i think this is flawed, tbh. it's true that up until this year, llms did not make people any faster at coding, or more able to tackle challenges outside their skillset. but that has changed, and if you're in uni now, you're not trying to be useful today, you're trying to be useful 4-5 years from now. it's wildly irresponsible imo to gamble your future on llms failing when the trend has been the exact opposite.

you also shouldn't gamble your future on llms being easily available forever. but you don't have to do either.

the real thing that matters isn't line-by-line coding ability, it's a deep understanding of technology. that's what you should be focusing on. it cuts juniors deep (and by definition, most coders at uni are juniors) because they tend to be the group that focuses the most on the superficial aspects of coding and not on the underlying reasons and technologies, but that focus was always wrong, it just used to be easier to get away with being wrong.

focus on being someone who understands systems and builds things, whether or not you have an ai to help you. programming is not an abstract field judged arbitrarily by how well your peers like you, at the end of the day it's very results-driven. we build real things, and it matters more what you can build than how you do it. and what you can build directly depends on your understanding of technology, regardless of how much ai is being used at building it. it does not matter if your c++ is 10% more idiomatic or 20% less llm-coded than that of your peers, what matters is if you managed to solve the problem.

even if we assume llms will be more capable in the future, not less, that direction will help you, because most non-coders don't know what the hell to ask an llm to get something done. they also didn't know where to start before llms were a thing. it's your willingness to learn and your understanding of where to go, what to reach for, that sets you apart from an mba with a claude subscription, not your ability to dance around the borrow checker.