AI は死んだ？脳器官が生命を持つ新事実

I’m going to let you in on a secret. Every cell in your body has the potential to get smarter. I don’t mean this metaphorically, or in a “body keeps the score” kind of way. I mean that if lab-coated biologists took a sample of your skin and very carefully manipulated the cells inside it, they could actually make a brain. They do it all the time.

Not a brain as complex as the one behind your eyes, of course, but a glob of gray matter nonetheless, with a few-million-odd neurons that can send and receive electrical signals. Biologists call these strange creations human brain organoids. Kept at a womblike 98.6 degrees Fahrenheit for eight months, they’ll produce repetitive oscillations—*brain waves*—nearly indistinguishable from those made by a premature baby.

The most ambitious, future-defining stories from our favorite writers.

In cell culture labs around the world, human brain organoids live out their short lives as neural guinea pigs, testing the effects of diseases, toxins, and new pharmaceuticals. But they may soon be on to more glamorous pursuits. At the University of San Diego, organoids are guiding spidery robots through mazes and taking hero doses of psychedelics. At Johns Hopkins, they’re forming the basis of novel biocomputing systems. And at a startup in Melbourne, they’re playing video games like *Pong* and *Doom*.

Biologists do the darnedest things. While the rest of us are distracted by large language models and AI agents, they’re going straight to the source of intelligence, cultivating living neurons and teaching themselves to program them with electrical signals and hits of dopamine. In the future, they wager, artificial intelligence won’t be artificial at all. It’ll be built from the stuff of life itself.

The most metal building at UC San Diego is the library. An inverted concrete ziggurat, the Geisel Library—named for the children’s author better known as Dr. Seuss—looms over an otherwise bucolic campus on spindly, two-story legs. On a recent afternoon, as a marine layer hung low in the eucalyptus groves, it looked particularly like the mothership of a brutalist alien race.

That day, the Geisel’s sunken lobby was hung with scientific images from the university’s collection. Among CGI renderings of folded proteins and macrophotographs of benthic sea creatures, one image stuck out. It depicted a clump of human brain cells, silhouetted in black against the milky white of a petri dish. A corona of axons, the threadlike nerve endings that transmit electrical impulses across the brain, stretched outward from the clump with palpable yearning.

Whether in our skulls or in a dish, neurons want nothing more than to find one another—and, across the emptiness, to forge the synapses whose electrical chattering forms the basis of thought. They’re very good at it. If you put loose brain cells together, they will multiply and interlink until they’ve cohered into autonomous globs of tissue. Human brain organoids practically make themselves.

A 20-minute walk from the Geisel, at UCSD’s Sanford Stem Cell Institute, they’re making themselves in the tens of thousands. “Whatever environment you put them in, the first thing that they do is try to connect,” said the Brazilian developmental biologist Alysson Muotri, as we gazed over the blue plane of Pacific outside his office window. “Connect with the dishes, connect with the electrodes, connect to each other. This is an intrinsic property of our brain, to connect.”

Muotri is dashing, with a surfer’s tan and the aquiline profile of a figure on an ancient Roman coin. Over the past decade, his lab has dramatically expanded the scope of brain organoid research. He and his colleagues have revived genetic material from the hominin fossil record to create “Neanderthalized” brain organoids. They have sent organoid payloads to the International Space Station to study what cosmic radiation does to astronaut brains. But the issue closest to Muotri’s heart is autism. His 18-year-old son is autistic and receives 24-hour care. By studying brain organoids grown from the cells of autistic donors—including his son—he hopes to pinpoint where the neural development of autistic children differs from their neurotypical counterparts.

This is not an invasive procedure. To make a brain organoid, all you need is that sample of skin I mentioned before. (Samples of blood, hair, or teeth work too.) You take the adult cells and introduce them to some special proteins that revert them to their embryonic state. Given a second chance to mature, these so-called induced pluripotent stem cells can become anything: tear gland organoids that cry, heart organoids that beat, or brain organoids that … well, that’s the question.

In utero brain development is, as one bioethicist told me, “a black box” of scientific knowledge. Historically, a lot of what we know about it is inferred from studies with mice. But with an organoid, the transformation of stem cells into neurons into brain tissue happens in full view. In theory, scientists could one day study how a colony of dividing cells comes together to create a mind—to make, from 86 billion neurons, a person named Alysson Muotri, for example.

Brain organoids are quarter-peanuts of lab-grown flesh.

