The Uncomfortable Question
As an engineer, I often ask myself whether mentoring junior engineers still makes economic sense. A few years ago, the path was predictable: juniors handled basic tasks, learned the codebase, and became reliable contributors within 6–12 months. The early period required guidance, but the return was clear and arrived within a predictable window.
AI tools changed that structure. Much of the work that historically built junior competence, such as small features, refactoring tasks, and routine implementation, can now be produced quickly through Claude, ChatGPT, or Copilot. This reshapes team expectations about where early productivity should come from.