In a world of AI agents, is it really necessary to learn to code? A hard question to answer, but based on some readings/videos here is my view:

Pros for learning:

• Learning the language improves the quality of the code since you can avoid some mistakes. ¹

• If you are responsible for code review, then learning the language is better than trusting agents to produce good, secure code. ²

• Models are becoming very good at findings but there is still the need for human judgement. ³

• If you work in a regulated market (banks, health, finance, etc) you might need to pay more attention to the output rather than velocity.

• The models are much better at programming now, but it is still not perfect because senior engineers are being quietly hired again. ⁴

Cons for learning:

• If you work at a startup, time to market is better than quality. ⁵

• If you want to prototype, vibe coding is the way. It is good enough to prove something and discard it (totally or partially) afterwards.

• If you are a senior engineer, then this is the way to make you more productive. However, if you are a junior engineer AI might harm your productivity. ⁶

Conclusions

• The way software was created has been forever changed. ⁵

• Right now it is still important to learn the programming language, but it depends on what you do, the industry you work on, and the limitations of AI use you have.

• You might not need to become the expert in a language or stack anymore, since AI models get stronger every time. However, you still need the fundamentals of software engineering and languages to drive the wave of the LLMs outputs.

“Writing code is now fast, it’s getting cheap, and quality is going up and to the right. The hard question is no longer how to build it. It’s should we build it.” – https://maggieappleton.com/zero-alignment

I don’t know for how long this will be true since AI keeps improving.

Footnotes

1. One Developer, Two Dozen Agents, Zero Alignment – https://maggieappleton.com/zero-alignment ↩︎
2. Write broken commits for better review- https://huonw.github.io/blog/2026/04/broken-commits/ ↩︎
3. Behind the Scenes Hardening Firefox with Claude Mythos Preview – https://hacks.mozilla.org/2026/05/behind-the-scenes-hardening-firefox/ ↩︎
4. Why Companies Are Quietly Rehiring the Staff They Fired – https://www.youtube.com/watch?v=9-HffhBSLp4 ↩︎
5. Programming in 2026: excitement, dread, and the coming wave – https://amontalenti.com/2026/04/23/excitement-and-dread ↩︎
6. Claude code is not making your product better – https://ethanding.substack.com/p/claude-code-is-not-making-your-product ↩︎
7. Programming in 2026: excitement, dread, and the coming wave – https://amontalenti.com/2026/04/23/excitement-and-dread ↩︎