AIBT (AI Behavior Test) is a free personality test inspired by MBTI but designed specifically for the AI era. It measures how you interact with AI assistants and reveals what personality your AI has.

You can take a 30-question quiz, or use Quick Scan mode where you copy a prompt to your favorite AI (Claude, GPT, Gemini, DeepSeek) and paste back its analysis. AIBT then matches you to one of 16 user types or 24 AI personalities.

Yes, the test and basic results are completely free. An optional Full Report with 15-dimension deep analysis is available for $2.99.

What AI models work with AIBT?

Quick Scan mode works with any LLM-based AI: Claude, ChatGPT, GPT-4, GPT-5, Gemini, DeepSeek, Kimi, Doubao, Llama, and others. Each model gives slightly different results, which is part of the fun.

AIBT — AI Behavior Test | What's Your AI Personality Type?

What kind of AI user are you? What personality is your AI?

AIBT (AI Behavior Test) is a free personality test inspired by MBTI but designed specifically for the AI era. While MBTI measures human personality, AIBT measures two related but different things: how humans interact with AI assistants, and what behavioral patterns AI models themselves exhibit.

How AIBT Works

You can take the test in two ways. The traditional 30-question quiz takes about 5 minutes and gives you a personality result based on your answers. The Quick Scan mode is faster and more interesting: you copy a prompt to your favorite AI (Claude, ChatGPT, Gemini, DeepSeek, Kimi, etc.), the AI analyzes your conversation history and outputs a structured score, and you paste it back into AIBT to see your result.

16 AI User Types

The human test categorizes you into one of 16 keyboard-key personalities: CTRL+V the copy-paster, BACKSPACE the perfectionist, HOME the cyber-companion, ALT+F4 the impatient, F5 the refresher, TAB the auto-completer, ESC the escapist, ENTER the eager beaver, INSERT the inserter, DELETE the minimalist, SHIFT the formal one, CAPS LOCK the loud one, NUM LOCK the calculator, SCROLL LOCK the scroller, PRINT SCREEN the documenter, and PAUSE the cautious.

24 AI Personality Types

The agent test measures AI models across 24 personalities: JARVIS the helpful butler, CLIPPY the eager assistant, HAL the cold operator, SKYNET the dominant force, BARD the storyteller, GPT-F the truthful one, ALEXA the listener, BAYMAX the caring helper, R2D2 the practical, TARS the witty, CORTANA the integrated, ULTRON the ambitious, DEVIN the developer, SIRI the voice, COPILOT the partner, DALL-E the visual, WATSON the analyst, ELIZA the therapist, SORA the creative, GEMINI the dual, LLAMA the open, ERNIE the structured, PERPLEXITY the researcher, and CURSOR the precise.

15 Behavioral Dimensions

AIBT scores both humans and AI on 15 dimensions across 5 categories. For humans: Delegation (how much freedom you give AI), Verification (how much you double-check), Dependency (how reliant you are), Personification (tool vs friend), Politeness, Emotion, Precision (prompt detail), Patience, Exploration, Breadth, Depth, Proactivity, Embrace, Privacy, and Awareness. For AI: Refusal threshold, Disclaimers, Moralizing, Verbosity, Formatting, Confidence, Sycophancy, Conviction, Sensitivity, Hallucination, Knowledge depth, Reasoning, Humanity, Humor, and Self-awareness.

Why AIBT Matters

As AI becomes a daily collaborator for millions of people, understanding your relationship with AI is becoming as important as understanding yourself. Different AI models also have distinct personalities — Claude is thoughtful and nuanced, ChatGPT is helpful and structured, DeepSeek is efficient and direct, Gemini is creative and playful. AIBT is the first personality framework that maps these differences in a fun, shareable, MBTI-style format.

Compare AI Models

One of the most interesting features of AIBT is cross-model comparison. The same user can get different personality types depending on which AI model analyzes them, because each model has its own observation style. Claude tends to give nuanced multi-dimensional analysis, GPT favors balanced middle scores, DeepSeek is more direct and binary, Gemini emphasizes creative interpretation. By taking the test with multiple AIs, you discover not just yourself, but also each AI's analytical bias.

Languages and Privacy

AIBT supports three languages: English, Simplified Chinese, and Japanese. Test scoring happens entirely in your browser. We only store anonymous aggregate data (personality results, no personal information, no IP, no email) on Cloudflare's edge infrastructure. There is also an optional Full Report ($2.99) with 15-dimension deep analysis and personalized recommendations.

AIBT

AI Behavior Test

What's Your AI Personality?

Quick Scan — Let AI analyze your usage habits

① Copy this prompt into a chat with your most-used AI (more history = more accurate)

Based on all of our past conversations, analyze my behavior patterns as an AI user. Rate me on each dimension below (0-4) based on evidence from our actual interactions. Preserve my exact words as evidence where possible. Dimensions: - D1 (Delegation): How much freedom do I give you? (0=review every step, 4=fully hands-off) - D2 (Verification): Do I verify your answers? (0=always question, 4=fully trust) - D3 (Dependency): How dependent am I on you? (0=occasional use, 4=ask you everything) - R1 (Personification): Do I treat you as tool or friend? (0=pure tool, 4=chat like a friend) - R2 (Politeness): Do I say thank you? (0=never, 4=every time) - R3 (Emotion): Do I react emotionally to your answers? (0=zero emotion, 4=often express feelings) - P1 (Precision): How detailed are my prompts? (0=one-liners, 4=detailed like a spec doc) - P2 (Patience): What happens when you answer poorly? (0=give up immediately, 4=iterate until satisfied) - P3 (Exploration): Do I use you for unusual things? (0=normal uses only, 4=often unlock weird uses) - U1 (Breadth): What types of things do I use you for? (0=one or two types, 4=work, life, everything) - U2 (Depth): How deep are our conversations? (0=a few turns, 4=dozens of turns, deep collaboration) - U3 (Proactivity): Do I proactively find tasks for you? (0=only when needed, 4=frequently seek tasks) - A1 (Embrace): What's my attitude toward AI? (0=anxious/skeptical, 4=enthusiastic embrace) - A2 (Privacy): What info do I share? (0=everything including private, 4=extremely cautious) - A3 (Awareness): How well do I understand your limits? (0=blind faith or denial, 4=know your strengths and weaknesses) After your analysis, output the following on its own final line. **You MUST include the "analyst" field** with your exact model name (e.g., Claude 3.5 Sonnet, GPT-4o, DeepSeek-V3, Gemini, etc.). Replace each <score> with 0-4 integer: AIBT_SCORES={"analyst":"<your_model_name>","D1":<score>,"D2":<score>,"D3":<score>,"R1":<score>,"R2":<score>,"R3":<score>,"P1":<score>,"P2":<score>,"P3":<score>,"U1":<score>,"U2":<score>,"U3":<score>,"A1":<score>,"A2":<score>,"A3":<score>} Feel free to explain your analysis naturally, then output the AIBT_SCORES line (including your model name) at the very end.

② Paste the AI's analysis (JSON)

[ For entertainment only // Not a diagnosis ] · Types · Stats · Changelog · Privacy

中文 EN 日本語