On LLMs in programming (blog.danieljanus.pl)

In a reflective blog post, Daniel Janus addresses the impact of large language models (LLMs) on programming, revealing a blend of skepticism and acknowledgement of their utility. He identifies himself as a “conscious LLM-skeptic," using LLMs occasionally for tasks like brainstorming solutions, but expresses concern about the rapid changes they bring to the software development landscape. Janus highlights the anxiety stemming from the evolving expectations of productivity in programming, likening this phenomenon to Parkinson's Law, where technology intended to save time paradoxically leads to increased demands. Janus emphasizes the importance of deep engagement with core programming principles and the joy of learning in his craft, suggesting that reliance on LLMs might detract from these valuable experiences. He articulates a desire to maintain focus on fundamental concepts such as algorithms and data structure, arguing that attention—rather than time—is the most precious resource for creativity and skill development in software engineering. By choosing not to fully embrace LLMs, he aims to preserve his mental space for what truly excites him in programming, emphasizing that human creativity remains irreplaceable despite the technology's rise.