AI Can Erode Engineering Skills
The Muscle We Cannot Afford to Lose
We talk about AI as if the question were simple.
Will it replace humans?
Will it take jobs?
Will engineers still be needed?
Those questions are loud.
And mostly irrelevant.
The real question for engineering leaders is quieter and more uncomfortable:
How do we make sure our teams don’t lose the ability to think, judge, and decide once AI starts doing most of the visible work?
Because AI doesn’t fail by being weak.
It fails by being good enough.
The Lens That Changed the Question
Some time ago, while reading User Friendly by Cliff Kuang, one example stayed with me longer than the others.
The book explores how automation and design don’t just make systems easier to use — they quietly reshape human capability.
In one case, automation made pilots safer and more efficient in normal conditions, while slowly eroding the very skills they needed when something went wrong. Not because they were careless, but because the system no longer required them to practice those skills daily.
Once you start looking at AI through that lens, the question stops being whether it replaces humans — and becomes whether it slowly trains them out of the profession they’re supposed to master.
This Isn’t Just a Feeling
Studies on the out-of-the-loop problem show that when people shift from doing the work to monitoring automated systems, their situational awareness degrades and their responses become slower and less reliable when something goes wrong.
Researchers Parasuraman and Riley captured this dynamic precisely in their “use, misuse, disuse, and abuse” model of automation, showing how overreliance on systems that usually work leads to complacency and judgment errors even among trained professionals.
The FAA has repeatedly warned about skill degradation in highly automated cockpits and explicitly requires ongoing manual-flight training to keep pilots capable when automation fails.
Automation Is Not the Problem. Atrophy Is.
We’ve automated before.
Compilers replaced assembly.
Frameworks replaced boilerplate.
Cloud platforms replaced data centers.
Each wave removed effort.
Each wave raised the abstraction.
But AI is different in one crucial way.
It doesn’t just automate execution. It automates reasoning-shaped outputs. Code, architectures, plans, analysis, documentation.
Things that used to train our professional judgment.
And that’s where the danger lies.
Not in replacement.
But in skill atrophy hidden behind productivity gains.
Full article and supporting materials are available to paid subscribers
You can use the training budget (here’s a slide for your HR).
Thanks for supporting Practical Engineering Management!
