'Writing Is Thinking': Why Heavy AI Users Insist on Writing Their Own Docs
lmoroney · x · 2026-10-14
Engineer lauren (@poteto), who works with agents daily, wrote that she never lets agents write for her—docs, tweets, even most prompts—because "writing is thinking."
A practical point: articulating what you want in your field's vocabulary makes agents more effective, much like engineers prompting for code. Gergely Orosz noted the most AI-heavy people at the labs do their own writing to keep thinking; @kepano added that delegating reveals what you uniquely care about, and writing is the best way to find that.
Suggested exercise: draft one doc fully by hand, then have AI review only for clarity and typos, noting ideas that surfaced only while typing.
Related event: Writing Is Thinking: Engineers Refuse to Let AI Write for Them(3 posts)→
More from coding & agent
- DeepSeek V4.1 Rates Itself Highly—Then Distrusts Its Own Subagents — teortaxesTex · 2026-10-14
- Dev ports Wright CAD workflow to ChatGPT Desktop, wiring Codex to SteveCAD for 3D design — burhop · 2026-10-14
- Dev one-shots RightClick-Utilities: open-source Windows 11 right-click menu for video compression and audio removal — rms80 · 2026-10-14
- Gemini CLI doesn't support Gemini 4 Argon yet; workaround via '-m gemini-4-argon' — peterjliu · 2026-10-14
- Gemini 4 Argon 'thinks' surprisingly little while doing accurate multi-agent research — ChrisUniverse · 2026-10-14
- Cramer: annotate interface constraints in comments or agents will stomp over them — zeeg · 2026-10-14