So I did go through a few articles regarding the same issue I had before I understood the issue I faced today with ChatGPT, and the change is very frustrating.
I have a chat that’s basically my journal, where I reflect on my day and asides from one major topic, there are other small things I share with it that are part of my day too. I’ve been doing this with the chat without any problems for the past 4 months, and only two weeks ago did I change to a new chat once the limit reached. On a daily basis, I chat about roughly 4,000-5,000 words, not exactly but an estimate of that. And it has been holding up very well for the previous chat.
The problem is: the context window hit its limit extremely fast this time, and you can’t even chat in that conversation anymore. It’s been two weeks only, yet my continuation journal chat has been restricted due to the limit and it forces me to get a summary. I went to other chats, and realised that the chat used to dedicate my learning for Korean Language, more topics has been locked for the same “context window full” reason.
Getting a summary isn’t the issue — for the long term, that would mean I would basically switch chats and generate summaries again an absurd amount of times. The chat could keep summarising, and eventually I might have 30 different chats on the same journal I have, just a continuation one after another after another. Besides, the major topic is remembered. But sometimes the minor topics aren’t remembered, and forgotten things that were unable to summarise would be repeated again. Before you know it, 1/4 of the chat is already gone. And it hasn’t even been a day. And this is an issue I find very frustrating.
This affects me very majorly, and I believe it affects many other ChatGPT users, regardless of what version (Free, Plus, Pro) they use, face this same problem. One: I can’t have a smooth and flowing journal without context needing to be carried over and over. Two: I can’t have a peace of mind because context is being forgotten and left behind in the original chat. Three: with my other research and study chats, especially my Korean learning chat, I would have to branch them out after a small piece of information is given. Then one research topic could branch out to a ridiculous amount of continuation chats. Even a few research topics would flood my entire ChatGPT chat history with hundreds of chats of just a few topics over the long term. And it affects my studying journey, my routine, and most definitely how I reflect on my day with the most prioritised journal.
What I think needs to be changed
For the many users out there including me that has long, dedicated chats to one specific topic, we would certainly need the conversation to continue and flow, and that we can return to one chat where everything is there and the memory is mostly retained, rather than branching out to more chats, and eventually things are all over the places in the history chats. What I think can be done is, for these recent changes of full context windows to be removed and reverted back to how it was like before these changes, and so that chats can sustain smoothly for at least a few months without needing to branch out as a continuation for a few days.
I certainly believe that even if the chat limit is extended for upgraded plans on ChatGPT, this problem still affects because previous chats could sustain a reasonably huge amount of texts, while context being full merely blocks the users out from using the chat again. Forgotten pieces of memory from weeks ago can be mentioned again and reinstated to the chat. We need chats that can flow, not continued and branched out chat after chat after chat.
And thus, this addresses the problem I’ve identified myself for my own experience and for other experiences from other people’s perspective too. And if these changes can be reverted and fixed, using ChatGPT would be way more convenient and less frustrating to simply use the app and the AI.