A prototype “biochip” from Johns Hopkins that integrates organoids and hardware.

Or another one named Claire L. Evans, who at this moment is leaning over Muotri’s shoulder to gape at a dish of floating brainlets under the microscope. Visually, organoids are not compelling; they’re opaque, snot-colored, the approximate size and shape of a chia seed. Muotri’s organoids contain 5 million cells, of which 2.5 million are neurons. (The rest are non-neural glial cells, which serve as scaffolding.) This, he reassures me, is the size of a bee’s brain. I suspect this is his way of tempering any ethical ick I may have. Although the evidence for insect sentience is growing, invertebrate research is still exempt from federal animal welfare laws. As John Evans, a sociologist and the codirector of UCSD’s Institute for Practical Ethics, put it to me when I wandered over to his office for the humanist take, “You don’t need to get permission to torture as many flies as you want.”

For now, you don’t need permission to torture brain organoids either. From a bioethical perspective, they aren’t people; they’re not even animals. If organoids someday graduate from bee to mouse size, this protocol would have to evolve. The trouble is, nobody knows quite where to draw a line. Sentience is hardly a settled idea; it’s not even clear you can have sentience without a body or a sensory experience of the world. And forget about calling an organoid conscious. “Imagine that you had spent your life in a glass tube—could you possibly even think of what you have as consciousness?” Evans asked me. “Philosophers of mind will tell you that what we call consciousness is not possible without experiences.” To make a conscious organoid, he added, “you first have to start having organoids have *experiences*.”

This isn’t impossible. We’ve all seen *The Matrix.* To the brain, the whole Ferris wheel of life is just pings of chemistry and electricity. To give an organoid an “experience,” all Muotri has to do is pluck one from its amniotic goo, place it on a conductive sheet of graphene, and—not to put too fine a point on it—*zap* it.

Do the organoids like being zapped? Muotri isn’t sure. What he does know is that they respond to electrical signals, remember them, and eventually come to anticipate them. For him, that’s evidence that they’re maturing, becoming more useful models of human development. But for other researchers, this electrical communication represents something else entirely: that living matter, like a computer, is programmable. And this is where things get *truly* weird.

It’s 3:30 pm in Los Angeles, which means it’s tomorrow morning in Australia. I’m sitting on the fire escape of my office building, peering at a grid of 59 squares on my laptop screen. I’ve been told that each of the squares represents an electrode in the Melbourne laboratory of the biocomputing startup Cortical Labs. And on each of those electrodes is a tiny culture of living human neurons. At the moment, my screen is registering fleeting spikes from those neurons—the spontaneous activity of brain matter in a vacuum. I click a square, zapping off an electrical hello to neurons 8,000 miles away. In response, all 59 electrodes spike at once.

For a moment, I’m giddy with a feeling of new power. I click around, sending those distant neurons hopping; my screen fills with the peaks and valleys of their electrical pulses. I page over to the neurons’ environmental settings. If I wanted to, I could drop their onboard temperature or nuke their precise gas mix of oxygen and CO2. If I did that, they’d certainly die. It’s about as significant a paradigm shift as I can imagine for computing: No matter how badly you mess up your code, things on the computer don’t normally *die in real life*.

But that’s the reality of the Cortical Cloud. In Melbourne, Cortical Labs cultivates flat neural cultures—the stem cells were donated by the company’s own founder—and loads them into sleek white biological “computers” called CL-1s. Each is the size of an elongated toaster and boasts an onboard life-support system capable of keeping a culture of up to a million neurons alive for six months. With the CL-1, Cortical Labs is aiming to become the Nvidia of neural computing, providing hardware and, let’s say, “neurons as a service” with a sub-millisecond delay.

For now, these neural computers are mostly of interest to researchers who want to work with neurons without taking on the tedious wet-lab husbandry themselves. Eventually, however, the company hopes that neurons will prove themselves to be an energy-efficient, resilient substrate for more general computing applications—including some tasks currently handled by AI, like image recognition and classification.

“When you think about what you want from AI, it’s biology,” said Brett Kagan, Cortical Labs’ chief operating officer, when I reached him over Zoom. “You want it to be self-repairing as much as possible. You want it to be adaptable. You want it to be long-lived. You want it to be energy-efficient. These are all features you get for free in biology.”

