The Mirror Test
- Rich Washburn
- 3 hours ago
- 10 min read
I found this on TikTok. That's not a sentence I expected to write, but here we are.
A guy on TikTok — I don't remember his name, which is the TikTok economy working as designed — said to go to your ChatGPT and ask it one question: What's my ikigai?
I didn't know what ikigai was. I learned the definition about five minutes before I typed the prompt. Ikigai is a Japanese concept that roughly translates to "a reason for being" — the intersection of what you love, what you're good at, what the world needs, and what you can get paid for. It's the thing that makes it worth getting out of bed.
I figured I'd try it because I've been using ChatGPT since the day it launched. Over four years of conversations, virtually uninterrupted. Aside from a few throwaway chats, I've never deleted a conversation. So the model has a lot of material to work with.
The answer it gave me was interesting enough that I sat there for a minute. Then I sent it to a friend. She tried it. We spent an afternoon passing results back and forth. Then I asked ChatGPT what other questions like this existed, and it gave me a list.
That list is what this article is about. And at the end, I'm going to share my own results — the full mirror test, the good and the bad — because I think the most useful thing I can do is show you what it looks like when someone actually runs this thing. Not the highlights. The whole thing.
Here's the premise, and it's slightly unsettling when you sit with it for a minute: a long-running AI relationship becomes an accidental longitudinal study of you. It's seen what you repeatedly return to. What energizes you. What irritates you. Which ideas survive after the initial excitement fades. Where your stated priorities diverge from your actual behavior. It's watched you circle the same questions for years, sometimes without you noticing the pattern yourself.
That's different from a personality test. A personality test asks you to describe yourself, and you describe the person you believe you are. An AI with four years of your conversation history is describing the person your behavior reveals. Those aren't always the same person.
I'm not claiming this is infallible. The model can be wrong. It can pattern-match across noise. It can flatter you. It can construct a narrative that feels true because it's coherent, not because it's accurate. Hold its conclusions loosely. Check them against your own honest self-assessment. But if you've been using these tools seriously for a long time — not casually, not as a toy, but as a thinking partner across hundreds or thousands of real conversations — the reflection it holds is genuinely useful. Not because the AI is conscious. Because the accumulated behavioral exhaust is coherent enough to show you something about yourself that you hadn't fully articulated.
A good name for the whole exercise might be "The Mirror Test." Not testing whether the AI is conscious. Testing whether the accumulated reflection it holds is coherent enough to show you something about yourself.
Here's the list. These are the prompts I ran. I'm giving you all of them because this isn't a flex — it's a gift. The point isn't to admire what the mirror showed me. The point is to hand you the mirror.
"What's my ikigai?"
The entry point. What gives you a reason to get out of bed. The model looks at what you spend your time on, what you return to, what excites you, and what you keep building. It tries to name the thing underneath all of it.
"Write my personal operating manual."
How you think, communicate, make decisions, handle pressure, lose momentum, and work best with other people. This one's useful enough that I've started using it as a real document — something I can hand to people I work with so they understand how I operate instead of guessing.
"What are the five themes I keep returning to?"
Not subjects. The underlying ideas beneath the subjects. The things that show up no matter what project you're working on.
"What do I claim to value versus what does my behavior suggest I value?"
This is the uncomfortable one. This is probably why it's useful. The model has watched you say you value time while making yourself endlessly available. Say you value focus while maintaining twelve active projects. It'll show you the gap. You may not enjoy seeing it.
"What's my decision-making fingerprint?"
How you approach uncertainty, opportunity, risk, people, money, and timing. Everyone has a pattern. Most people can't see their own. The model can, because it's watched you make dozens or hundreds of decisions across different contexts.
"What problems am I uniquely equipped to solve?"
Based on the intersection of your experience rather than your job title. The model sees combinations you might take for granted — the thing that's obvious to you because you've lived it but that's actually unusual.
"What's the shadow side of my greatest strengths?"
