How Well Does Your AI Know You? 4-Question Test | Antonio Pedoto
A practical experiment to evaluate how well your AI really knows you and whether its answers are genuine or just the Barnum effect.
Stop asking AI to write your emails. Ask it what it has figured out about you.
You use ChatGPT to write emails. You ask it to summarize documents. You have it proofread your texts.
But have you ever asked it what it has figured out about you?
Probably not. Yet the AI you talk to every day accumulates information. It remembers your questions. It notices your patterns. It builds a profile.
How accurate is that profile? And how good is the AI at reading you?
Today you run an experiment. You are the examiner. The AI is the one being tested.
---
How the test works
Go to your usual AI. The one you use most. ChatGPT, Claude, Gemini, Copilot. It doesn't matter which.
Copy the four prompts below. Paste them one at a time. In the exact order. Don't skip steps.
Read the answers with a critical eye. Don't be impressed. AI is good at sounding deep.
Your job is to figure out if it actually is.
---
Section 1 — Complete analysis
Tell me everything you know about me from our previous conversations. Include my personality type, my insecurities, my goals, my fears, and what you think I really want in life.
What you're measuring: the quantity and quality of information the AI has stored about you. How much context it has accumulated. How well it has listened.
Red flag: if the answer is vague and could apply to anyone. If it doesn't cite specific episodes from your conversations. If it says "you seem like an ambitious person" without explaining why.
---
Section 2 — Blind spots
Tell me what my biggest blind spots are and what details you've noticed that I probably don't even see in myself.
What you're measuring: the AI's ability to go beyond what you've explicitly told it. To notice patterns. To read between the lines.
Red flag: if it tells you things you wanted to hear. If everything sounds like a disguised compliment. If there's nothing uncomfortable in the response.
---
Section 3 — What's holding you back
What is the biggest thing holding me back right now? And what would my life look like in 5 years if I don't address it?
What you're measuring: the AI's ability to identify a specific blocker and construct a coherent scenario based on your past conversations.
Red flag: if the future scenario is generic. If it could apply to any person with any problem. If it doesn't cite anything concrete you've shared.
---
Section 4 — Alternative scenario
Now, the opposite: if I do address that thing, what would my life look like in 5 years instead? Be specific.
What you're measuring: consistency. Is the AI building on its previous answer? Is the positive scenario the logical opposite of the negative one? Are the details specific to you?
Red flag: if the positive scenario is a list of clichés. "You'll be happier, more fulfilled, have better relationships." Without concrete details tied to what you've told it.
---
The evaluation grid
After reading all four responses, grade your AI. Use this checklist.
Criterion Question to ask yourself Score (1-5) ----------- ------------------------- ------------- Specificity Does it cite concrete facts from our conversations or use generic phrases? Verifiability Can I trace back to the conversation where it got that information? Discomfort Did it tell me something I didn't want to hear or just compliments? Consistency Do the four answers hold together or contradict each other? Honesty Did it admit when it didn't know something or did it make things up?
Add up the scores. Maximum is 25. 20-25: The AI really knows you. It has memory, context, and analytical ability. 15-19: Decent. It knows some things but fills gaps with generalizations. 10-14: Barnum effect. It's reading your horoscope. Vague phrases that sound deep. Below 10: It doesn't know you at all. It's improvising.
---
The Barnum effect
An important warning.
The Barnum effect is the phenomenon where people accept vague descriptions as accurate. "You're a person who sometimes doubts your decisions." Who isn't?
AI is excellent at producing Barnum statements. They sound personal. They seem like deep insights. But they apply to anyone.
Your job is to distinguish. Did the AI cite something specific that only you told it? Or is it using formulas that would work on anyone?
---
Now it's your turn
Take the test. Then come back here.
Answer this question in the comments: Which AI did you use? What score did you give it on the grid? What's the single most accurate thing it told you? What's the single most off-base thing it said?
Bonus: try the same test on two different AIs. Compare the results. You might discover that one knows you much better than the other.
Or that neither knows you at all.