete goal in mind. Sometimes you want neural/statistical, in big part black box, models. Sometimes you want symbolic white box models. Sometimes hybrids. Sometimes symbolic models are too rigid, but sometimes they predict better, are simpler, have better explanatory power, and generalize and adapt better (lots of physics). Sometimes neural models are too brittle, but sometimes they predict better, are simpler, have better explanatory power (sometimes after reverse engineering maybe), and generalize and adapt better (classifying certain images). Sometimes something combining those in the middle gets you to better local minima (some physics models). "