I had to fix a batch of vulnerabilities in one of our applications at work the other day.
I thought to myself, "I'll get AI to do this for me".
I knew what needed to be done, dependency upgrades that could potentially break the build. It felt like the perfect task for AI: Upgrade the dependencies. Fix the build. Save me the time.
I prompted the AI tool, I observed it's reasoning process, really impressive and methodical. When it finished, I noticed over 20 files had been modified, with additional 6-8 dependencies added to the POM file.
Technically, my vulnerabilities were resolved.
However, from experience, I knew there was some sort of overengineering going on.
What AI had done was to patch each vulnerable dependency individually. The symptoms had been treated.
I knew upgrading the actual parent dependency would transitively resolve most of the vulnerabilities in one clean move, thereby treating the root cause.
So I discarded all of AIs changes, and upgraded my parent dependency. A few build fixes and test later, work done.
This experience reminded me of something important:
Experience is becoming the differentiator.
AI can do almost everything I can do, usually faster and maybe even better.
In a job market increasingly shaped by AI, the real competition isn't speed. AI will win that every time.
The real value is in judgment.
Knowing when not to patch 20 files.
Know when a solution could be cleaner.
Knowing what will hurt you six months from now.
The engineers who thrive won't be the ones who compete with AI. They'll be the ones who know how to use it and guide it for optimal result.
Top comments (1)
Real value is Judgment. Thank you Gracie