AI assistant tackles cosmology's trickiest math ⚡ экспресс
Large language models, equipped with computer algebra systems (programs for symbolic computation), can handle algorithmic tasks from theoretical physics. Researchers hooked up the Claude model to the Maple environment and tested it on calculations of cosmological perturbations in modified gravity theories. It turns out that with solved examples at hand, a modern language model successfully tackles most tests, though it makes typical correctable mistakes. This work reveals both the promise and limits of this symbiosis—like training an intern who learns from existing solutions.
The Universe's expansion is accelerating, and maybe gravity doesn't work the way we think. Physicists come up with alternative theories, but to test them, they need to calculate how ripples after the Big Bang grew into galaxies. Such calculations involve long equations with hundreds of terms, where it's easy to get lost. Georges Lemaître and Edwin Hubble laid the groundwork, but now a machine takes over the grunt work.
The neural network Claude, hooked up to the math package Maple, works like a robot chef: you give it a recipe (the initial theory), it measures out ingredients (simplifies formulas), mixes them (substitutes expressions), and bakes up predictions. The physicist just has to taste-test the result—compare it with observations. On the problem of cosmic perturbations, after a couple of examples, the AI handled most new cases, but sometimes slipped up—flipping signs or dropping terms.
Such an assistant could speed up the hunt for the secrets of dark energy many times over—or even show how the expansion can happen without it. For now, human oversight is still essential, but it's clear: the future of cosmology demands a symbiosis of living and machine intelligence.
🎯 The first calculations of cosmic ripples were done with pencil and paper. One arithmetic slip, and a discovery could be delayed by a decade.