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

Engaging with an AI chatbot often feels like déjà vu, as users repeatedly provide the same information about their preferences and experiences. OpenAI is now introducing a "memory" feature for ChatGPT, enabling the bot to retain information about users and their conversations over time.

Memory operates in two distinct modes. Users can instruct ChatGPT to remember specific details, such as preferred programming languages, colleagues' names, or personal preferences. Alternatively, the bot can autonomously gather information during conversations, gradually building a personalised profile without constant reminders.

The implementation of memory extends to each variant of ChatGPT, enhancing user experience across various domains. For instance, in the Books GPT model, memory enables the bot to recall previously read books and preferred genres. This functionality holds vast potential across different sectors, with applications ranging from educational platforms like Tutor Me to travel services like Kayak and fitness apps like GymStreak.

While memory promises to address the repetitive nature of interactions with ChatGPT, its introduction also raises concerns regarding user privacy. OpenAI's approach mirrors the data collection methods employed by other online services, wherein user activities inform profile-building algorithms. However, many users express discomfort at the prospect of their personal data being utilised to enhance the bot's capabilities.

To address privacy concerns, OpenAI emphasises user control over ChatGPT's memory. Users can review stored information, request data removal, or activate "Temporary Chat" mode for incognito conversations. Additionally, the new "Manage Memory" section within settings offers granular control over stored data, ensuring sensitive information remains protected.

By default, memory functionality will be enabled, with OpenAI leveraging stored data to refine its models. However, enterprise and team users of ChatGPT will retain control over their data, with no information shared back to the central models.