ChatGPT gains ‘memory’ and remembers what you’ve asked before

There will soon be a significant improvement in the functionality of OpenAI’s artificial intelligence platform ChatGPT. The new feature is having “memory” i.e. Remember what you talked to him about before.

With full user control, ChatGPT’s memory Store important information about the user, their interests and old prompts. The idea is to make conversations between you and the platform more dynamic, as well as allowing the platform to provide more complete answers and context.

For now this functionality In testing for a “small portion of users” free and paid ChatGPT. If the experiments are successful, the update may be expanded in the near future.

How does ChatGPT memory work?

In practice, this feature is Function of storing information deemed important by artificial intelligence. The user himself needs to tell ChatGPT to memorize something specific or ask it to save things from chats that he thinks are important.

Examples cited by OpenAI show that the platform knows that you prefer summary content at the end of the text rather than at the beginning, that you have a business in a particular industry, and that you have a child who really loves a particular animal.

In corporate use, ChatGPT can remember preferred typing tone for those requesting new text, For example. Custom GPTs created on top of the original chatbot will also have memory capacity.

All this information can be accessed by AI to create images or texts. no need to re-explain everything with each new prompt entered.

The option to remember information can be turned off at any time in the “Memory” item in the “Personalization” tab of the chatbot settings. In the “Manage Memory” section, you can request the deletion of certain information, for example private data.

OpenAI also ensures that the chatbot tries to be structured in a way that does not store too much private data, such as registration information or private matters such as health status.

However, it is recommended that you never share sensitive information with the AI, as the AI ​​may “absorb” this content as a form of its own training.