Talking to an ai chatbot can feel a little like Groundhog Day after a while, when you tell it for the umpteenth time how you like the format of your emails and which of those “fun things to do this weekend” You've already done it six times. OpenAI is trying to solve this problem and customize its own bot in a big way. It is implementing “memory” for ChatGPT, which will allow the bot to remember information about you and your conversations over time.
Memory works in two ways. You can tell ChatGPT to remember something specific about you: you always write code in Javascript, your boss's name is Anna, your son is allergic to sweet potatoes. Or ChatGPT may simply attempt to collect those details over time, storing information about you as you ask questions and get answers. In any case, the goal is to make ChatGPT feel a little more personal and a little smarter, without needing to be reminded every time.
Each custom GPT you use will also have its own memory. OpenAI uses Books GPT as an example: with memory enabled, it can automatically remember which books you've already read and which genres you like best. In fact, there are many places in the GPT Store where you can imagine the memory could be useful. Tutor I could offer a much better long-term course load once I know what you know; The kayak could go directly to your favorite airlines and hotels; GymStreak could track your progress over time.
Memory is a feature that ChatGPT desperately needs, and a total minefield
In many ways, memory is a feature that ChatGPT desperately needs. It's also a total minefield. OpenAI's strategy here is a lot like the way other internet services learn about you: they watch you operate their services, learn about what you search for, click on, like, or whatever, and develop a profile of you over time.
But that approach, of course, makes many people uncomfortable! Many users are already wary of OpenAI absorbing their questions and missives and returning them to the system as training data to help further customize the bot; The idea of ChatGPT “knowing” users is both interesting and creepy.
OpenAI says it keeps users in control of ChatGPT's memory and has trained the system not to remember sensitive things like information about your health. You can always ask ChatGPT what it knows about you and tell the bot to forget something or delete it in the new Manage Memory section of the settings. OpenAI is also launching Temporary Chat as a sort of incognito mode: a way to have a quick conversation without affecting ChatGPT's memories of you. You can also disable it completely for your entire account.
By default, memory will be on and OpenAI says that memories will be used to train your models in the future. (Businesses using ChatGPT Enterprise and Teams will not send their data to the models.)
For now, the memory is just a test, open to a “small portion” of users, the company said on its blog announcing the feature. But it's easy to imagine how quickly this could become a central part of how we interact with ChatGPT, for better or worse. Robots are getting smarter and they are getting to know us very quickly.
