In a recent post on the Haskell discourse forum, a programmer outlined methods for staying engaged with software development despite the increasing presence of Large Language Models (LLMs). The author expressed concerns regarding "AI burnout" and the potential for developers to become mere "cogs in a machine" if they rely entirely on generated code.
How to Maintain Programming Joy in the Era of LLMs
The proposed workflow focuses on using AI agents to handle incidental or routine tasks—such as cleaning up code, performing small refactorings, or managing repetitive work—while keeping the core design and implementation under human control. Rather than following a "plan first, then let the agent code" approach, the author suggests planning together with the AI but performing the actual coding manually. This method aims to use LLMs as specialized tools for research and boilerplate reduction, rather than autonomous agents that generate complete pull requests (PRs).
The author emphasized the importance of maintaining ownership of the codebase and the importance of human-centric communication. They noted that while LLM assistance can significantly increase productivity, relying too heavily on it risks the degradation of the developer's professional satisfaction and the quality of the software.
Sources
- How to keep enjoying programming in a world of LLMs （Hacker News Frontpage, 2026-09-26）