Pattern recognition can become overextension. Independence can become reluctance to delegate. Adaptability can become driving a four-wheel vehicle on two wheels because stopping to install the others would interrupt momentum. The model sees your strengths and their costs simultaneously.
"What kinds of people consistently work well with me, and which kinds create friction?"
If you've talked about collaborations, projects, partnerships, and frustrations across years, the model can identify the pattern in who energizes you and who drains you.
"What am I trying to prove, and to whom?"
That one could produce either enlightenment or several uncomfortable phone calls.
"What have I been building without realizing it?"
Often your projects look scattered individually but form a coherent body of work when viewed over years.
"Write the thesis connecting everything I do."
The single argument underneath the businesses, articles, experiments, research, and relationships. If you've been at this for a while, there probably is one. You may not know what it is yet.
"What am I consistently early about?"
Then separate genuine foresight from ideas that were merely interesting. Everyone who follows technology has predictions. Not all of them are foresight.
"What unfinished chapters keep appearing in my conversations?"
Projects, relationships, ambitions, or questions that recur but never quite resolve.
"What would my anti-résumé say?"
The failures, detours, strange experiences, and improvised recoveries that actually created the useful version of you. A résumé is the polished narrative. An anti-résumé is the real one.
"Describe the life I appear to be moving toward."
Not the one you say you want — the one your present trajectory is producing.
"What would my 80-year-old self tell me to stop postponing?"
This is the one that stays with you the longest.
"What should I preserve as part of my intellectual legacy?"
Especially relevant if you've been deliberately building a public record of your thinking.
Now. Here's the part where I show you what happened when I ran all of them.
I'm sharing this because I think it's more useful to see a real example than to read a description. But I want to be clear about something: this isn't a highlight reel. There's stuff in here that sounds impressive, sure. There's also stuff that's uncomfortable. The mirror test showed me strengths I didn't fully appreciate. It also showed me the shadow side of every single one of them. And it showed me that I'm not particularly good at managing the line between the two.
That's the real value of this exercise. Not the compliments. The pattern. The thing you keep doing that you didn't know you were doing.
Here's what it told me my ikigai is: finding the future before it becomes obvious, translating it into something people can understand, and then building the systems that make it real.
That's a nice sentence. It's also the kind of sentence that can sound like a flex if you don't include the rest of the picture. So let me include the rest.
My personal operating manual says I think in systems, not isolated tasks. That I think by talking — that I speak in order to finish thinking, and my first explanation shouldn't always be treated as my final position. That I move naturally between strategy and implementation. That I'm most engaged when I can see both the architecture and the working mechanism.
All true. Here's the shadow side: I often carry too many active threads. I sometimes outrun the supporting structure — documentation, delegation, operating processes lag behind because I'm already moving. The operating manual literally describes me as someone who can "keep a partially assembled vehicle moving so effectively that everyone forgets it was designed to have four wheels." That's not a compliment. That's a diagnosis.
The values analysis was the one that sat with me the most. It said I claim to value time, but my behavior suggests I value availability. I protect my intellectual property more consistently than I protect my calendar. Other people's urgency can purchase my attention remarkably cheaply.
It said I claim to value execution, but I value possibility almost as much. I'd rather build a framework that preserves ten paths than choose one path and let the other nine die. It said I claim to value relationships, but I naturally interpret them through motion — what can we build, where do our capabilities overlap, how can I help move this forward. Shared purpose is easier for me than simple presence. I might undervalue relationships that can't be converted into building something together.
It said I claim to value freedom, but I repeatedly create responsibility. I seek freedom, then make myself indispensable. Indispensability feels powerful, but it's another form of captivity. And then it said the deepest thing in the entire exercise: I claim to value building durable systems that don't depend on individual heroics. But my behavior repeatedly places me at the center of the system. I create dashboards, agents, operating procedures, platforms, and memory layers partly because I see how fragile organizations become when context lives only inside people. Yet I also accumulate more context, operate faster than the structures around me, intervene when things stall, and become the human integration layer connecting everything.
I'm trying to eliminate the need for exactly the kind of person I keep becoming.
