@snowboat84: To add a supplementary note, regarding the phenomena emerging from AI—scaling laws, emergence, double descent, representation geometry—the papers discussing them are already numerous. But there is a big problem: they are all thinking in the way of computer scientists, not physicists. What is a computer sci…

To add a further explanation: the phenomena that have emerged in AI—scaling laws, emergence, double descent, representational geometry—have already generated a mountain of papers. But there is a big problem: they are all thinking in the way of computer scientists, not in the way of physicists.

What is the computer scientist’s way of thinking? See a phenomenon, pick a set of mathematical tools, run a small experiment, try to piece together an explanation—maybe you catch something, maybe you don’t. Then switch tools, switch data, and try again. Ask them what a necessary and sufficient condition is, and they don’t care. This is not science; this is modeling.

The physicist’s approach is completely different. From a mess of phenomena, you distill what you believe to be the most fundamental set, idealize it, and construct an isolated, almost nonexistent but conceptually clear “ideal system.” Then you thoroughly understand the laws within this ideal system, and only step by step do you expand to the real world.

Newton’s first law was built this way: “An object not subject to any external force remains in uniform rectilinear motion.” But this law does not hold just anywhere. It does not hold on Earth because Earth rotates; it does not hold in the solar system because the entire solar system orbits the center of the Milky Way; if you are running, sitting on a swivel chair, or riding in an accelerating car, none of those are good reference frames because they are all accelerating. Newton’s approach was: only relative to the infinite distant stars does an “absolute reference frame” exist, and the law holds strictly there. This is an ideal state suspended in midair—unattainable in reality, but conceptually necessary. All real mechanical problems are “deviations from this ideal state.” Gravity, centrifugal force, Coriolis force, friction—all arise because the reference frame we are in is not an inertial frame.

Without this ideal state suspended in midair, classical mechanics could never be built.

That is why I introduced the concept of Cyber Space. Only by placing AI phenomena in the independent world of Cyber Space can we ask questions in the physicist’s way: What is the “inertial reference frame” of Cyber Space? What is its “ideal gas”? What is its ideal state—the starting point that almost does not exist in that Space, but must first be established conceptually?

No one is seriously asking this question right now. But only by asking it can new mathematics and new theories find a point of departure. Otherwise, we will just keep watching AI papers pile up like a mountain, each one “kind of explains something,” yet we can never build a new foundational theory.

Exactly right.

Haha, a bit dry, isn’t it?

Thank you for your summary.