ChatGPT is getting ‘memory’ to remember who you are and what you like

OpenAI introduces memory for ChatGPT, enabling user-specific recall and potentially personalized interactions. Privacy concerns arise, but user control tools are offered. The impact on future GPT interactions remains to be seen.

ChatGPT is getting ‘memory’ to remember who you are and what you like

Chatbot interactions can become repetitive, prompting OpenAI to introduce "memory" for its ChatGPT. This allows the bot to remember user preferences and conversation history, aiming for a more personalized and seamless experience.

Memory operates through explicit user instructions or implicit information gathering during interactions. Each custom GPT, like Books GPT, will retain its own user-specific memory. This holds potential for various applications, such as personalized learning with Tutor Me, travel recommendations with Kayak, or fitness tracking with GymStreak.

A screenshot of ChatGPT’s memory feature.
Image: OpenAI

However, concerns regarding user privacy and data collection are inevitable. While OpenAI emphasizes user control and sensitive information exclusion, the concept of ChatGPT "knowing" its users can be intrusive.

To address these concerns, OpenAI offers memory management tools, including memory removal, "Temporary Chat" mode, and complete opt-out options. Users also retain the right to inquire about stored information. By default, memory is active and contributes to model training, with enterprise users exempted from data contribution.

Currently, memory is a limited test feature, but its potential impact on future GPT interactions is significant. As bots evolve and gain insight into users, navigating the balance between personalization and privacy will be crucial.