String theory PhD from Oxford, Maxime Gabella has a vision for the role of ML in theoretical physics.
“There's a long-standing suspicion that space and time are not fundamental, but emerge from something else.”
It's counter-intuitive to try to imagine what this something else may be, but Maxime has an idea.
"Learning may well be the most fundamental process in the universe.”
(He cites his motivation, arXiv:1709.01223)
QM calculates probabilities of observer measurements (i.e. learning), so he decided to study the “laws of learning” in the form of a start up 'MAGMA Learning'.
He developed an AI tutor, Ari, which learns how to teach you. It constructs a 'Learnet' to visualise learned against unknown knowledge - literally in a 3D universe.
He understands that things are learned in relation to one another; he uses colour closeness to cram a further 3 dimensions into the Learnet, as if to lull the user into learning even more!
We discuss how people have unique Learnet signatures and play with the idea of "optimally complementary collaborators” - people matched up to form dream teams.
As you learn more, your Learnet evolves as a growing galaxy.
Just like gravity is the curvature of spacetime by mass, learning is the curvature of reality by knowledge.
Interesting comment from Sione Palu:
"Space and time are relational concepts that emerge from existence. That is, once matter exists then these relational concepts make sense. These concepts are not objects or substance of some kind. They are simply relations that describe the motions of object A to B or C, etc,... Given a Universe that voids of objects, then the concepts of space and time are meaningless. Physicists have reified these non-material relational concepts into kind of material which is absurd. Space-Time curvature? Does 2 minutes curve? How about 3 seconds? If time is curve, then curve of what? I looked at my wristwatch 10 minutes ago and I didn't see a curved 10 minutes."