When I first started using ChatGPT actively to review my texts and YouTube scripts, I made a mistake: I edited my drafts according to its recommendations. Thank god I never took its texts wholesale, but even these partial fixes created two problems:
- The texts became less alive.
- I got exhausted — because, as we all know, ChatGPT is incapable of saying “okay, now the text is perfect,” even if it wrote the entire thing from start to finish.
One change that made my writing much stronger was this: I almost stopped using ChatGPT’s feedback on current drafts and started using it for future ones instead. Same story with Ilyakhov’s “Glavred” tool — I use it to notice recurring mistakes, not to polish the text into “info style.” By the way, info style in personal blogs is triple-digested crap — change my mind.
This kind of feedback loop works much better: instead of taking pre-chewed instructions like “remove the personal pronoun” or “the paragraph is too long and hard to read,” I’m forced to develop internal rules. Texts become strong not because the text was improved, but because the author was. You feel the difference?
The same funny effect appears in other areas too — in my case, vibe-coding or making YouTube thumbnails. At the same time, I still use ChatGPT for certain tasks where I act like a monkey with a mouse and a keyboard: my brain’s resources are limited, unlike the electronic one’s.