The Brain Is Not a Server Farm

There's a prototype sitting in a lab right now that uses actual human neurons to compute. Not simulated neurons. Not a neural network in the metaphorical sense that every tech press release has been abusing for fifteen years. Actual biological brain cells, grown and wired into something that processes information, courtesy of a partnership between the National University of Singapore's Yong Loo Lin School of Medicine, DayOne, and Cortical Labs. You can call it a biological data center if you want a phrase that sounds clinical and manageable. You can also call it what it is, which is one of the stranger sentences anyone has typed in a while: we are using human brain tissue to run computations, and we are doing this now, not in some speculative future where the ethical frameworks are presumably already sorted out.

The applications they're pitching are real and some of them are genuinely useful. Drug discovery. Humanoid robotics. Fraud detection. Cybersecurity. These aren't nothing. Drug discovery in particular is the kind of problem that kills people while it's being solved slowly, so if neurons-as-compute actually accelerates that, the calculus gets complicated fast, because "this makes me uncomfortable" and "this might find the thing that cures the thing that's killing someone's kid" are not easy to weigh against each other and anyone who tells you they are is either lying or hasn't thought about it long enough. The discomfort is real. So is the potential. Holding both of those at once is the only honest position available.

What keeps nagging at me, though, isn't the technology itself. It's the pace. The gap between what we can build and what we've thought through has been widening since at least the moment someone decided to put a social engagement algorithm in front of three billion people without asking what "engagement" would eventually mean for democracy, for mental health, for the basic epistemological question of whether shared reality is still a thing we have. We didn't figure that out first. We figured it out, slowly and badly, while the consequences were already compounding. And now we're here, with brain cells in a server rack (roughly), and the ethical conversation is happening in the same breath as the press release, which is not the same as the ethical conversation happening first.

The question of what we owe the neurons is not a joke, even if it sounds like one. These are human cells. They came from somewhere. They process. Whether they experience anything is a question that neuroscience cannot currently answer with confidence, and "we don't know" is not the same as "definitely not," and "definitely not" is the threshold you'd want before you start scaling this into a data center. The researchers know this. Cortical Labs has been public about the fact that they're working on ethical frameworks alongside the technical development, which is more than most tech ventures bother to do, and I'm genuinely not being sarcastic about that. Credit where it's due. But frameworks developed concurrently with the thing they're supposed to govern have a way of being shaped by the momentum of the thing, rather than the other way around.

I've been thinking about the old questions this raises, the ones that get called philosophical when people want to dismiss them and get called urgent when the thing they're describing is already built. What constitutes a subject? What do we owe something that processes but may not feel? Where's the line between a tool and something that has, however dimly, an interest in its own continuity? Religion has been arguing about the soul for millennia and hasn't settled it. Neuroscience has been arguing about consciousness for decades and hasn't settled it. We're not going to settle it in the ethics appendix of a grant proposal, and we probably shouldn't pretend we can. What we can do is slow down enough to ask the question seriously before the answer gets made for us by the fact that the thing is already deployed at scale and the contracts are already signed.

None of this means stop. I don't think stop is the right answer, and I'm suspicious of anyone who reaches for it reflexively, because some of the suffering this technology might address is happening right now, in bodies that can't wait for the philosophical consensus. It means build carefully. It means the people in the room when these decisions get made should include more than the people who stand to profit from the decisions. It means "we don't know yet" should be allowed to pause things, not just footnote them. It means the neurons deserve at least as much consideration as the IP.

We've been at edges like this before, places where what we could do outran what we'd decided we should do, and the record is mixed. Sometimes we figured it out. Sometimes we're still cleaning up. The fact that this one involves the literal stuff of human thought seems like a reason to be more careful, not less, and to have the argument loudly, in public, before the prototype becomes the product and the product becomes the infrastructure and the infrastructure becomes the thing nobody remembers choosing.