Will AI replace programmers and should I still learn to code?

Yes, still learn to code — but learn it knowing the job has changed shape. AI has made writing code dramatically faster and has not replaced the part of the job that was always the hard part: deciding what to build, why, and whether it's correct. What AI genuinely does well now: boilerplate, standard patterns, converting between formats, first drafts of functions, explaining unfamiliar code, writing tests, debugging suggestions. This is a real productivity change, not hype — a developer using these tools well is meaningfully faster on routine work. What it does not do: - Decide what to build. Requirements come from people who don't know what they want and change their minds. Translating fuzzy human needs into a precise system is most of the job. - Hold a large system in context and understand how a change ripples through it. - Take responsibility. When it's wrong — and it's confidently wrong regularly — someone has to notice, and noticing requires knowing the subject. - Navigate the social parts: negotiating scope, weighing trade-offs, arguing for the approach that's cheaper to maintain. The honest concern, stated plainly: the very bottom of the market is genuinely harder than it was. Tasks that used to be a junior's first six months are exactly what these tools do best, so the bar for a first job has risen. That's a real change and it's fair to be uneasy about it. What that means for how you learn: 1. **Don't skip fundamentals.** If you can't read code, you can't review AI output, and reviewing is now a core skill. People who learn by accepting suggestions they don't understand accumulate a codebase they can't debug — this is the actual career risk, not the tools themselves. 2. **Use AI as a tutor, not a vending machine.** 'Explain why this works' and 'what's wrong with my approach' compound your ability. 'Write it for me' doesn't. 3. **Get good at the things that stay scarce:** debugging, system design, understanding a domain, communicating with non-engineers. 4. **Build real projects with real users.** Judgement about what to build is now more valuable relative to typing speed, not less. The historical pattern is worth noting: compilers, high-level languages, frameworks and Stack Overflow each raised abstraction and each was predicted to eliminate programmers. Demand went up every time, because cheaper software means more software gets built. That's not a guarantee, but 'this tool removes the tedious part' has a much better track record than 'this tool removes the profession'.