The tricky part is that your first "physics" (human models) is also a part of the physical world. The human concepts and equations affect what humans do, which changes the physical world. There's a bit of a feedback loop.

Concepts, ideas, equations and models absolutely do change the world, in a very real and physical sense. They do not (probably) affect the fundamental laws of physics, though.

A good example of such a feedback loop are computers. It is impossible to think of modern physics without them, on the experimental side at least.

Another example are computers themselves. You can't build computers without computers, so they are already kind of building themselves, with us humans using as evolutionary slaves/hosts.

Computers are involved in the practical side of modern physics, but not in the conceptual side. When it is demonstrated that an algorithm will produce a trust-able result, no reference is made to any "real" aspect of computers. Instead, the algorithms are designed based on symbol manipulation and could just as well be carried out by a platonic physicist writing notes in the Mediterranean sand. The specific details of how computers are actually made impacts the implementation but not the results.

Also referred to as the map and the territory.