A quiet letter for engineers who've been running in place.
You show up every day. You ship. You review the PRs. You answer the questions. You've been doing this for months.
And yet.
There's a specific kind of exhaustion that doesn't come from working hard. It comes from working hard and feeling like you're not getting anywhere. You've heard this called "burnout." But burnout implies you were going somewhere and burned out from the journey. This feels different. This feels like you're on a treadmill — moving fast, going nowhere.
This is the consistency illusion.
Here's what AI has done to the engineering experience:
It has separated velocity from learning.
In the old world, shipping a feature meant wrestling with it. The difficulty of the problem was part of the process. You learned something in the friction. The hours you spent were also, almost always, hours of skill development.
AI collapses that. AI can get you to "shipped" without the "learned." You can close tickets faster than ever before. Your velocity metrics look great. And somewhere underneath, your skills are quietly not keeping up with your output.
The cruelest part: you don't notice until you try to do something without AI. Then the gap is stark.
From our survey of 2,147 engineers (fielded Q1 2026):
The consistency illusion isn't in your head. It's a structural feature of how AI-assisted work separates doing from learning. When shipping becomes independent of skill development, your habits stop being a proxy for growth.
If you're early career and AI is doing the hard parts, you don't have a baseline for what you're missing. You never built the thing the long way. It feels normal.
If you're senior, you remember building things the long way. You have a clear baseline. And you can feel, with uncomfortable precision, that the work you're doing now isn't building the same things it used to.
The 8-to-10-year engineer who started programming when debugging meant reading stack traces for hours — that engineer has a very different relationship to AI fatigue than the new grad who started with Copilot built in.
The senior engineer knows what the AI is replacing. That knowledge is its own kind of weight.
When you learn something the hard way, it sticks in a way it doesn't when an AI explains it to you. Cognitive science calls this the "desirable difficulty" — the struggle that makes knowledge stick. AI removes the struggle. Sometimes that's mercy. Sometimes it's at the cost of durable learning.
When AI generates something for you, you understand it at a surface level immediately. This creates an illusion of comprehension. You read the code, it makes sense, you move on. But surface comprehension isn't deep ownership. You won't be able to recreate it from scratch — which means you won't be able to debug the edge cases either.
Your manager sees: features shipped, PRs merged, velocity up. You know the hidden truth: the AI did most of the implementation. The metric that everyone celebrates — shipping fast — has been decoupled from the skill that matters — being able to ship well without help.
Your idea of "what a good engineer looks like" is quietly being reset. As AI takes over more of your work, your new reference point is... AI-assisted work. You're comparing yourself to a standard that keeps declining, because more and more of the work is being done by a model you didn't have to train.
This isn't about using less AI. It's about being deliberate about what you let AI do — and what you insist on doing yourself, the slow way, even when it's inefficient.
The standard advice in tech is: show up every day, ship every day, be consistent.
That advice wasn't written for an era where AI does 70% of what "showing up" used to mean.
The question isn't: did you show up today?
The question is: did you get better at something that matters to you?
If you've been consistent for months and the answer is no — that's not a motivation problem. That's a signal. The treadmill is still running. You just haven't been going anywhere.
The way off isn't hustle. It's intentional practice in the gaps AI doesn't fill.
Stop measuring your days by what you shipped.
Measure them by what you learned.
P.S. If you're wondering where to start: take the AI Fatigue Quiz. It takes 3 minutes and gives you an honest picture of where you stand. No email required.
P.P.S. This issue follows up on last week's data section — find the full survey results and methodology on our statistics page.