They don’t even explain it in physics class. That is kind of the schtick of the Copenhagen interpretation. You just assume as a postulate that systems are in classical states when you look at them and in quantum states when you do not, and from those two assumptions you can prove using Gleason’s theorem that the only possible way the former can map onto the latter is through the Born rule. But there is no explanation given at all as to how or when or by what mechanism this transition actually takes place.
Many Worlds isn’t much better because they posit that the classical world does not even exist, yet that clearly contradicts with what we directly observe in experiments, so if that is true it necessarily means that the classical world is an illusion, and so then you still have to explain how the illusion comes about, which they do not. Dropping the postulate that there is indeed a classical world also disallows you from deriving the Born rule through Gleason’s theorem, and so it then becomes unclear how to do it at all without some arbitrary additional postulate, and the arbitrary nature of it means there are dozens of proposals of different postulates and no way to decide between them.
Modern physics is of the form (1) there is a quantum state, (2) you look at it, (3) a miracle happens, (4) you perceive a classical state, and then you are repeatedly gaslit into believing quantum mechanics is a complete theory of nature and it’s impossible for there to ever be anything more fundamental than it and any physicist who thinks there might be, even if they are literally Albert Einstein, is a crank crackpot. They then take on the same playbook as the Christians where when you point out their explanation seems to be logically incoherent, they say, “God has no obligation to make sense to you” as an excuse to be incoherent and making no sense, but just replace “God” with “nature” and the same argument is repeated verbatim.
The interference pattern disappears if anything becomes entangled with the which-way information at all, it doesn’t need to even be an “observer” (unless you are using “observer” broadly enough that it can include even a single particle). You can replace the entire measurement device with a single particle that interacts with the particles at the slits in such a way that it becomes perfectly correlated with the which-way information that the observer has no awareness of (such as if a moat of dust interacts with the particle because the experimenter did not isolate it well) and that is sufficient for the interference pattern to disappear.