You are right on both counts. I do think that it's however different on the first concept. Before, they would be ahead but still capped by their university. If you come from a uni in a 100k person city, you probably would not have the material nor the best teachers. Now you can have literal Stanford quality education (by accessing Stanford's open source lectures) as well as the collective aggregated knowledge of humanity in the chat interface. The curiosity/intrinsic motivation is the only limit except for perhaps compute.

As for the other question, its mixed. I think about 20% of students understand that they are fucked if they just delegate it all to LLMs, they still go through the ropes and show up to class but do the minimum. However most are down the deep end in various degrees. I have seen students with 5 different 3000-line files for 5 questions for the same lab where each file has 3 lines of code different. This never happened even when the students cheated by accessing old labs online or plagiarizing before.

I believe that what will happen (because universities move really slow on policy and education on LLM use), is that pre-LLM, the university had a normal distribution of skills upon graduation. A company could trust that someone with a degree knew X and Y. With this however, you have more of a bimodal distribution, some know nothing and some know it all, so then the trust in universities deteriorates. I think we will see much more IQ-test/practical tests in hiring processes as the trust falters for that a degree equals something.