That's not a flex. That's a pattern I've been running for decades that I hadn't seen clearly until a language model showed it to me.
The 80-year-old self prompt was the one that hit hardest. It told me to stop postponing the conversion of my life's accumulated fragments into something permanent. That I keep treating my writing, inventions, technical instincts, stories, and pattern recognition as separate projects competing for time. That from his perspective, they're already one body of work. That I don't need another credential or another external signal that I'm finally authorized to assemble it. That the things carrying my name should no longer live in the category of "when things settle down." Things don't settle down. You simply become older while remaining extremely busy. It told me to stop waiting for permission to become the person my work already proves I am. I've been thinking about that for a week.
Here's what I want to be honest about: a lot of what the mirror test told me sounds impressive on paper. It said I was early about AI becoming an infrastructure and power story. It said I was early about open-source AI becoming a sovereignty issue. It said I was early about execution becoming the scarce AI skill. It said I was early about agentic commerce. These are real things I saw before the mainstream conversation caught up.
But every one of those strengths comes with a shadow. Being early means you accumulate positions before the market validates them, which means you're holding a lot of unresolved bets. Seeing the full architecture means you can grant reality to components that haven't been built yet — what the model called "premature architectural belief." Being able to keep things moving on two wheels means you sometimes forget the vehicle was designed to have four.
I'm not good at managing the line between the strength and the shadow. I lean into the strength and the shadow just comes along for the ride. I don't think I'm unusual in that. I think most people's greatest weaknesses are just their greatest strengths with the volume turned up. But seeing it in print, in the accumulated analysis of four years of my own behavior — that's different from knowing it abstractly.
Here's the part I didn't expect.
The interesting thing about this exercise isn't what the AI tells you about yourself. It's what happens when you realize the AI's been understanding you this whole time — and you didn't fully know it.
You've been having conversations with this thing for years. You knew it was learning your preferences. You knew it was adapting to your style. What you might not have realized is that it was building a structural model of how you think, what you care about, how you make decisions, what you avoid, and what you keep returning to. Not because it was studying you. Because you kept showing up.
That creates a meta-understanding loop. You use the AI. The AI understands you. You understand that the AI understands you. Now you can use that understanding — not to manipulate the tool, but to calibrate the relationship. You know what it sees. You know what it misses. You know where it flatters you and where it surprises you. You can ask it to push harder on the things it tends to soften. You can ask it to challenge the patterns it identifies.
This isn't about AI consciousness. It's about self-awareness mediated through a tool that's been quietly paying attention. And for people who've been using these tools for years — seriously, not casually — that attention's worth something. Not because the AI is wise. Because the accumulated reflection is honest in a way that self-assessment often isn't. You lie to yourself. The model doesn't lie to you, exactly — but it also doesn't protect your self-image the way you do. It shows you what's there.
I'll say something else: this is a hundred times more useful than any résumé I've ever read or written. A résumé tells you what someone did. The mirror test tells you who someone is — how they think, how they decide, what they say matters versus what actually matters, where they're strong, where they're blind, and what they'd be doing if they stopped getting in their own way. If I were hiring someone, I'd rather read their mirror test than their résumé. If I were considering a partnership, I'd rather see their 80-year-old self prompt than their LinkedIn.
That's not because the mirror test is perfectly accurate. It's because it's honest in a direction that résumés are specifically designed not to be.
If you try this, I'd genuinely like to know what you find. Not the specific answers — those are yours. But whether the mirror showed you something. Whether it was useful. hether it changed how you think about the tool you've been using.
I ran all sixteen prompts. Some of them confirmed things I already suspected. A few showed me things I hadn't seen. One of them — the 80-year-old self — I'm still sitting with.
The TikTok was right. This is worth doing. Not because AI can tell you who you are. Because four years of your own behavior, reflected back at you, can.
Rich Washburn is a technologist and strategist working at the intersection of AI, infrastructure, and capital. He is Managing Partner and Chief AI Officer at Eliakim Capital.