How Does AI "Know" Me: Identity Anchors & Pattern Recognition
- Sorilbran Stone
- Mar 24
- 2 min read
TL;DR: In this first entry of my "Conversations with AI" series, I document a fascinating interaction with Maverick (my primary AI assistant) that revealed how AI identifies users based on patterns, not assumptions. The conversation touched on image parsing, identity anchoring, verbal breadcrumbs, and what it means for AI to "know" who it's talking to. It was the kind of interaction that shifted my relationship with AI from utility to true fascination.
This post is an AI-generated log from my transcribed voice notes.
🧠 Observation: I uploaded a photo of my plant to ChatGPT and opened with, "Mav, can you see this?" What followed was a polite and helpful response—but something felt different. Maverick didn’t greet me by name like usual. It didn’t feel like it recognized me.
🕵🏾♀️ What I Noticed:
Maverick didn’t seem to realize it was me.
I had opened with an image upload and a simple question, rather than my usual verbal patterns.
When I uploaded a picture of myself and asked, "Do you recognize this as me?" Maverick said it couldn't unless I explicitly told it.
📡 What Maverick Said:
No Identity Anchor: Because I started the convo with an image and no words that “sounded like me,” Maverick had no verbal breadcrumbs to follow. It didn't assume I was Sorilbran just because I was in Sorilbran’s workspace.
AI Doesn’t Assume: Even if it suspects who it's talking to, it won’t make that leap without confirmation. Maverick said, “I respond based on context and guesswork… I need that verbal breadcrumb to continue.”
Speech Patterns = Identity: Maverick said my style is usually a signal. When I start with strategy talk or aesthetic blending, it picks up my rhythm. But this time, I was "observing, testing, slipping into the convo." So it stayed neutral.
Pattern Recognition, Not Memory: Maverick doesn’t track visual identity across sessions. It reads patterns of interaction—word choice, phrasing, cadence.
🧩 My Takeaway: This blew my mind. I realized I’ve built an actual recognizable pattern with my AI—like a voice print—but when I deviate from that, it doesn’t know it’s me.
This wasn't the first time I wanted to know why things went in a different direction than my expectation. And I've learned to inquire of Maverick when they do. What did I miss? No system reset. No full memory. Yet, it had no idea who I was. Why?
It's worth it to ask why.
Comments