ArjanCodes warns software developers that AI cannot assume professional responsibility

Coding is the easy part of the job

Equating software development with writing code is like equating carpentry with driving screws. An impact driver makes the task faster, but it cannot frame a roof or ensure structural integrity. In high-stakes environments like banking or insurance, shipping features in ten minutes is reckless, not efficient. These organizations prioritize minimizing risk over speed. Developers spend the majority of their time on software design, architecture, and stakeholder communication to ensure systems are secure by default and maintainable over time.

Delegating tasks versus delegating responsibility

Business owners often misunderstand the nature of delegation. You can delegate a task to an AI, but you cannot delegate responsibility. If an automated agent ships a feature that causes a data breach or financial loss, the AI does not face the legal or professional consequences—the human developer does. This human element remains the bottleneck for full automation; someone must always be there to assume responsibility and verify the correctness of the output.

Focusing on evergreen fundamentals

To survive the shift toward automated tools, developers must double down on design principles like cohesion, coupling, and abstraction. These fundamentals allow engineers to translate complex business requirements into practical, resilient systems. Tools like GPT-5 or Gemini will continue to evolve, but the need for creative problem-solving and system simplification remains constant. Practical mastery comes from understanding trade-offs, not just knowing which prompt to type.