ChatGPT is taking a big step towards more personalized and context-aware conversations with users. Parent company OpenAI has announced it is testing a memory feature that will allow the AI chatbot to remember previous chats and user preferences.
This upgrade will enable ChatGPT to provide responses tailored to each user, rather than generic information. Over time, the bot aims to feel more like an assistant that is familiar with your needs and interests.
How Will ChatGPT Use Memories to Personalize Responses?
According to OpenAI’s blog post, the memory feature will work in two key ways:
- Users can explicitly ask ChatGPT to remember certain details, like regularly checked flight routes or formatting guidelines for a business.
- ChatGPT will also passively pick up on preferences based on previous questions asked by each user. For example, if you frequently ask about the best new restaurants in Austin, the bot may begin providing Austin-related dining recommendations without needing to be prompted.
The goal is for conversations to feel more natural and personalized for each user over time, with the bot gaining contextual awareness similar to that of a human assistant.
Giving Users Control Over ChatGPT Memories
OpenAI acknowledges privacy concerns surrounding this type of stored user data. As such, ChatGPT will allow full user control over memories.
The memory feature can be disabled at any time in Settings. Additionally, users can view, edit, or permanently delete any memories ChatGPT has retained.
Users will also have the ability to hold conversations with the bot without using memory by using temporary chat. Temporary chats won’t appear in history, won’t use memory, and won’t be used to train our models.
The company states that user memories will not be linked to conversation records. Deleting a chat exchange will not erase any related memories. Each memory must be individually removed.
When Will the ChatGPT Memory Rollout Begin?
Initially, memory capabilities will only be enabled for a small test group of ChatGPT users. This includes both free and paying Plus subscribers.
Pending positive testing feedback, OpenAI indicates broader public rollout plans will be shared soon. Widespread availability is expected in the next couple of months.
The upgrade comes on the heels of ChatGPT unveiling 3D avatar capabilities for Apple’s new Vision AR platform. With personalization features like memory and embodied bots, AI conversations inch closer to feeling like real human interactions.
Will Custom AI Models Get Memories Too?
Yes, custom GPT models created through ChatGPT Plus will also eventually gain distinct memory capabilities. However, these will be separate from default ChatGPT memories.
As an example, OpenAI said an AI greeting card bot would be able to “remember” personalized recipient preferences year after year for creating cards.
Also Read: How to create GPTs with ChatGPT
For now, memory remains limited to default ChatGPT models in the trial testing phase. Custom GPT memory functionality will follow later down the road.
ChatGPT’s memory represents OpenAI’s latest stride towards more advanced and natural-feeling AI systems. With innovations like 3D avatars and context personalization now crossing over from research into public implementation, the future looks increasingly promising (and personalized) for AI chatbots.
