AI is mastering the laws of physics by playing in simulators rather than reading textbooks.
By training on synthetic data from physics engines, models improved IPhO scores by 5-10%. This suggests high-level reasoning can be grounded in simulated physical reality, bypassing the need for scarce human-curated datasets.
Solving Physics Olympiad via Reinforcement Learning on Physics Simulators
arXiv · 2604.11805
We have witnessed remarkable advances in LLM reasoning capabilities with the advent of DeepSeek-R1. However, much of this progress has been fueled by the abundance of internet question-answer (QA) pairs, a major bottleneck going forward, since such data is limited in scale and concentrated mainly in domains like mathematics. In contrast, other sciences such as physics lack large-scale QA datasets to effectively train reasoning-capable models. In this work, we show that physics simulators can ser