Openai has just launched an important memory update for chatgpt. Although subtle in a way, I think it could mark a significant change in the way people get involved with AI, certainly in the long term.
Before now, the chatgpt memory was limited to the current session unless Chatgpt decided that it should be added to the long -term memory or if it did manually. Otherwise, each new conversation was a clean slate.
Now, Chatgpt can extract from all your chat history in each session to respond to your last consultation. Know your atmosphere and can track your projects. He will remember the things of his discussions, even if he has forgotten.
He still has the memory saved by the user who deliberately asks him to store, but now, every small comment and asks will also be part of how Chatgpt processes conversations with you, like an educated robot intern that secretly maintains a newspaper. If you want to find out what is the chatgpt image of you, you can ask you to “describe it depending on all our chats.”
You may not think that this is such a big change, but as someone who has become a regular chatgpt user, I can easily imagine how it will benefit me. When I ask for an idea of recipe, Chatgpt will now extract previous recipes that are provided and ask me if I liked the result, presenting new meals ideas based on my opinion about the previous one.
The same goes for the ideas of the history of bedtime. I never want to write one completely, but I inspire myself in the premises that suggests that Chatgpt, and now it will be better for the suggestions I have said before.
While the new features and improvements in AI chatbots may sometimes seem a lot of noise for something that is not so great and persistent memory feels like real progress just because it is a characteristic constructed in the long term. Maintaining the context in all interactions facilitates that the general “relationship” feels more significant.
It also opens the door to new use cases. Imagine the tutoring that adapts to your learning style over the weeks. Or the therapy diary with an AI that remembers what you said three sessions ago. O Productivity planning that does not need to be re -applied every Monday morning. It does not need AI to be sensitive whenever it is consistent.
Memorable movements
However, the improvement of chatgpt memory is not without complications. Make an AI remember you over time inevitably raises questions about privacy, autonomy and, frankly, how much information you want your partner to have.
Yes, it is useful that you remember that you are Kosher and you like some spices in your dishes, but you don't want me to assume too much.
This is quite specific to me, but I do many chatgpt tests and its characteristics, and not all tests are based on my real life. I will not travel to Japan next week; I just wanted to see how Chatgpt would do it when designing an itinerary. Then I have to eliminate that session or explain to the AI that I should not use that question by formulating answers to other questions.
There is also a philosophical element. The more the memory imitates, the easier it will be anthropomorphic. If you remember your favorite sports team, the name of your pet or your disgust for the semi -colones, begins to feel like a person, and it is vital not to attribute self -awareness to an algorithm that is far from achieving it. It is easy to trust a tool that reminds you. Maybe it's too easy in this case.
However, for better or worse, I maintain that the integral memory of Chatgpt is one of the most consistent updates this year so far and probably still when it has finished 2025.
Memory is a powerful trick, even if it does not allow you to make a version of Ghibli Studio of yourself. Memory is what converts an inert tool into a long -term assistant. Even if your assistant is just a digital emulation of a brain that floats in a cloud, it is good to remember the little things.
