Not long ago, junior engineers learned by working closely with more senior ones.
Code reviews, design discussions, and mistakes created friction, and that friction built understanding.
AI changed that.
Today, answers are instant. Code is generated. Progress looks fast, but depth is optional. A junior can ship without ever fully understanding why something works. That’s powerful, and risky.
During our last internship season, our intern chose not to use AI tools while coding. Short-term productivity suffered. Tasks took longer.
But something more important happened: they struggled, debugged, and built intuition. When they would introduce AI tools, the effect would be dramatic. With experience in place, AI will become a multiplier instead of a shortcut. That’s the real distinction.
AI is great at accelerating experienced engineers. It’s bad at replacing the process that creates them. If we remove friction too early, we don’t get better engineers, just faster output with hidden gaps. And those gaps surface later, when systems scale or fail.
This isn’t about banning AI. It’s about being intentional. Let juniors struggle a bit. Accept short-term productivity loss.
Otherwise, we’re not training the next generation of engineers; we’re just shipping code faster.
Top comments (0)