OpenAI has begun testing a new feature for ChatGPT - memory, which allows information from previous conversations to be saved. This innovation aims to simplify communication, reduce the need to repeat information, and enhance the relevance of responses. Users have full control over ChatGPT's memory: they can specify what should be remembered, find out what is already stored, and delete unnecessary information through dialogue or settings.
At this stage, the feature is available only to a limited number of ChatGPT users, including both free and Plus account holders, as part of the initial testing. OpenAI plans to expand access to memory soon and share plans for its further implementation.
ChatGPT's memory is designed to improve over time and with increased interaction, becoming better at adapting to your preferences and needs. For example, if you indicate that you own a coffee shop, ChatGPT will use this information to form more relevant suggestions in the future.
Users can disable the memory feature in settings at any time and manage stored information: view, delete individual memory fragments, or clear it entirely. A temporary chat is offered for conversations that should not be saved in memory and will not be used for model training.
OpenAI has updated its privacy and security principles considering the introduction of the memory feature and is taking steps to minimize risks and biases, as well as to protect users' sensitive data.
🔗 Link: https://openai.com/blog/memory-and-new-controls-for-chatgpt
#ai #agi #gpt #llm
