You don't need to know everyone's level, just a rough idea about what the worst student of the class might need.
The first exam that students in test-prep schools take is before classes start, usually a mock university test that can be used precisely to assess your current level and to gather information about the class overall.
You don't need to know who are the worst performers. But if you know that your class of 250 has an uneven distribution of aptitude, you'd try to break the class apart.
I've taught some college classes and guest lectured in many others. The bottom 10% is trivial to identify in a matter of minutes.
You don’t need to know which specific 50 students are in the bottom two deciles, just that they exist and that your policy is to teach to what you’ve estimated is the 3rd decile and above.
You will find out who they are in the exams, but you know they exist in next year’s class even before that class is admitted to the uni.
Where this matters is in something like a lab or music practicum where the communication between instructor and students is the class. But I can tell you from experience that the variance among different years of students in a single university is large enough to disprove your uncle's point. I had one class that could have been confused for graduate students. Yet another might as well been sent by the Trump administration with the goal of destroying the institution of ear training.