what the diffrenece between custom instruction and memory of ChatGPT
#what the diffrenece between custom instruction and memory of ChatGPT
earnest bone
split fiberBOT
Hi @earnest bone. I'm kapa.ai, your friendly support bot. I will try to answer your question by searching through the knowledge base. :wave:
Custom Instructions and Memory in ChatGPT serve different but complementary roles in guiding the model's behavior and responses.
Custom Instructions allow you to provide direct guidance to ChatGPT on what you'd like it to know about you and how you'd like it to respond. These instructions can be explicit information or instructions that you add to your Custom Instructions. They are used to fine-tune the interaction without needing to modify the underlying code or model. Custom instructions are available on all plans on Web, iOS, and Android, and they are added to new conversations going forward. However, there is no API for custom instructions; in the Chat Completions API, system messages should be used for a similar effect. Custom instructions should also influence the output from plugins (source (https://help.openai.com/en/articles/8096356-custom-instructions-for-chatgpt)).
On the other hand, Memory in ChatGPT is a feature that allows the model to remember details between chats, making its responses more relevant and personalized. As you chat with ChatGPT, it will remember details and preferences from your conversations, and its memory will improve the more you use it. You can teach it to remember something new by chatting with it. Memory works similarly to Custom Instructions, but the models are trained to update the memories rather than requiring users to manage them. When you share information that might be useful for future conversations, the model can add a summary of that information to the notepad. The model has also been trained to manage memories, including by intelligently updating pre-existing memories, consolidating memories where appropriate, and forgetting memories when instructed by the user. You’re in control of ChatGPT’s memory and can reset it, clear specific or all memories, or turn this feature off entirely in your settings (source (https://help.openai.com/en/articles/8590148-memory-faq)).
In summary, while both Custom Instructions and Memory help guide the model's responses, Custom Instructions are more about providing direct guidance and instructions, while Memory is about the model learning and remembering details from your conversations over time.
Note: you can ask me a follow up question by @ mentioning me again :speech_balloon:
wet pier
@split fiber what limit does the memory feature have?
split fiberBOT
Let me check the knowledge base for you. One moment! :books:
The Memory feature in ChatGPT has several limitations and considerations:
-
Geographical Limitations: As of the information available, Memory has been rolled out to ChatGPT Plus, Team, and Enterprise users, except for those located in Europe and Korea. OpenAI is working on making Memory available to all users (source (https://help.openai.com/en/articles/8590148-memory-faq)).
-
Privacy and Safety Considerations: Memory brings additional privacy and safety considerations, such as what type of information should be remembered and how it’s used. OpenAI is taking steps to steer ChatGPT away from proactively remembering sensitive information, like health details, unless explicitly asked to. Users can always review and clear specific memories (source (https://help.openai.com/en/articles/8590148-memory-faq#h_e5c7bd7a3c)).
-
Workspace Limitations: If you merge your personal workspace into your Team workspace, all data, including conversations and memories, will be migrated to your Team workspace. However, memories are tied to each individual account and cannot be transferred to another account (source (https://help.openai.com/en/articles/9295112-memory-for-chatgpt-team-faq)).
-
Control Over Memory: Users have control over ChatGPT’s memory. They can reset it, clear specific or all memories, or turn this feature off entirely in their settings. If they’d like to have a conversation without memory, they can use Temporary Chat (source (https://help.openai.com/en/articles/8590148-memory-faq)).
-
Memory Persistence: The GPT should act as a memory aid, designed to help users recall personal details, appointments, and events based on prompts. It should provide reminders and store information briefly during interactions but won’t retain information after the conversation ends (source (https://community.openai.com/t/731698)).
-
Memory Availability: Memory is available to individual accounts via their Settings, but only Enterprise workspace owners can limit workspace access to personalization and memory features for the entire workspace (source (https://help.openai.com/en/articles/9295112-memory-for-chatgpt-team-faq#h_b2151f0f48)).
Note: you can ask me a follow up question by @ mentioning me again :speech_balloon:
wet pier
@split fiber Re "won’t retain information after the conversation ends " I thought that was exactly what the memory feature did.
@split fiber Re "won’t retain information after the conversation ends " I thought that was exactly what the memory feature did
split fiberBOT
Let me check the knowledge base for you. One moment! :books:
You're correct, and I apologize for the confusion. The Memory feature in ChatGPT does indeed retain information after a conversation ends. This information is stored separately from your chat history, which means that even if a chat is over or has been deleted, the memories created during that conversation can still be used in future interactions (source (https://help.openai.com/en/articles/8590148-memory-faq#h_1657929092)).
The statement "won’t retain information after the conversation ends" was taken from a forum post and may have been referring to the default behavior of GPT models without the Memory feature enabled. With the Memory feature, ChatGPT can remember details between chats, allowing it to provide more relevant responses (source (https://help.openai.com/en/articles/8983136-what-is-memory)).