Every AI Knows a Different You [Raw Session]

00:00 — Opening

Hey, welcome back to Slow Builds.

In one of my earlier videos, I was talking about the idea that maybe my digital self isn’t really me. Maybe it’s just the version of me that’s easiest to measure.

After I finished recording, though, another thought hit me: I don’t even think I have one digital self. I don’t think anyone has one.

I think we all have a bunch of them, and every company, every AI, and every service is building a completely different version of who they think we are.

It’s easy to imagine that somewhere there’s this complete profile of us. One giant file that says, “This is you. Here’s everything you like, everything you do, and how you think.”

But when you stop and think about it, no one actually has that. Not even close.

Instead, everyone has tiny pieces or different versions of you.

00:51 — Different Versions of Me

Take AI, for example.

Gemini probably knows things about me that ChatGPT doesn’t. Gemini sees my Gmail, bills, appointments, calendar, and travel confirmations. Maybe it understands that side of my life: my organization and day-to-day activities.

Claude knows me completely differently. It knows my work, integrations, coding style, architectural decisions, and problem-solving skills. It probably thinks software is my entire personality.

Then there’s ChatGPT. This one might know me best from a personal point of view.

Gemini probably knows me best from an organizational and day-to-day perspective. Claude probably knows me best from my work life.

Another service probably knows my health activity better than anybody. It also knows which apps I like. It doesn’t know what’s happening inside those apps, but it knows which ones I use and where I spend most of my time.

My bank knows how I spend money. Amazon knows what I’m buying. Google knows what I’m searching for, and YouTube knows what I’m watching.

02:36 — Different Windows Into My Life

Each company is building its own little profile, separate from the others. Every one of those profiles is built through a different window into my life.

It’s almost like asking ten different friends to describe you. None of them are lying, but none of them can describe the whole person.

Then you realize something else: we’re actually making this harder.

02:54 — Intentionally Breaking Connections

For years, everyone has said—and they still say it—be careful online. Protect your privacy. Don’t share everything. Use a VPN, block trackers, browse in incognito mode, use different email addresses, and keep work separate from your personal life.

Don’t use your real birthday everywhere. I do that one. And I lost my spot again.

But I do all these tricks. I change the answers to my security questions. I don’t connect every account. I use passkeys for many things. I use burner email addresses.

That sounds like you don’t trust the systems, and the funny thing is that all of that advice is good.

But every one of those decisions breaks another connection. It makes another piece of your digital life invisible to the other parts.

03:48 — Personalization Versus Privacy

That’s the tension I can’t stop thinking about.

On one hand, everyone wants AI to become more personal. We want it to understand us, remember us, know our preferences, save us time, help us make better decisions, and give us better recommendations.

On the other hand, we’re spending more effort than ever making ourselves harder to observe.

We’re hiding behind VPNs and blockers. We’re limiting social media, using fake accounts, deleting old posts, and separating our work and personal lives.

These two ideas are pulling in opposite directions.

We’re asking technology to know us better while intentionally revealing less of ourselves.

Honestly, maybe that’s exactly how it should be.

Privacy and protecting yourself are important. Understanding how these systems use your information, allowing different parts of your life to remain separate, and not relying completely on these systems are all good things.

But it also creates that tension.

04:43 — Fragmentation as Protection

I don’t know if I want one company to know everything about me.

Maybe I’d rather no one know the whole picture. Maybe I’d rather every AI understand only one part of my life.

Maybe fragmentation isn’t a flaw. Maybe it’s actually protection.

If one company had access to my work, finances, health, conversations, travel, purchasing, thoughts, and even relationship information, that would become something very different.

It could become something capable of duplicating or impersonating you, and that’s very scary.

Maybe this separation isn’t a flaw. I believe it has a purpose, but it changes how we should look at these systems.

05:30 — A Different AI for Each Part

I’m off script here, but if you want an AI that understands your work habits, maybe that’s your work AI. It’s like having a separate work phone.

Then maybe you have something else for cooking, chores, and the practical parts of your home life. Something that helps you plan, reminds you to put out the garbage, knows which meals you like, or notices what’s missing from the fridge.

I think there may be different places for everything.

06:04 — Intentionally Scattered

Back to the script.

I have another question: maybe there will never be one AI that really knows me.

Not because the technology isn’t good enough, but because I’ve intentionally scattered myself across different services, devices, accounts, conversations, and parts of my life.

Honestly, maybe that’s true for almost everyone.

We talk a lot about personalized AI, but at the same time, we’re becoming less observable than we’ve ever been before.

06:32 — Dozens of Incomplete Versions

I think that’s a really interesting tension, even though it creates a lot of friction between these systems.

Maybe there isn’t one digital version of me. Maybe there never will be.

Maybe there are dozens of versions of me. Each one convincing, each one useful, but each one a little incomplete.

Maybe that’s exactly how it should stay.

If one system knows you completely, then technically, it is you. Maybe that’s a scary thought.

It’s also a useful thought. It might be something we could strive for, but maybe it isn’t where we should want this technology to go.

I believe we should keep some personality and humanity in it.

Like I said in my previous videos, my memories are my memories. Things happen in my life that aren’t digital, so an AI will never know them.

I believe that’s what keeps the humanity there.

All right. Thanks for watching. Bye.