Feedback on ChatGPT's Memory and Correction Consistency

Hello,

I’d like to report an issue with ChatGPT’s memory and consistency regarding custom instructions. I have specifically requested that ChatGPT correct all my errors in grammar, syntax, conjugation, punctuation, spelling, and sentence structure at the start of every conversation. However, I’ve noticed that it doesn’t always apply this rule systematically and often “forgets” to correct my mistakes unless I remind it.

That being said, I really appreciate the way ChatGPT provides corrections when it does. The corrections are precise, detailed, and well-explained, which helps me improve my writing over time. I like how ChatGPT not only fixes mistakes but also suggests alternatives, explains why a correction is needed, and adapts to my writing style. This approach is very effective for learning, and I’d love for it to be applied consistently across all my conversations.

For example, I expect corrections on:

Missing or misplaced punctuation (commas, semicolons, hyphens, etc.).

Incorrect verb conjugation or agreement.

Syntax errors, such as unnatural phrasing or awkward sentence structures.

Spelling mistakes, including typos or missing accents.

Overuse of specific words or filler phrases.

Since memory is supposed to help maintain user preferences across discussions, I believe this should work more consistently. It would be great if ChatGPT could reliably follow user-defined correction rules without needing manual reminders.

Thank you for your attention to this matter!

Best regards,

Yeah. that is a long known problem and it is not easy to solve - my idea was to add shorter custom instructions to each request (auto injecting them on each request after your prompt).