I am a paying user of GPT-4. During my use, I’ve encountered numerous issues. On November 2nd, for example, when processing longer code snippets or asking it to translate articles, it sometimes responds with a blank output—just an empty response with no content. Additionally, I’ve noticed a significant decline in response quality. Responses are becoming shorter and less informative, especially when I use Chinese. Sometimes I receive only 200 characters or less, or occasionally up to 400 characters, but it still fails to fulfill the task I assigned or provide the information I need. Instead, it often gives meaningless pleasantries.
In the past, it could provide detailed answers, suggestions, and deeper insights. Now, it feels like it’s just giving superficial responses. Sometimes, when I ask GPT to write part of an article or draft a short program, it tells me to wait until it’s ready to complete the task, but it doesn’t actually proceed with the modifications or coding. Only after repeating my command three, four, or even five times does it finally respond. Furthermore, when I ask it to help revise a piece of code or check a specific code snippet, it sometimes just copies the code I sent and pastes it back without any changes.
Additionally, at times when I ask it to look up certain information online, if the content isn’t available in Chinese, it may start fabricating details instead of continuing the search. If I switch to English or ask it to use English for the query, it only retrieves partial information, which sometimes includes incorrect details.
Lastly, the quality of translations has drastically declined. The Japanese translations I request no longer follow the style I specified in my memory bank, and sometimes the translations are completely off. These are all issues I experienced on November 2nd.