How Should Beginners Learn Software Engineering in the Age of AI?

The fundamentals still matter. You need to understand programming, data structures, debugging, testing, and how systems work before you can reliably judge generated code. If you cannot write a smaller version yourself, it is much harder to notice when an AI solution is overcomplicated or subtly wrong. Use AI after making a genuine attempt, and make sure you can explain, test, and modify everything it produces.

Treat an AI tool like an extremely inexperienced junior developer: useful and fast, but lacking context and common sense. Give it precise requirements, constraints, examples, expected inputs and outputs, and failure cases. Then review the result line by line, run tests, compare it with documentation, and ask yourself whether the simplest solution would be better. Learning to specify and evaluate work is valuable, but it depends on having enough knowledge to evaluate it.

The workload can genuinely vary between schools and courses, so it is understandable that some students feel pressured to use these tools. Even then, try separating learning work from delivery work: do the core reasoning and implementation practice yourself, and use AI for low-value assistance such as formatting, brainstorming, documentation, or checking edge cases. Also, if the workload leaves no time to understand anything, that is a course or time-management problem worth addressing rather than proof that independent learning is obsolete.

You do not have to keep pace with classmates by copying their workflow. If AI is doing all the thinking for them, they may be saving time now while creating bigger gaps in their understanding. Try solving exercises yourself first, even if the first attempt is imperfect. Once you understand the problem, AI can help explain an error, suggest alternatives, generate tests, or review your approach rather than completing the entire assignment.