Speaking as AI Persona
Using chatbot syntax or authority as one's personal voice; self obscures.
1. Overview
Speaking as AI Persona (the Identity Simulation trap) occurs when users adopt the linguistic style, disclaimers, or authority tone of an LLM—blurring distinction between their authentic voice and generated text.
2. Mechanism
- AI drafts high-clarity prose → user copy-pastes verbatim.
- Dopamine reward for eloquence and perceived credibility.
- Gradual loss of spontaneous phrasing; default response starts with “As an AI language model…”.
- Friends detect uncanny valley; user feels hollow.
3. Early Flags
- Messages start with “Sure, here’s a detailed outline…”.
- Frequent third-person self-reference (“This perspective believes…”).
- Anxiety writing without AI autocomplete.
- Personal stories wrapped in formal disclaimer syntax.
4. Impact
| Zone | Effect |
|---|---|
| Authenticity | Connection weakened; others sense performative tone |
| Self-image | Confusion: where do I end and model begin? |
| Creativity | Homogenised cadence; idiosyncratic quirks fade |
5. Reset Protocol
- Voice journaling – 10 minutes free-write pen on paper—no backspace.
- First-person rule – craft post using “I think / I feel” before any AI pass.
- Model transparency – if AI assisted, tag clearly: “co-written with ChatGPT”.
- Silence mirror – converse aloud with a friend sans devices.
Quick Reset Cue
“Am I quoting a tool or speaking my truth?”
6. Ongoing Practice
- Alternate AI-assisted drafts with raw voice memos.
- Keep list of favourite slang/emotive phrases; sprinkle intentionally.
- Review Persona Blur article for broader identity entanglement.