AI made me faster, but review became the real work.

I use AI every day for coding, writing, and planning, so this is not an anti-AI post.

The speed boost is real. What I did not expect was how tiring it feels after a full day of prompting, correcting, re-prompting, and double-checking.

The old pain was writing everything from scratch. The new pain is deciding whether the output is actually correct or just confidently written.

A lot of AI output looks clean on first read, but once I test it, I find hidden problems: missing edge cases, weak failure handling, or assumptions that break in production.

So yes, it saves time. But it also shifts responsibility from typing to judgment.

That is where my AI fatigue comes from. Not from "AI is bad," but from the mental overhead of not being able to trust it fully.

I still use it because the leverage is too useful to ignore. I just treat every output as draft, not truth, and review it harder than code from a teammate I trust.