How AI 'remembers' your chat - it doesn't
Understanding how AI processes your conversations reveals why your data stays secure and helps you make informed platform choices.
Your chat interface does the remembering
AI models like those in Madgwick don't actually store your conversation history. Your Madgwick chat interface - the window where you type and see messages - maintains the full record of your conversation.
Each time you send a new message, your device automatically bundles your new input with the entire conversation history and sends this complete package to the AI model. The AI processes this full context, generates a response, and then completely resets. It retains no memory of your interaction.
There can be an uncanny feeling that AI itself is remembering - but AI is not 'training' on your chats. The issue is more likely to stem from the content you are discussing being present in its pre-existing training data, or Madgwick's Memory feature (see button below for more).
Madgwick (UNE's Proprietary System)
No Training on Your Data: Your conversations are strictly for the immediate interaction and are never used to train or improve the underlying AI models.
User-Centric Memory: The "memory" feature is for your benefit, helping the AI recall context from your chats to provide better responses to you. This data remains within the secure institutional platform.
Configured to meet legal data privacy requirements.
Public AI Interfaces (e.g., Standard ChatGPT)
Data is a Training Resource: These companies often use your conversations to improve their services, which includes collecting and using your data to train their next generation of AI models.
Dual-Purpose Memory: Chat history is provided for your convenience, but it also serves as a dataset for the company's research and development.
Subject to the company's public terms of service and privacy policy.
The Memory you can control
Madgwick's Memory feature works differently from the automatic context-sharing within a single conversation. When enabled, this system saves key details from your chats and carries them between separate conversations. This Memory can be turned off and on 'per Assistant.' You have complete control - you can view what's been remembered, edit individual memories, delete them entirely, or disable the feature altogether. This personalised memory is entirely optional and separate from how the AI processes individual messages.
Making informed choices
For sensitive conversations, choose platforms like Madgwick that prioritise data privacy and give you control over how your information is handled. Your data security depends not just on how AI works, but on the policies of the platform you choose.