A/B testing is a learning tool, not an optimization tool.

In the original will, invention is given the same weight as discovery.

One could also argue that in 1895, applied computer science and information theory would be considered physics.

When Alfred Nobel wrote his will in 1895, there was no computer science or information theory.

One could argue it's closer to mathematics than physics, but if you'd say to him that someone made sand think like a human he might even put it under the medicine category.

It's not clear to me that we have "made sand think like a human" (yet).