If you are studying Computer Science and read this, please stay in school. LLM models are trained on existing code, which has bugs. There will always be bugs to fix.
There’s been a shift away from hiring juniors but not because of AI. Everyone wanted seniors because they erroneously believe that they can get to market faster by shunning less experienced developers.
Building software is more solving problems than writing code. Understand the problem, formulate a solution. Current gen LLMs generally regurgitate solutions based on prior examples, and preform poorly in novel situations.
And that’s OK. Focus on learning the theory. Build experience knowing when, where and how to apply that theory. Perfect how to prompt the system and let the algorithm write the grunt code.