Brett Kagan, of Cortical Labs, believes his neural cultures display a form of sentience.PHOTOGRAPH COURTESY OF CORTICAL LABS

Kagan is a new father; as we spoke, his toddler ran riot in the background. But he’s no stranger to young and unruly forms of intelligence. In 2022, using a system similar to what currently powers the Cortical Cloud, Kagan grew a neural culture on a microchip and trained it to play the 1972 Atari game *Pong,* rewarding the neurons with predictable electrical pulses when they made correct decisions and punishing them with chaotic bursts when they made mistakes.

The technique served as a minimal proof of concept for a theory, proposed by the neuroscientist Karl Friston, that self-organizing biological systems tend to minimize surprise whenever possible. By showing that neurons will reorganize themselves to avoid chaotic stimulus, Cortical Labs demonstrated one possible approach for programming living matter. But the experiment also signaled that the CL-1 could be considered hardware for testing theories of cognition. “Not to try and make our stuff sound so grandiose,” Kagan told me, “but I would say the CL-1 is to theoretical neuroscience as the Large Hadron Collider was to theoretical physics.”

Kagan, to be fair, enjoys a grandiose claim. In the *Pong* paper, he and his colleagues claimed that the neurons, “embodied” within the game, displayed a form of sentience. Many in the research community balked at this cavalier use of language; one particularly polemic response, published in the journal Neuron, accused Cortical Labs of “hijacking” the very concept of sentience.

But as Alon Loeffler, a scientist at Cortical Labs, later explained to me, our brains are embedded in an environment, and responding to that environment in real time is what brains do—it’s what brains are *for*. Since neurons in a dish don’t benefit from those constant feedback loops of action and experience, the game fills the gap. “A game is just a version of the world,” Loeffler said.

“The world’s first code-deployable biological computer,” says Cortical Labs.

Cortical Labs is aiming to become the Nvidia of neural computing.

The *Pong* program comes preinstalled on the Cortical Cloud, as a package of easily deployable Python code. When I ran the Australian neurons through a two-hour session, I watched as they incrementally improved their game in real time. In the end, their longest rally was 10 accurate shots, which beats my tennis game handily. It was impressive, but now the neurons are leveling up.

For one, they have learned to play Doom.

When I set out to report this story, I was sure it was about consciousness: the eerie moment a quarter-peanut of flesh sparks with self-knowledge, and what that precipice means for the researchers responsible. I imagined long dark nights of the cell and tiny funerals for spent neurons. What I found, however, was that nearly all scientists who keep organoids see the consciousness question as a distraction.

Most bristle when asked. They gesture to the organoids themselves—tiny balls bobbing in liquid solution like droplets of olive oil in vinegar—as if to say, *give me a break*. “Consciousness is so qualitative,” complained Annie Kathuria, an organoid researcher, when I visited her lab at Johns Hopkins. “How am I supposed to measure something qualitative on a tissue that’s floating in a dish? Someone has to define it. That’s what I say to everyone: Define to me what consciousness is, in quantitative terms.”

The request is rhetorical, of course. Humans have been trying to define consciousness since the days of Plato and nobody’s come close to nailing down the general idea, let alone a list of quantitative metrics. But it reflects a strong tendency in biology labs to want to focus on the practical, rather than the philosophical. When I spoke to Kathuria, she sipped from a Venti Starbucks tea and regularly glanced at the whiteboard over my shoulder, covered in a list of dozens of drug screenings her lab was on the hook for. Defining the nature of mind was the last thing on her mind.

That’s a job for the philosophers. Publishing papers about the edge cases of sentience has become something of a cottage industry in contemporary philosophy. *Do zombies have moral status? Should robots have rights? What is it like to be a bat? Is an organoid sentient?* Opinions vary. The philosopher Tim Bayne worries that organoids, isolated in their petri dishes, could become “islands of awareness.” Jonathan Birch, a professor of philosophy at the London School of Economics, thinks we’re in trouble if organoids ever grow a brain stem, which he argues functions as a “power cable” for consciousness.

Others suggest thresholds based on the number of neurons, or organoid size. But thresholds are difficult to quantify too. “You cannot draw a barrier that will hold based on a number,” explained John Evans, the sociologist at UCSD. “You could say ‘you can’t grow anything larger than 4 millimeters.’ Well, how about 5 millimeters? There’s no morally relevant distinction between the two.”

