But ... I think with AI, if you have an insight that isn't motivated by math per se, but is motivated by some other problem intuition, you can still attempt to implement it, and the proof is sort of in the pudding.
With physics, even the experts can go on developing theory which (as I understand it) arguably might never be testable, or is only testable after some absurdly expensive new machine is planned and built which might take decades. So an armchair physicist with an idea can have no route to falsifying their idea.
With physics, even the experts can go on developing theory which (as I understand it) arguably might never be testable, or is only testable after some absurdly expensive new machine is planned and built which might take decades. So an armchair physicist with an idea can have no route to falsifying their idea.