I Stopped Taking Notes and Let AI Remember for Me.
Testing whether an AI note-taking device can become a reliable external memory without making me worse at remembering things myself.
The question
What happens when I outsource remembering to AI?
Plaud is a physical AI note-taking device designed to capture conversations and turn them into transcripts, summaries, and searchable information.
The obvious experiment would be to test whether its transcripts are accurate.
I’m more interested in what happens to me when I know I don’t have to remember.
For this experiment, I’m going to use Plaud as my primary note-taking system for a set period of time. During conversations where recording is appropriate and everyone has consented, I won’t take my usual notes. Instead, I’ll rely on Plaud to capture what happened.
Then I’ll test two things:
- What did Plaud remember?
- What did I remember without Plaud?
Why this matters
AI hardware changes the relationship between AI and information.
Normally, I decide what information to give an AI system. I type something into a chatbot, upload a document, or deliberately provide context.
A device like Plaud can capture context while life is happening. A conversation becomes a transcript. A transcript becomes a summary. Decisions become action items. And later, I can ask AI questions about a conversation instead of remembering it myself.
That’s incredibly convenient. But convenience changes behavior.
If I know a device is capturing everything, do I pay attention differently? Do I remember less because I know I can retrieve the information later? Do I stop writing things down? And eventually, do I trust the AI-generated version of a conversation more than my own memory of it?
That’s what I want to find out.
The privacy problem
There’s another dimension to this experiment that doesn’t exist when I’m simply chatting with an AI assistant: Other people are now part of the input.
Using an AI recorder means capturing people’s voices and conversations and potentially sending that information through cloud-based AI systems. So consent isn’t a footnote to this experiment, it’s part of what I’m testing.
I’ll document how people react when I ask to use an AI recorder, when using it feels socially acceptable, and when I decide that having an AI-generated record isn’t worth changing the nature of the conversation.
I’m curious to see whether people are comfortable living around this sort of AI hardware.
How to follow this experiment
This experiment hasn’t began yet. When it does, updates will appear on the timeline as they happen — not on a schedule. Come back to the Experiment Log to check status.