Dear Development Team,
I have been using ChatGPT for a while, and one key improvement I would like to suggest is the introduction of submemory. Submemory would allow users to organize their memories more clearly and utilize them more effectively during longer, in-depth conversations. Currently, memory operates on a single level, which can make managing complex and broad information more challenging.
The Concept of Submemory:
- Management of Categories and Subtopics:
- Users should have the ability to create their own memory structures, such as categories, topics, and subtopics, where specific memories can be stored. This would increase the flexibility of memory management and give users more control over the information stored.
- Deeper Memory Hierarchies:
- Submemory would enable the creation of hierarchical structures for memory, where one memory entry could contain multiple subheadings. This would be useful when users are dealing with larger concepts or themes that are interconnected.
- Visibility and Control of Memory:
- Users should be able to control when memory entries are displayed in full detail and when they are shown in a summarized format. This would provide flexibility, especially when dealing with large amounts of information.
- Users should also be able to clearly view all stored information without restrictions or automatic summarization, ensuring transparency and trust in memory handling.
Benefits of Submemory:
- Better Organization: Multiple topics and projects would be more easily managed, and users could navigate their memories effortlessly by moving between categories and subtopics.
- Adaptability and Efficiency: The ability to customize memory structures and visibility makes the user experience smoother and allows for quicker access to key information.
- Deeper Interaction: Submemory would support larger, multi-level conversations where memory plays an essential role in long-term information management and reflection.
I believe that the introduction of submemory would provide users with a valuable tool for memory management and open up new possibilities for more in-depth conversations. I hope this proposal sparks interest and leads to improvements in future updates.
In example you can code it like:
First layer of memory is memory that stores how/what development team wants AI to memorise.
Second layer of memory (probably won’t be second really, but in example) is memory that stores what user want to keep in AIs memory and is editable by user. Including chance to make categories.
Thank you in advance for your consideration!