Should you still learn to code?
The honest answer from a school that teaches programming, including the parts that argue against us.
We teach programming for a living, so treat what follows with the suspicion it deserves. We have tried to write the version we would want to read if we were deciding whether to spend six months of evenings on this.
The case against, stated properly
It deserves to be stated properly, because the weak version is easy to knock down.
Writing code, the physical act of turning a known solution into syntax, is being automated quickly and well. If your plan was to make a living converting clear specifications into CRUD endpoints, that plan is in trouble. It was already in trouble before the models arrived, but the timeline has shortened.
Junior roles feel this first. A lot of what a junior developer was once paid to do, a competent engineer with a good model now does in an afternoon. Anyone telling you the entry-level market is unchanged is not looking at it.
What the argument misses
It treats programming and typing code as the same activity. They never were.
Most of the job is deciding what to build, noticing that the requirement contradicts itself, working out why the thing that passed in staging fails in production, and choosing which of four workable designs you will still be happy with in two years. A model helps with all of these. It does none of them for you, because each one requires knowing things about your system, your users, and your constraints that were never written down anywhere.
There is a second, sharper point. To use these tools well you have to be able to tell when they are wrong. Models produce fluent, confident, plausible code that does not work. Spotting that takes exactly the skill the tools were supposed to make unnecessary. The people getting the most leverage out of AI are the ones who could have written the code themselves, just slower. That is not a comforting observation but it appears to be true.
What changed, concretely
The floor moved up. Producing basically-working code is no longer a marketable skill on its own. That is a real loss for people who were counting on it.
The ceiling also moved up, and further. One person who understands systems can now build and operate something that used to need a team. The gap between knowing a little and knowing enough has become much more expensive to sit in.
The learning curve got gentler at the start and steeper in the middle. You get to a running program on day one, which is genuinely motivating. Then you hit the part where you have to understand what you built, and the model cannot do that for you.
So, should you?
If you want a stable job doing what you are told: probably not this one, and probably not most others either.
If you want to be the person who decides what gets built, who can look at a system and see where it will break, who can take an idea and make it exist: yes. That has arguably never been more worth learning, because the tools that amplify it have never been this good.
The thing to be honest about is that it is hard, it takes longer than the ads suggest, and there is a stretch in the middle where you feel stupid every day. That was true before the models and it is true now. What is different is that you no longer have to be stuck alone at two in the morning, which turns out to matter quite a lot.
If you want to do it with someone alongside you, that is the whole reason our courses are live rather than recorded. Have a look at what we teach, or tell us where you are starting from and we will tell you honestly whether we can help.