There can be macro objects with non-determinism. The cat in the box in the Shrodinger experiment has deterministic neurons but you don't actually know the state

There can be macro objects with non-determinism. The cat in the box in the Shrodinger experiment has deterministic neurons but you don't actually know the state of the cat as a whole due to the initial quantum event that set the entirity of the cat on one path or another. This is where the entire problem with the thought that systems are deterministic or not comes into play. If there's a single photon that can hit a pixel on a sensor (or retina) or not and that ultimately has a cascading effect to make that larger object do something completely different or not then that larger object is non-deterministic as a whole. Yes as others have stated there's ways in which non-deterministic inputs can end up not mattering - the system has the same output. But there's also systems, eg. the cat in the box experiment, where the macro object ends up taking on the non-determinism of a single quantum event. An AI that can behaive one way or the other based on what it read into the sensor input is absolutely as non-deterministic as that input. Likewise humans are as non-deterministic as their inputs. If a photon can hit your eye today and cause a flash of light that sets you on the path of doing something you otherwise wouldn't have done it doesn't matter if your neurons are essentially a machine or not. You are still non-deterministic. The focus on neurons being deterministic or not kind of misses the bigger picture. We absolutely live in a non-deterministic world and those inputs from this world cause ourselves to ultimately be non-deterministic. This really gets down to the regularly discussed thought experiment of simulating a human. If you fully simulated a human with non-deterministic inputs everywhere you probably would indeed have a fully deterministic system. But that's just not how the real world environment a human is in behaves.