Conveniently, organoids can’t grow much bigger than 5 millimeters anyway. Once they reach a certain size, without a vascular system to pump in oxygen and pump out metabolic wastes, they develop a “necrotic core” and suffocate. This has created an upper ceiling for the organoid debate, but it won’t hold much longer. At Johns Hopkins, the nanotechnologist David Gracias is developing biomimetic artificial arteries; across town, at the Medical School, Kathuria creates rough blood vessel networks from endothelial organoids. Researchers are keeping organoids alive longer and longer. Muotri’s record, of three years, might’ve gone longer had someone in his lab not dropped the dish.

“The goal is to get to a 1-centimeter brain organoid,” said Thomas Hartung, as he walked me around the Center for Alternatives to Animal Testing, or CAAT, at Johns Hopkins, where he’s experimenting with new perfusion systems to get around the bloodlessness problem. One centimeter is roughly the size of a mouse brain—“a completely different beast,” Hartung said, than the 500-micrometer organoids he’s accustomed to.

Hartung, a German-born toxicologist, has been working with organoids for over a decade. He’s the picture of a genial scientist. When we met, he sported an oversized cardigan and a beaded necklace made by his 9-year-old daughter. His wife, Lena Smirnova, is a neurotoxicologist and runs a lab at CAAT. Where Hartung is bearish and avuncular, with the loping Teutonic accent of a laid-back Werner Herzog, Smirnova is petite, with a bowl cut and a punctilious manner. The odd coupling seems to work for them. When I asked Smirnova if she’d ever make an organoid out of her own cells, she laughed so abruptly it sounded like she barked. “This is actually something between the two of us,” Hartung said, sheepishly. He’d offered her some of his stem cells, as an act of love, and she has so far refused to do anything with them.

Hartung and Smirnova are leading figures in organoid research and have been vocal public supporters of the idea that organoids will revolutionize computing. In 2023, after organizing an international scholarly community, they published a splashy “Declaration” for a field they christened “organoid intelligence,” or OI. During my visit, however, the organoids in the CAAT lab weren’t doing much computing. They were mostly being dosed with heavy metals. OI is a buzzy idea, and it has drawn a lot of media attention. But practically, it’s a long-term goal. For now, organoids remain what they have always been—glorified lab rats.

Nearly every organoid researcher I met reporting this piece told me the same fact: that in human clinical trials, the failure rate for neuropsychiatric drugs is close to 95 percent. The pipeline for new medications for conditions like depression, Alzheimer’s, and epilepsy is long, and often dry. That’s because drugs are ordinarily tested on animals, not people, and animal testing has never been the most physiologically relevant way to ensure the efficacy of drugs—it’s just been the most practical one. Organoids have changed that calculus.

“Think about nonhuman primates, which is the model closest to us—these monkeys with electrodes in their heads,” Hartung said. “At best, you have a battery of five, six monkeys sitting there and doing their experiments. We’re producing, in one six-well plate, up to 4,000 brain organoids.” In this context, OI isn’t an end as much as it is a means. An organoid trained to play *Pong*, for example, is a perfect test subject to study how drugs affect learning and memory.

In his office, Hartung showed me a simple video game his team is working on—a platformer with varying obstacles for the brain organoids to “jump” over. The idea is to train the organoids to respond to electrical signals representing those obstacles. Once they’ve learned this task, Hartung can ask questions like: How long do brain organoids remember when to “jump”? What effect do Alzheimer’s drugs have on the strength and duration of those memories?

Underneath his desk, Hartung keeps a cardboard box filled with bottles of red wine, which he suggested we crack open twice before I left. He believes all the best conversations happen after 5 pm, and he paid for his graduate studies, back in Germany, doing stand-up comedy (he compares his comedic style to Bill Maher’s). The years of gigging gave him a head for PR.

Organoid intelligence is, above all, good PR—a savvily branded attractor for funding and talent in a wide-open field. “I think the pie is too big for anybody to eat,” he said, leaning back in his chair. “What is really important is to tell the people about the opportunity of the pie so that we jointly have a feast. I will try as quickly as possible to eat my piece, which is the use for this in drug development, or in identifying substances which are impacting brain diseases. That’s my business.”

Business is booming. In July 2025, the National Institutes of Health announced that it will no longer award grant funding to research that relies exclusively on animal testing, encouraging consideration of “new approach models,” like organoids, instead; in September, the NIH announced an $87 million investment in building a Standardized Organoid Modeling Center. Fortunately for the NIH, since organoids are made with induced pluripotent stem cells from consenting adult donors, they don’t inspire the same religious ire as embryonic stem cell research did in the 1990s. And unlike lab rats or primates, organoid testing doesn’t irk the animal rights crowd.

