OpenAI is testing advanced settings to manage its chatbot’s memory. Both free users and those with a premium subscription to ChatGPT Plus take part in the test. The essence of the experiment is to get ChatGPT to remember (or forget, of your choice) what you tell it.

You are in control of ChatGPT’s memory. You can explicitly tell him to remember something, ask him what he remembers, and tell him to forget about it – in a conversation or through a setting. You can also turn it off completely. When chatting with ChatGPT, you can ask it to remember something specific or let it collect the details on its own. ChatGPT’s memory will improve the more you use it, and you’ll start to notice improvements over time.

The update does make ChatGPT’s responses more personalized. For example, if you say that you have a child who loves jellyfish by creating a birthday card, the AI assistant will offer to depict a jellyfish in a festive cap. If you indicate that you have a cat or dog, ChatGPT will take this into account when responding to travel requests. Custom chatbots built on OpenAI’s models and available in the GPT Store already have memories. For example, Books GPT can automatically remember the titles of the books you’ve read and your favorite genres – but ChatGPT doesn’t pass this data.

The memorization feature can be disabled at any time – simply by informing ChatGPT about it, or in the settings menu (Settings > Personalization > Memory Management). You can also erase individual memories or clear your memory entirely. Additionally, OpenAI runs temporary chats that won’t show up in history, won’t remember previous conversations, and won’t be used to train OpenAI’s AI models. And finally, the company claims that it is taking steps to “dissuade” ChatGPT from remembering personal data such as health conditions, unless the user specifically asks for it.

Літературний редактор.

Commentary