Enhancements for the Memory Feature

I have been exploring the Memory feature extensively and believe there are several improvements that could make it even more useful:

  1. Currently, when conversing in Japanese, I have to specifically request that memories be saved in Japanese. It would be more convenient if the memory could automatically save in the language of the conversation.

  2. The memory saving feature, particularly in Japanese (as I cannot read English), seems imperfect. For example, when I send a specific text and ask for it to be remembered exactly as it is, it appears correct in the memory update section. However, when I access the Manage Memory option, the content sometimes changes. Thus, it would be beneficial to add an editing function to the Memory feature to allow for corrections and updates.

  3. Currently, unless explicitly instructed to use the memory feature, ChatGPT defaults to generating responses based on its training data without pulling from the memory. I suggest adjusting the priority to first refer to the stored memories and then to the broader GPT knowledge. This change would ensure that interactions are more personalized and contextually relevant based on the stored memories.

These enhancements would significantly improve user experience, making the Memory feature more intuitive and efficient.