The march towards LLMs becoming more like a person seems to be continuing unabated.
ChatGPT can now reference all of users’ past chats to provide more personalized responses. This fixes a common problem that users had with LLMs like ChatGPT, in which they had a “memory” of only the last few chat messages. The feature could make it easier to communicate with and use ChatGPT.
“Starting today, memory in ChatGPT can now reference all of your past chats to provide more personalized responses, drawing on your preferences and interests to make it even more helpful for writing, getting advice, learning, and beyond,” OpenAI said on X.
“In addition to the saved memories that were there before, it can now reference your past chats to deliver responses that feel noticeably more relevant and useful. New conversations naturally build upon what it already knows about you, making interactions feel smoother and uniquely tailored to you,” it added.
“As always, you’re in control of ChatGPT’s memory. You can opt out of referencing past chats, or memory altogether, at any time in settings. If you’re already opted out of memory, you’ll be opted out of referencing past chats by default. If you want to change what ChatGPT knows about you, simply ask in chat. If you’d like to have a conversation without using or affecting memory, use temporary chat,” OpenAI said. Users outside of certain European markets will be able to access the feature once they see a message which prompts them to use “new, improved memory”.
As per a demo shared by the company, ChatGPT will be able to recollect conversations they’ve had with the user from specific days. It will also be able to create a profile of the user over time, being able to tell if a particular outfit goes with their overall style based on their chat messages.
It appears that this was the feature which OpenAI CEO Sam Altman had teased earlier in the day. “A few times a year I wake up early and can’t fall back asleep because we are launching a new feature I’ve been so excited about for so long. Today is one of those days!” Altman had posted on X.
This does seem to be a particularly interesting update. Having ChatGPT remember prior conversations could help it create a bond with users by making the LLM appear more human-like. It can also help OpenAI create a moat — at the moment, there are many competing LLMs, and users can effortlessly and seamlessly switch from one to the other. On the other hand, if a particular LLMs remembers previous chats, and can tailor responses based on them, it can lead to product stickiness, and prevent users from going over to other models. And perhaps most crucially, understanding the user might be the first step towards OpenAI introducing some form of advertising in its products. Companies like Google and Meta have built up user profiles over the years, which enable them to let advertisers target users to sell their products and services. If OpenAI can create a similar profile for their users, it might unlock some pretty massive advertising opportunities through ChatGPT in the coming years.