OpenAI is testing a new one memory functionality for ChatGPTwhich will allow the chatbot to remember details of previous conversationsand to use them to offer more targeted and personalized chats.
How will ChatGPT remember conversations?
ChatGPT’s memory works in two ways. The first is that the chatbot can collect details from the conversations it has with users. To give a Valentine’s Day-themed example (since it occurs today), if a user asks ChatGPT to write a letter for his girlfriend called Simona, the chatbot will remember that the aforementioned user has a girlfriend called Simona. So, the next time the user asks him to write a similar message, the chatbot will use the name Simona in the message. In short, hope you never break up, otherwise paranoia starts with every chat. But don’t worry, as we will see shortly it is possible to remove some elements from ChatGPT’s memory.
The second way is that the user himself can ask the chatbot to remember something specific. For example, if a user prefers to receive results in a certain format (e.g. in a bulleted list), you can ask the chatbot to remember this preference. So, the next time the user asks him to explain a topic, the chatbot will present the content in a bulleted list.
It will be possible to manage ChatGPT memory
If the prospect of seeing an ex’s name appear in chats has alarmed you, don’t worry. In fact, if a user can do like the Man In Black and flash the chatbot (or cast an oblivion spell on it for Harry Potter fans). Basically it will be possible to make ChatGPT forget certain information previously assimilated. All from the chatbot Settings.
Not only that: functionality temporary chat will allow users to use the chatbot without it remembering anything. A kind of incognito mode.
And privacy? Will conversations with ChatGPT be safe?
OpenAI specified that will not use the chat memory function to train its models.
It is currently unclear when the memory feature will be available to all users, as this is still in the testing phase.
