Learning Programming in an Age of LLMs
- ID
- 25535
- Status
- summarized
- Published
- 16 Sep 2026, 5:12 PM
- Fetched
- 18 Sep 2026, 4:53 PM
- Provider
- Hacker News
- Category
- dev-community
- Original URL
- https://blog.ploeh.dk/2026/09/16/on-learning-programming-in-an-age-of-llms/
- Source URL
- https://hnrss.org/best
Summary
- Score
- 8.0
- Created
- 18 Sep 2026, 4:54 PM
- Tags
- Audience
- developersvibe_coderssaas_foundersai_agent_users
What happened
Mark Seemann publishes a reader's letter from someone with no formal CS background who used LLMs to build a large TypeScript/JavaScript system (APIs, PostgreSQL, LLM pipelines, research automation, multi-model workflows), then hit a wall moving to production: 'I may have built a system that is above my own level of understanding.' Seemann, with 30+ years of experience, admits he leans toward disliking LLMs while acknowledging they may be unstoppable, and is most resentful when they perform best.
Why it matters
If you are vibe-coding a production system, this letter names the exact failure mode you will hit: the gap between 'it works' and 'I understand why it works' becomes visible only when it breaks, and at that point you cannot debug without another model. Decide now which parts of your stack you must understand deeply enough to own under pressure—database schema, API contracts, deployment, security boundaries—versus where AI assistance is sufficient. Founders shipping AI-built products should budget time for deliberate comprehension, not just feature velocity.
Discussion angle
Share your own 'above my understanding' moment: when did AI-built code work until it didn't, and what did you have to learn the hard way to regain ownership? Compare strategies for forcing comprehension—reading every generated line, writing tests first, pair-programming with a human—versus accepting AI as a black-box dependency.