How to keep enjoying programming in a world of LLMs
Programmers are grappling with how to keep software development meaningful and enjoyable now that large language models can generate and refactor vast amounts of code. Some embrace AI as “power tools” that remove drudgery and free them to focus on design, architecture and personal projects, while others feel their core craft, identity, and job security are being eroded as roles shift toward supervising agents and reviewing AI output. Across viewpoints, recurring themes include skill atrophy, code quality and maintainability, historical parallels to past waves of automation, and the idea of treating hand-crafted programming more like a hobbyist or artisan pursuit.
Role of Programming: Tool vs Craft
- Many distinguish between programming as a means to an end vs an end in itself.
- Those who see it as a tool often welcome LLMs: they offload boilerplate, parsing, tests, docs-search, and let humans focus on design, APIs, architecture, and product thinking.
- Those who enjoy the act of coding itself feel that LLMs hollow out the part they love—typing code, debugging, reading source, refining logic.
Code Quality and Reliability
- Supporters: frontier models can generate large amounts of working code rapidly, write tests, explore big codebases, and act like “StackOverflow on steroids.”
- Skeptics: LLM code is inconsistent, bloated, breaks invariants, introduces hidden debt, and can’t yet be trusted like compilers or “power tools.”
- Some report teams moving faster but “in the wrong direction,” with more bad decisions and workarounds stacked on top of each other.
Jobs, Status, and Economics
- Strong anxiety about displacement: “coding is over,” junior-level work is automated, and teams that orchestrate many agents may outcompete others.
- Others argue there will still be demand for people who deeply understand code, especially where correctness and safety matter.
- Comparisons are made to textile machinery, print automation, and modern car repair: fewer humans, different skills, more disposability.
Skill Atrophy and Cognitive Effects
- Several posters describe losing confidence in design and problem-solving after leaning heavily on LLMs.
- There is concern that “prompting” and reviewing becomes a new meta-skill while core programming muscles weaken.
- Some intentionally avoid LLMs in personal time, or build their own languages/tools, to keep skills and joy intact.
Hobby vs Work and Emotional Impact
- At work, many feel forced into AI usage and into acting as “managers of bots,” losing flow and attachment to code.
- As a hobby, people either:
- Use LLMs selectively to remove friction and ship more personal tools, or
- Reject them entirely and treat hand-coding like woodworking or playing an instrument.
- A recurring theme is grief: missing code as a “canvas” and feeling one’s hard-won expertise suddenly devalued.