I understand that each chat in GPT4 is like the movies 50 first dates. That means each time I start a new chat it does not have access to my other chats. is this correct? If yes, then if my chat thread gets too long it slows down the responses and misses stuff at the beginning. On the other hand GPT4 is well trained on a lot more data than sits in one of my chat threads. What is the best way for me to reference a prior chat when working on a new chat? For example, I have an R&D thread going where I came composing emails to the same group of people and I want the new message to remember both the past content and my writing style.
#Access previous chat in a new chat
violet pilot
neon notchBOT
Hi @violet pilot. I'm kapa.ai, your friendly support bot. I will try to answer your question by searching through the knowledge base. :wave:
Yes, you are correct. Each new chat with GPT-4 does not have access to your previous chats. The AI model itself doesn’t store a history of what you talked about. Individual AI model calls are standalone, and must be told again what you were talking about recently. This is why if a chat thread gets too long, it can slow down the responses and miss stuff at the beginning due to the model's context window limit (source (https://community.openai.com/t/426336)).
However, there are ways to manage this. One approach is to use "conversation compression", which involves using an additional language model call to summarize the previous conversation and make it take up fewer tokens in the chat history. This is a lossy approach and has its own issues, but it can help manage the token limit (source (https://community.openai.com/t/426336)).
As for referencing a prior chat when working on a new chat, you could manually include relevant parts of the previous chat in the new chat's history. This would allow the model to have some context about the previous conversation. However, this would also consume tokens and could make the new chat reach the token limit faster.
Regarding your writing style, GPT-4 does not have a built-in mechanism to remember your specific writing style across different chats. However, it is trained on a diverse range of data and can generate text in a variety of styles. You could potentially guide the model to write in a certain style by providing explicit instructions or examples at the beginning of the chat.