{
"$type": "site.standard.document",
"bskyPostRef": {
"cid": "bafyreiex43sgtqlocgv6akctjhcbmj5vizypvkbzvfdd2fnwiqp6ok5c34",
"uri": "at://did:plc:lk3jfj3zq4k4wxnk474axylu/app.bsky.feed.post/3medhmsv6qgy2"
},
"path": "/t/feedback-on-chatgpt-s-handling-of-long-term-context/1373634#post_1",
"publishedAt": "2026-02-08T07:31:22.000Z",
"site": "https://community.openai.com",
"textContent": "I would like to provide feedback on ChatGPT’s handling of long-term context and confirmed facts.\n\nSpecifically:\n\n * ChatGPT often assumes “Sunday = day off,” even though my work schedule is irregular and I already shared a confirmed one-month shift.\n * When conversations are resumed after a gap, ChatGPT seems to only look back a few days instead of reviewing at least one month of prior context.\n * As a result, I frequently have to correct basic misunderstandings that should have been known.\n\n\n\nI suggest:\n\n * When users explicitly ask for “date confirmation + premise confirmation,” ChatGPT should review at least the last month of confirmed facts before responding.\n * Better use of user-provided, already-confirmed information (e.g., work shifts, car status, delivery dates, etc.).\n\n\n\nAdditionally, in the mobile app, pressing the thumbs-down button does not open a comment box, so I could not submit detailed feedback in-app.",
"title": "Feedback on ChatGPT’s Handling of Long-Term Context"
}