What should we learn now? (ducktape.blot.im)

A recent discussion highlights the dual impact of large language models (LLMs) on coding practices within the corporate tech landscape. While LLMs have improved coding efficiency and potentially increased code quality, there is growing concern that reliance on these tools diminishes opportunities for software engineers to learn and develop their technical skills. The friction traditionally encountered while working on legacy codebases has been eliminated, leading to a potential atrophy of essential coding competencies as engineers juggle multiple projects with the assistance of AI agents. To counteract this trend, the article suggests drawing parallels with other industries, such as aviation, where practical experience remains crucial despite automation. It emphasizes the importance of hands-on learning—encouraging engineers to tackle projects manually, explore new programming languages, and engage deeply with coding concepts beyond their current level of abstraction. By honing their skills independently and avoiding over-reliance on LLMs, engineers can maintain a competitive edge and a broader understanding of their craft, ultimately leading to more innovative solutions and a deeper mastery of technology.