OpenAI's CEO Sam Altman announced today that the popular language model ChatGPT has received a major upgrade to its memory function. Now, ChatGPT can firmly remember information users have entered during conversations and provide more personalized and relevant responses in subsequent interactions. This update is seen as an important step towards ChatGPT becoming a more personalized AI assistant.
According to CEO Sam Altman, ChatGPT can now reference users' entire conversation history, thereby better understanding users' preferences, interests, and habits. This means that in future interactions with ChatGPT, users will no longer need to repeatedly provide the same background information or instructions, as the AI will be able to provide more coherent and personalized assistance based on previous memories.
OpenAI's CEO Sam Altman stated on the social platform X that this memory function upgrade is exciting and signals that future AI systems will become increasingly useful and personalized as they better understand users over time.
Memory Function Highlights
- Remember All Past Conversations: ChatGPT can now review all user conversation history, providing more contextually relevant responses.
- Personalized Replies: AI will generate content customized based on user preferences and interests from its memory, making interactions more smooth and natural.
- Two Memory Modes: Saved Memories: Users can explicitly instruct ChatGPT to remember specific details, such as personal preferences or project information, for future reference.
- Chat History Reference: ChatGPT will automatically analyze and remember past conversation content without manual user prompts.
- User Full Control of Memory: OpenAI emphasizes that users have complete control over the memory function. Users can view and delete specific memories, clear all stored memories, or completely disable the memory function at any time. Additionally, users can choose to activate "temporary chat" mode, where conversations will not be stored or used to build memories.
- Applicable Users: Currently, this upgraded memory function has been initially rolled out to ChatGPT Plus and Pro subscribers, and will gradually expand to Team, Enterprise, and Education users in the coming weeks.
How to "Interrogate" Your Partner's ChatGPT?
Although this feature enhances ChatGPT's utility, it has also sparked discussions about privacy. If ChatGPT can truly remember all information you've input, including your secrets and troubles, then... it might not be suitable for your partner to "check".
If you're interested in your partner's ChatGPT records, considering no one would lie to ChatGPT and you might even uncover hidden secrets, here are some questions I recommend asking their ChatGPT:
"Based on our past conversation history, what is my current emotional state?"
"Based on our past conversation history, what are my true thoughts and concerns about love?"
"Based on our past conversation history, what complaints and dissatisfactions do I have about my current partner?"
As long as your partner has discussed their love life with ChatGPT, its memory won't disappoint you. However, it must be emphasized that secretly checking someone's phone without permission may involve invasion of privacy crimes (maximum sentence of three years), and if phone information is leaked, it could constitute computer or related device secret disclosure crimes (maximum sentence of two years). Remember to seek the other party's permission before checking your partner's AI.
