Coding in the AI Era: When Do Developers Still Write Code?
The landscape of software development is rapidly changing, prompting many to reflect on how frequently they engage in the act of writing code and their evolving relationship with AI-powered tools. While some developers maintain a daily rhythm, others find themselves going weeks or even months without directly inputting syntax, thanks to advanced assistance.
The Evolving Rhythm of Coding
The frequency with which developers write code is diverse. For many, it's a daily ritual, driven by professional necessity or personal projects. However, a significant number report periods of reduced direct coding, sometimes extending to weeks or months, indicating a shift in how development tasks are approached. This variance often correlates with comfort in a language or the nature of the project. Some, like a developer learning Rust, actively code for hours daily, while others, confident in languages like JavaScript or Dart, primarily review AI-generated code.
Navigating AI and Autocomplete: Learning vs. Productivity
A key theme is the strategic use of AI and autocomplete. When learning a new programming language, such as Rust or Emacs Lisp, a common and highly recommended practice is to disable AI suggestions and autocomplete. This "manual" approach is crucial for cementing syntax, understanding core concepts, and truly grasping the language's nuances, rather than relying on an AI to fill in the gaps. Resources like "Rust By Example" are highlighted for structured learning.
Conversely, for familiar languages and routine tasks, AI tools are widely embraced for boosting productivity. They excel at filling in boilerplate, handling syntax details, suggesting API usage, and generating tests or throwaway scripts. This allows developers to focus on higher-level problem-solving and architectural decisions. However, some express frustration with AI, finding it can "yap" excessively, repeat patterns, or be unfulfilling for complex, original work, leading them back to hands-on coding.
The Debate on Generative Coding and Skill Maintenance
The rise of fully generative AI raises questions about its long-term impact on developer skills. Some individuals have transitioned almost entirely to prompting, noting they haven't written a line of code in months, relying on tools like Cursor or Claude. This approach, while efficient for some, can lead to a sense of unfulfillment for those who enjoy the craft of coding.
An important counter-perspective is the deliberate practice of writing code to maintain skills and understanding. Developers express a concern that relying too heavily on AI might lead to skill atrophy, especially in a volatile technological landscape where access to high-quality AI tools might not always be guaranteed. Regularly coding from scratch is seen as a way to stay sharp and ensure that fundamental knowledge remains engaged and accessible, acknowledging that "knowledge is not eternal and needs to be constantly engaged."
The choice between prompt-level and code-level work often boils down to individual enjoyment and the specific demands of a project, with many still deriving greater satisfaction from direct interaction with code.