This site is part of my thinking on where the boundary of the mind lies and what happens when it gets thinner

Getting it out of my head

My head isn't good at storage. It's good at inventing, connecting, noticing. But holding a dozen half-started ideas in memory is hard for it: each one takes up room and quietly reminds you it's there.

So I offload them. When an idea comes, I hand it off right away, by voice or in text, and stop remembering it. From there the agents pick it up.

I have separate projects, and the agents sort every thought into them: what it belongs to, what it connects with, what it might turn into. Then I decide what to do with it. Most thoughts just sit and wait. Some I start to grow, and the next project is born from them.

This site went through exactly that path: a project grew out of offloaded thoughts that agents helped me sort. In a sense, the site is a continuation of my thinking, only written down in files rather than in neurons.

It doesn't feel strange to me. If you don't treat your brain as something sacred and inseparable, the system looks extensible. A thought started in my head, continued in an external system, an agent sorted it, and the decision about it was made by me again. Where in all this is "my" mind, and where does it stop being mine?

Not dumber, just slower

The simplest way to check whether a tool is part of you is to take it away in your mind. Without external systems I don't get dumber: same head, same ideas. I get less productive.

The difference matters. Dumber means something has disappeared from me. Slower means the channel I usually act through has disappeared. Like someone used to cycling who suddenly has to walk: same legs, same city, the destination is just farther away.

Philosophers have argued about this for a long time, and I wrote about the idea of the extended mind in more detail in The Mind as an Interface. Here I'm more interested in the practical side: what part of my thinking I've already handed over, and what came of it.

What I trust the machine with

The list is already long. I trust the machine with design, development, writing. I don't write code myself anymore: I set the frame, the rules and the goal, and hand the implementation to AI.

But there's one place where I hit a wall every time. The machine can't yet convey an idea the way I see it. I explain, get a result, see that it's not quite it, clarify, explain again. It doesn't understand me one hundred percent yet.

I suspect this is less a question about AI than about how we talk to it. To pass on a picture from your head, you probably have to see it clearly yourself first. And the clarifications the machine asks for make the picture clearer for me, too.

In the other direction, though, it works beautifully. When I have a solution, I deliberately ask AI to suggest other options, each with its reasoning. Among them there's often one that's better than mine. That way I solve the problem and also expand my own map: I see moves I wouldn't have reached on my own.

The bottleneck is bandwidth

Right now I talk to external systems in two ways: by voice and by keyboard. Both are slow. A thought appears in a fraction of a second, but to pass it on I have to sit down, open the laptop, type or dictate. So for now external systems extend my mind only very remotely: there is a connection, but it's narrow.

There's a curious twist here. Neuroscientists Jieyu Zheng and Markus Meister calculated the rate at which a person puts information out into the world: typing, speaking, playing, memorizing. It came to about 10 bits per second Source 1. A bit is the answer to one yes-or-no question. Ten such answers per second is our entire throughput, even though our senses take in incomparably more.

So perhaps it's not just the keyboard: conscious thinking itself may be what's slow. But for me the main thing isn't bits. It's friction. Every time I want to continue a thought outside my head, I have to interrupt what I'm doing and sit down at a desk. Remove that friction, and the boundary between "thought it" and "wrote it down" almost disappears.

Such devices already exist, and the best known is made by Neuralink. Its implant reads signals from the brain areas that control the muscles of the hands and turns them into cursor movements. For now these are clinical trials for people with severe paralysis and amyotrophic lateral sclerosis (ALS), not a product. In January 2026 the company reported twenty-one participants worldwide Source 2.

What I find most interesting in the company's report isn't the technology but how participants describe the experience. At first a participant tries to move their hand, and the cursor follows the attempt. After a few minutes, many forget about the hand: the cursor simply ends up where they wanted it. The company writes that people start acting as if the interface has become a natural extension of the body. Some type by thought at up to forty words per minute by imagining the movements of ten fingers Source 2.

Other researchers are going straight to text. In 2025 a group at Stanford learned to decode inner speech: the words a person says to themselves without trying to say them aloud. Accuracy reached up to 74 percent. And a safeguard came right along with it: the system reads inner speech only when the person has switched it on themselves, like a phone you unlock with a password Source 3.

Think it, and the thought is written down as text. Think it, and the light comes on. I suspect that at some point this will stop feeling any different from raising my hand. I raise my hand by intention too; it's just that the path from intention to movement was laid by nature, not by engineers.

Would I get a chip

Yes. But not today's.

Today's implant means surgery, trials and a very narrow task: giving control over a computer back to people who lost it to illness or injury. For them it's a huge step. For me, in its current form, it doesn't fit.

I'd get one when it becomes simple and reliable. When controlling external systems through it is as easy as moving my hand, with no effort and no thought about the device itself. I don't know when that will be, and I don't want to guess.

What I can picture clearly is the scene I want it for. I'm walking around the apartment, thinking, and suddenly I remember a task I wanted to move forward. Right now I'd have to sit down at the laptop and put my hands on the keyboard. In that future I'd just connect to the system with a thought and keep working without ever stopping my walk around the room. I think, talk to the agents, make decisions. Thinking doesn't pause just so I can get to the tool.

What "I" is, then

If part of my thinking is already outside, it's natural to ask: so where am I?

My answer may sound cold. "I" is a construct the brain uses to describe itself. Something like an internal linguistic label, a convenient name for the whole system. If part of my thinking is outside, then in a sense "I" is already a little bit outside too.

I don't cling to this label and don't identify with it. Beneath it there's what feels real to me: a subject that is aware of itself and has intention. Through intention it changes thoughts, actions, and through them the environment around it. Whether you call it "I" or something else doesn't matter that much. What matters is the essence.

This makes it clear why external systems don't scare me. If "I" were equal to the contents of my skull, every agent, every note, every future implant would take a piece of my personality away from me. But if "I" is an intention that acts through different tools, new tools don't dissolve it. They give it more hands.

The far horizon

Here I move from experience to imagination, and I ask you to read it that way.

If intention can be passed directly into an external system, bypassing the muscles, then agency is already reaching slightly beyond the body. Today it's a cursor on a screen. Maybe later it will be the light, the door, a fridge that tells me what it's missing when I've merely thought about it.

And if you extend the line further, a question appears that interests me a lot. Could agency one day end up entirely outside a biological body? Could a new branch of mind split off from the human species, built not on organic chemistry but on silicon and metal? Then the next stage of evolution would be technological rather than biological.

I don't know the answer and won't try to predict it. It just seems more honest to me to keep this question open than to assume in advance that the mind is forever locked in the form in which it first appeared.

What I take from all this

Put it all together and I end up with only a few claims I'm ready to stand behind today.

1. 1

   Part of my thinking is already outside: in notes, agents, projects. This site is one of its results.

2. 2

   Without tools I'm not dumber, just slower. The boundary of the mind runs where fast exchange ends.

3. 3

   I hand the machine design, development and writing. It can't yet convey an idea the way I see it, and that's more a question of how we talk to it.

4. 4

   The bottleneck right now is bandwidth and the friction between thought and action. Brain-computer interfaces are trying to widen it.

5. 5

   "I" is a convenient label. Beneath it is intention, and intention can act through any tool.

I'll keep writing about how the boundary between human and machine is shifting: what works for me, what researchers are finding, and what's still an open question. If you'd like to think about this together, you can subscribe to new texts.

In the meantime, I have one question for you. What part of your thinking have you already handed over, and did you notice when it happened?