Recent content by manuel-sh

you mean gemini LLM? I've tested Claude Opus and chatGPT5, and both make many mistakes. You can see examples here where it messes even the format: https://theoria-dataset.github.io/theoria-dataset/entries.html?entry=clausius_clapeyron_equation.json

hey that's a good reference, thanks! I found many other related threads like https://www.physicsforums.com/threads/why-chatgpt-ai-is-not-reliable.1053808/

Hi, I've been using recently AI to generate physics results derivations and, in general, even frontier models such as GPT5 or Claude Opus are quite bad at that, which is surprising. By bad, I mean they are not able to derive common formulas (say the Black Body radiation, Kepler's law, Lorentz...

Thanks Dale! Indeed! And in fact I am not surprised: most of the existing physics datasets are either compilations of papers, books, wikipedia articles, which are quite unstructured. Ask to frontier LLM to derive for example the black body radiation law, and will do a lot of mistakes until it...

Hi everyone, I’m Manuel. I studied physics originally and now work mainly in AI. Recently I’ve been combining both interests in an open project called TheorIA (Theoria Dataset), which collects structured derivations in theoretical physics, formatted so they can be used for reproducibility...