It's been a few years since I read http://lesswrong.com/lw/qj/einsteins_speed/ and the rest of the quantum physics sequence, but I recently learned about the company Nutonian, http://www.nutonian.com/. Basically it's a narrow AI system that looks at unstructured data and tries out billions of models to fit it, favoring those that use simpler math. They apply it to all sorts of fields, but that includes physics. It can't find Newton's laws from three frames of a falling apple, but it did find the Hamiltonian of a double pendulum given its motion data after a few hours of processing: http://phys.org/news/2009-12-eureqa-robot-scientist-video.html

New Comment
1 comment, sorted by Click to highlight new comments since: Today at 11:13 AM

So am I correct in inferring that this program looks for any mathematical correlations in the data, and returns the simplest and most consistent ones?

New to LessWrong?