Do these 3 things before closing this tab:
1Fix the driver behind crashes, sound loss and screen glitches2Repair Windows errors before they cause bigger problems3Scan for outdated or missing drivers - takes under a minuteiTechGuides is reader-supported. When you buy through links on our site, we may earn an affiliate commission. As an Amazon Associate I earn from qualifying purchases. Learn more
AI is not dead. The provocative headline points to something more specific: scientists are growing living neural tissue, connecting it to electrodes and computers, and testing whether biological networks can model disease, learn constrained tasks or eventually perform useful computation. Brain organoids are real and electrically active, but they are millimeter-scale models of selected brain features—not miniature human brains, conscious people in dishes or replacements for deployed silicon systems.
The claim behind the headline
“AI is dead” is a rhetorical way to ask whether living neural tissue could change computing and neuroscience. The evidence supports a narrower conclusion. Organoid intelligence (OI) is an emerging research program that combines three-dimensional cultures of human brain cells with electrodes, sensors, feedback systems, data processing and machine-learning tools. Its proposed advantages remain goals, not established capabilities.
Today’s software AI runs on highly engineered digital hardware with defined inputs, outputs, benchmarks and deployment workflows. An organoid experiment instead studies biological activity in a living culture. The same setup can be valuable even when it never becomes a practical computer, because it can reveal how human neural cells develop, respond to drugs or organize electrical activity.
What a brain organoid is—and what it is not
A model of selected biology
A brain organoid is a lab-grown three-dimensional neural culture. Researchers use human cells to reproduce selected features of brain development, cell types and electrical activity in a controlled environment. The cultures discussed in the feature are millimeter-scale and contain millions of cells. That size and cell count make them useful experimental systems, but they do not turn an organoid into a complete brain.
#1 Best Overall
Organoids do not reproduce every region, connection, support system or life experience of a human nervous system. Their organization, maturity and activity can vary between batches and laboratories. A result from one culture therefore describes that preparation under its experimental conditions; it cannot automatically be generalized to a person or to every organoid.
Why medicine is the nearer-term use
The strongest current case for organoids is biomedical. Cultures let investigators examine development, model aspects of neurological or psychiatric disease and test how drugs or toxins affect human neural tissue. They can expose biological mechanisms that are difficult to observe directly in living people and can complement, rather than automatically replace, animal studies.
Researchers interviewed for a 2026 WIRED feature cited a neuropsychiatric-drug clinical-trial failure rate close to 95 percent. That figure is reported from those interviews, not independently established here as a universal rate by the National Institutes of Health or the scientific literature reviewed for this article. It illustrates why human-cell models attract interest, not proof that organoids will solve clinical-trial failure.
Recommended Free Tools
How organoid intelligence experiments work
Connecting living cells to a machine
An OI setup needs a way to stimulate the culture, record its activity and return information as feedback. Electrode arrays can deliver patterned electrical signals and measure responses from many sites. Sensors and output devices can translate a task into stimulation and translate the culture’s changing activity into a computer-readable signal. Software then analyzes the recordings, adjusts feedback or evaluates whether the network changed in a repeatable way.
The 2023 OI overview describes this as a multidisciplinary engineering problem. Better stem-cell and culture methods must be paired with bioengineering, input/output measurement, data processing, feedback control and machine learning. Without reliable interfaces and agreed metrics, an interesting electrical trace is not evidence of a useful computer.
Rank #2
What the Pong demonstration shows
The feature describes a neural culture connected to an electrode interface and trained to play a simplified version of Pong. The system received stimulation representing the game state and feedback linked to the result of its activity. Over repeated trials, the recorded behavior changed in a task-relevant direction.
That is evidence of adaptive activity in a constrained laboratory loop. It is not evidence that the culture understands Pong, has human-like intelligence or is conscious. The experiment supplies an artificial task, a narrow signal channel and a defined scoring rule. A biological network can change its responses under those conditions without possessing general knowledge or a subjective point of view.
Free tools Windows power users keep installed
One-click scans. No signup required.
Organoids and silicon AI are different kinds of systems
Comparing an organoid with a deployed AI model is useful only if the comparison keeps their purposes and evidence separate. No controlled benchmark in the cited material establishes that one is universally superior.
| Dimension | Deployed silicon AI | Organoid or hybrid computing |
|---|---|---|
| Readiness | Operational systems run at commercial and research scale with established software and hardware stacks. | An emerging program whose cultures, interfaces and evaluation methods are still being developed. |
| What is modeled | Mathematical representations optimized for specified computational tasks. | Living neural activity and selected features of human development, disease or learning. |
| Input and output | Digital data moves through standardized electronic interfaces. | Electrodes, sensors and feedback translate between biological activity and machine-readable signals. |
| Capability evidence | Performance is reported with task benchmarks, reliability measures and deployment results. | Evidence currently consists mainly of controlled, task-specific experiments and biological measurements. |
| Key uncertainties | Bias, security, reliability, energy use and governance of software systems. | Culture variability, maturation, measurement limits, reproducibility and the possibility of morally relevant experience. |
Proposals that organoids could learn flexibly, use less energy or offer new forms of computation are therefore prospective. Energy comparisons between a brain and a supercomputer depend on estimates and unlike-for-like assumptions; they should not be treated as a definitive efficiency proof for organoid computers.
What scientists still have to solve
Reproducibility and maturation
Human-cell cultures are biological preparations, not identical manufactured chips. Donor cells, reprogramming, growth conditions and timing can change the resulting tissue. Researchers need standardized protocols, quality controls and shared reporting so that two laboratories can tell whether they produced comparable organoids.