Recent studies suggest that ordinary people, by and large, aren’t much bothered by organoids. When they do lodge an objection, it’s not because globs of neural cells might someday achieve human-level sentience. It’s because brain organoids are creepy. They muddy the line between person and thing, which, like the difference between humans and animals, or between the dead and the living, is a fundamental distinction that transcends cultures.

In 2017, a pair of neuroscientists at University College London took a small sample of skin cells from the science journalist Philip Ball’s arm and made them into a brain organoid. Eight months later, gazing at their creation under the microscope, Ball felt an “unsettling frisson” at the idea that this living tissue, formed from his flesh as Eve from Adam’s rib, pulsed with the same electrical signals that animated his own brain. Although their structure was comparatively primitive, the neurons in Ball’s organoid were nonetheless communicating with one another. “These are not ‘thoughts,’” he told himself. “But they are what thoughts are made from.”

Cajoled into becoming a mini-brain, Ball’s cells shifted ontological categories. Nobody could deny they were still alive, but it wasn’t Ball’s life they were living anymore. As Ball observes in his book How to Grow a Human, life is supposed to be a one-way journey: Sperm fertilizes egg, egg becomes zygote, and from this single cell cleave the trillions of facets that make us. This “seems conceptually unproblematic, so long as we can regard it all along as a coherent, boundaried organism with a unique history,” Ball writes. “But if you can take a tiny part of that mass of cells, rewind the tape and make a fresh start, then this unity starts to dissolve.” Kept warm and well fed in a petri dish, our cells carry on quite happily without us. Are they still ours?

Microphotographs of wired-up neural cultures.

Somewhere between the few million neurons in a brain organoid and the billions more that make up our minds, selfhood emerges—and absolutely nobody knows when. It reminds me of something the gnomic science-fiction writer Ted Chiang observed in a 2021 interview. “Suffering precedes moral agency in the developmental ladder,” he said. “In the process of developing machines that are conscious and moral agents, we will be inevitably creating billions of entities that are capable of suffering. And we will inevitably inflict suffering on them.”

Chiang was talking about AI, but his point stands. After all, as digital interfaces between neural wetware and silicon hardware improve, we’ll have a harder time differentiating between the two, and it’s reasonable to imagine that efforts to engineer and standardize biological intelligence will inform the design of future AI systems. But what struck me, as I visited one cell culture lab after another, is the way their values *inverted*, rather than mimicked, those of the AI business.

For the AI people, machine intelligence seems achievable, even inevitable. They “feel the AGI.” Part of this is hype; part of it is a willingness to take intelligence for what it does, as opposed to what it is. “If it quacks like a duck, it is a duck,” said John Evans, the UCSD sociologist. “The Silicon Valley types are super pragmatic. If AI is capable of doing what a human with consciousness does, they’ll call it consciousness.”

Biologists who work with neural cultures, however, are far less likely to use such a word without plenty of agonizing caveats. These biologists cultivate the raw meat of mind every day; they’re in the privileged position to understand how colossally complex it really is. As biology and AI converge, the culture shock between these worldviews is likely to be bracing.

For now, it seems that in the race toward conscious machines, there will be two lanes. One will be paved with rare earth minerals and silicon—forced, at enormous financial and energy cost, to model the brain from the top-down, in the form of artificial neural networks. Although there’s a lot these systems can do well, they will never be brains. Nothing forged from silicon ever could. As Cortical Labs’ Brett Kagan put it, “How can you have a truly adaptable, flexible system in an inflexible substrate?”

Which leaves the other path. Admittedly, this one will be slippery and meandering—adapting, as all living things do, to the journey as it goes. It might not be the most direct route, but in the evolutionary history of life on Earth, it’s the only thing that has ever led to intelligence.

So I’m betting on the neurons. I made up my mind back at that strange mothership of a library in San Diego, when I saw those axons reaching for one another across the void. Something in the innocence of their striving moved me. After all, my cells and yours have been doing the same thing since the day we were born. Trying to connect. And one thing we can count on—through the rise and fall of silicon—is that they’re never going to stop.

*Let us know what you think about this article. Submit a letter to the editor at* *mail@wired.com.*