The hurdles to AI breakthroughs in physics

Two posts offer explanations for what they see as AI's limits in physics. One argues that much practical physics relies on long-established equations and capable numerical methods, while major breakthroughs require insights beyond current human knowledge—and largely beyond current AI's reach. The other claims physics lacks the formal verification available to math and software, preventing reinforcement learning with verifiable rewards and recursive AI improvement in the field. That author says Lanyon AI is rapidly closing the verification gap.