Computing Life · YageMay 12, 2026, 08:00 UTC+8

How AI Caused and Fixed My Insomnia

AI如何导致和修复了我的失眠问题

The author used AI to build a HealthKit export app in about 5 minutes and run multivariate regression, finding that the last post-dinner AI usage time correlated negatively with sleep duration; after avoiding AI at night, average sleep increased by 1 hour and 40 minutes.

Why it matters: HKR-H/K/R all pass: a first-person quantified experiment links post-dinner AI use to shorter sleep, then reports +1h40m after stopping. Personal-blog scope keeps it below major industry-update territory.

Read the original ↗Export Markdown

Take

This is not wellness content; it is a heavy AI user treating himself like a production system. He used AI to build a HealthKit export app in 5 minutes, pulled Apple Watch, caffeine, alcohol, glucose, blood pressure, and hourly activity data, then ran multivariate regression. The strongest negative correlate was not coffee. It was the last post-dinner AI session. After cutting night AI use, average sleep rose by 1 hour and 40 minutes.

I buy the direction, not the strength. This is one person over dozens of days, and the piece does not give coefficients, p-values, or the control table. Causality stays in self-experiment territory. The useful mechanism is sharper: AI lowers the cost of coding, cleaning, and modeling, so people keep doing high-cognition work at night. The Cursor-style productivity story rarely prices that in: tasks do not disappear; the stopping point moves later.

Source: Computing Life · Yage · yage.ai
Published: May 12, 2026, 08:00 UTC+8 · May 12, 2026