If You're Tired, You're Not Broken — You're Paying Attention

I have fifteen years of experience. I've shipped compilers, distributed systems, a database that didn't catch fire in production. Then, about eighteen months ago, I started noticing something: I couldn't debug things the way I used to.

Not because I forgot how. Because I'd reach for a problem and AI had already solved it before I got there. The answer was just… sitting there. I'd accept it, ship it, and move on. But I couldn't tell you why it worked. Not really.

The moment that broke me was in a code review. A junior asked me why a particular optimization was correct. I had no idea. I had approved it because it looked right and the tests passed. I'd become a pattern-matching stamp.

What saved me: I started doing one hour every Friday with zero AI assistance. Just me, a problem, and whatever I could figure out. First weeks were uncomfortable. Now it's the best part of my week.

I learned to code in a twelve-week bootcamp in 2023. I graduated into a market where every job listing said "experience with AI tools a plus." So I learned to code with Copilot running from day one.

Here's the problem nobody talks about: I never developed instincts. When something breaks in a way that's weird, I genuinely don't know if it's a simple typo or a systemic architectural problem. I can't tell when I'm looking at something I should know. I just… prompt.

I got to about nine months in and realized: I can build things, but I don't know what I know. That gap — between being able to produce code and actually understanding what you're producing — is the most unsettling feeling.

The quiz on this site was the first thing that made me feel like maybe it wasn't just me. The "Explanation Requirement" strategy helped. Making myself explain, out loud, how a piece of generated code actually works — that's uncomfortable and also the only thing that's actually building real understanding.

I managed a team of eight engineers at a mid-stage startup. We adopted AI coding tools company-wide in Q1 2024. Within three months, velocity metrics looked great. Within six months, I noticed my best performers were withdrawing.

The tell was code review. People who used to debate trade-offs would approve things with no comment. The conversations that used to happen — about why one approach was better than another — had gone silent. We'd achieved velocity at the cost of exactly the kind of thinking that made us a good engineering team.

What hurt most: I could see it happening and I didn't have the language for it. HR thought I was worried about nothing. The CEO pointed to the velocity numbers. I felt like I was watching my team slowly lose something and nobody else could see it.

What helped: I started running monthly "no-AI" retros where the team deliberately discussed what they'd figured out themselves that month. I changed our code review culture to require explanation of AI-generated sections. Eventually I had to leave — the company's velocity obsession was too deep. Found a team with healthier norms. Night and day difference.

I'm a remote engineer at a US company, working from Eastern Europe. The overlap window with my team is already small. AI tools made it easy to stay in that window — I could get answers to questions that would have required scheduling a meeting or waiting until tomorrow.

What I didn't notice happening: I stopped talking to my teammates. Not because I didn't like them. Because AI was a faster answer. And in a small overlap window, speed felt critical.

About eight months in, I realized I couldn't remember the last time I'd had a real conversation with a teammate about a hard problem. I'd become efficient in a way that was quietly removing something I actually loved about being an engineer: the shared wrestling with something difficult.

What helped: I set a rule — any question that takes me more than three back-and-forths with an AI tool, I ask a teammate. Even if the AI could eventually answer it. The first week was awkward. Now it's the thing that makes me feel like I'm actually on a team again.