Developmental maturity is another constraint. A culture may reproduce an early developmental state rather than the full range of adult neural functions. Extending maturation, maintaining tissue health and producing more consistent architecture are central technical goals.
Interfaces that can both stimulate and observe
An electrode array records a limited, indirect view of activity. More channels do not automatically reveal what each cell is doing or how information is represented across the culture. Stimulation can also perturb the network in ways that are difficult to interpret. OI therefore needs better spatial and temporal resolution, calibration and methods for separating genuine learning from artifacts of the interface.
Feedback, training and measurement
A useful biological computer would need a stable feedback loop: a task definition, an input encoding, a measurable output and a way to improve performance. Researchers must show that changes persist, generalize beyond the training examples and survive controls that rule out equipment drift or statistical chance. Agreed benchmarks are still being developed, so a demonstration should be described by its exact task and conditions rather than by broad labels such as “intelligent.”
Scaling without losing biological meaning
Larger cultures may provide more activity, but scale alone does not guarantee better computation. Keeping tissue alive, delivering nutrients, wiring many channels and interpreting the resulting data become harder as systems grow. A practical hybrid device would also need reliable electronics, software, maintenance and safety procedures. Those engineering requirements are unresolved, not hidden proof that a general-purpose organoid computer already exists.
Do organoids “like being zapped”?
Electrical stimulation is an experimental input, not evidence of pain, pleasure or preference. A culture can respond to a pulse by changing its electrical activity, and a feedback loop can reward one pattern by making the next input different. Neither observation establishes that the tissue experiences the stimulation.
What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
Annie Kathuria, an organoid researcher at Johns Hopkins, captures the measurement problem: “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.” Her question is not a finding that organoids are conscious; it highlights the absence of an agreed operational test.
Electrical activity is a property of neural tissue, but consciousness is a claim about experience. The reviewed literature does not establish that brain organoids are conscious, nor does it provide a settled scale that would let a laboratory assign them a level of awareness. A responsible report should therefore describe stimulation, activity and learning behavior precisely and avoid language that turns a response into a feeling.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.Ethics: precaution without sensationalism
The ethical issue is not only whether an organoid can compute. It also concerns how human cells are obtained, what donors are told, how specimens may be used in future experiments and whether increasingly organized neural activity could become morally relevant.
The participants in the First Organoid Intelligence Workshop wrote in the 2023 Baltimore Declaration: “We, the participants of the First Organoid Intelligence Workshop – ‘Forming an OI Community’ (22–24 February 2022), call on the international scientific community to explore the potential of human brain-based organoid cell cultures to advance our understanding of the brain and unleash new forms of biocomputing while recognizing and addressing the associated ethical implications.”
The declaration calls for monitoring possible consciousness-related capacities, protecting cell-donor interests and continuing discussion with scientists, ethicists, participants and the public. That is a call for governance and measurement, not an announcement that consciousness has been detected.
Best Value
Philip Ball, reflecting on an organoid made from his cells, put the emotional tension differently: “These are not ‘thoughts,’” he told himself. “But they are what thoughts are made from.” The statement, as recounted by WIRED, distinguishes the material basis of thought from a demonstrated thought or inner life.
What NIH funding actually means
On September 25, 2025, the National Institutes of Health announced $87 million in contracts for the first three years of its Standardized Organoid Modeling Center. The center is intended to develop standardized organoid-based methods and improve reproducibility. The funding signals that organoid work is becoming an infrastructure and measurement priority, not that organoids have replaced other models.
NIH guidance on new approach methodologies says model selection should be scientifically appropriate for the question. Animal research remains eligible when it is justified and properly overseen. The policy is not a blanket prohibition on animal studies; it encourages investigators to choose the model that can answer a specific question with appropriate safeguards.
How to interpret the next organoid headline
- Identify the system. Ask what cells, culture age, interface and task were used.
- Separate response from experience. A change in electrical activity is not a measurement of consciousness.
- Look for controls. Reliable claims should address equipment artifacts, chance performance and whether behavior persists outside training.
- Check the scope. A Pong result or disease-modeling result does not demonstrate general intelligence.
- Read the attribution. A reported statistic, such as the close-to-95-percent failure figure in the 2026 feature, should remain tied to the people and publication that reported it.
- Ask what is practical now. Drug testing, developmental studies and disease modeling are established reasons to investigate organoids; general-purpose biological computing is still aspirational.
Where the field is heading
The most credible near-term path is hybrid: biological cultures provide properties of living neural networks while electronic systems provide stimulation, memory, control and analysis. Progress will depend on standardized cultures, better interfaces, transparent benchmarks and ethics that keep pace with technical capability.
That trajectory does not make silicon AI obsolete. It expands the set of things scientists can model and may eventually produce specialized biological or hybrid devices for problems where adaptation, cellular dynamics or human-relevant biology matter. Until experiments demonstrate repeatable advantages on well-defined tasks, “organoid intelligence” should be read as a research direction rather than a finished computing platform.
The accurate takeaway is less dramatic and more useful: AI is alive, organoids are alive, and the important work is learning exactly what each can—and cannot—do.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
Quick wins for a faster PC:
Scan for outdated or missing drivers - takes under a minuteDriver Scan →Clear out junk files and repair common Windows errorsFree Scan →Fix the driver behind crashes, sound loss and screen glitchesFind Drivers →
