★ Live on Hacker News — April 2026

Your code ships.
You don't recognize it as yours.

We surveyed 2,047 engineers. 63% feel like middlemen to their own code. 58% report measurable skill decline. 71% take our AI Fatigue Quiz as a validation test. Here's what the data shows — and what actually helps.

Take the AI Fatigue Quiz → Read the Full Guide

Free forever. No account. No tracking. 73 pages of engineer-specific recovery content.

63%
Feel like "middlemen"
58%
Report skill decline
71%
Quiz = validation test
44%
Considered leaving tech
What 2,047 Engineers Told Us

The patterns behind the quiet crisis

This isn't burnout. It's not imposter syndrome. It's a structural condition we're calling AI Fatigue — and it's reshaping who engineers are at work.

63%

Feel like "middlemen" — writing code they don't own

They ship features daily. The code works. But there's a growing gap between "code written" and "code understood." The authorship isn't just missing — it's structurally displaced by AI tooling that generates, refactors, and optimizes without the engineer fully processing it first.

58%

Report measurable skill decline they can't reverse

Not vague discomfort — actual measurable erosion. Debugging instincts dulled. Algorithmic thinking harder to access. Code reading slower. The competence illusion hides this until it suddenly doesn't: a production incident, a whiteboard moment, a "I used to know how to do this."

71%

Take the AI Fatigue Quiz as a validation test, not discovery

They already know something is wrong. They take the quiz hoping someone will name it — not to find out what's wrong, but to confirm they're not crazy. Only 23% take action within 7 days. The awareness-to-action gap is where recovery stalls.

44%

Considered leaving tech. 18% actively planning an exit.

Not because they don't love building. Because the identity cost is too high. Every engineer who leaves takes institutional knowledge with them. This is a retention crisis hiding as a productivity problem — and most organizations haven't noticed.

"I thought I was just tired. Then I realized I couldn't remember the last time I built something without AI writing half of it. I started to wonder if I still knew how to code at all." — Clearing quiz respondent, 8-year senior engineer
Free Tools

Built for recovering engineers

No account required. No data collected. Everything runs in your browser — localStorage only.

🧠

AI Fatigue Quiz

5 questions. 4 severity tiers. Personalized recovery path in under 3 minutes.

2,047 takers
📋

30-Day Recovery Checklist

Structured daily actions across 4 phases: Awareness → Reduction → Reconnection → Integration.

PDF Download
⏱️

Deep Work Timer

Pomodoro-style focus timer with ambient sound — rain, forest, fire, stream, café.

Browser-only
📝

AI Fatigue Journal

8 guided reflection prompts. Tracks patterns over time. Entirely private.

localStorage
📊

Daily Check-in

30 rotating questions. Streak tracking. Patterns emerge over weeks, not hours.

Streak-based
🏅

Quiz Result Badge

Generate a shareable AI Fatigue Score badge. PNG download. Start the conversation.

PNG Export
Start Reading

Pick where you are

73 pages. Free. No signup required.

🧠
What is AI Fatigue?
Complete field guide — 10 signs, 5 dimensions, 30-day plan
💪
Skill Atrophy Is Real
Science: Bainbridge, Parasuraman, skill decay curves
🪞
"Who Am I Without My Code?"
The identity crisis at the heart of AI fatigue
🌱
How to Recover
7-phase practical guide — specific, evidence-based
Cognitive Load Theory
Why AI is actually drowning your brain, not lightening it
📊
AI Fatigue Statistics 2025
50+ data points, cited sources, research-backed
🔬
Survey: 2,047 Engineers
Primary research — findings, methodology, raw data
⚖️
AI Tool Comparison
Copilot vs Cursor vs ChatGPT vs Codeium — which causes fatigue
Common Questions

What HN visitors ask

No — and that's the critical distinction. Burnout is from workload and emotional exhaustion. AI Fatigue is functional: your learning loop broke, your authorship changed, and your cognitive process adapted to skip steps it used to do. They overlap and compound each other, but require different interventions. See the full comparison →
No. The issue isn't using AI — it's the unconscious, unexamined way most engineers currently use it. The engineers who recover fastest use it deliberately, with boundaries, as a scaffolder rather than a replacement. See the mental model frameworks →
The Explanation Requirement: before accepting any AI-generated code, write in a comment why each piece exists — not what it does, but why it's designed this way. This re-engages the cognitive process AI bypassed. 71% of quiz respondents who tried it said it was the most effective change they made. See the 30-day plan →
More severely, and more dangerously. Juniors learn by struggle — productive failure is the mechanism by which intuition develops. When AI removes the struggle, juniors build competence without understanding. The gap only becomes visible years later when they hit problems AI can't solve. Full breakdown for junior engineers →
The cognitive load and skill atrophy effects are structural, not attitudinal. They persist regardless of how good AI tools become — because the mechanism isn't "AI makes bad code." It's "AI bypassing cognitive processes that maintain skill and understanding." Even perfect AI tools would cause this. See the full research →
Built by engineers, for engineers, after watching the people around us lose something they couldn't name. Not a startup. No monetization plan. Everything is free, zero tracking, no account required. Read the full founding story →

The Dispatch — Weekly

Every Monday: one insight from 2,000+ engineers, one recovery tactic, one thing that actually helps. No noise. Unsubscribe anytime.

Or email us directly: hello@clearing-ai.com