In a sterile lab at Johns Hopkins University, petri dishes sit in incubators alongside server racks. The researchers here are growing neurons — and studying their computational properties.
The field is called organoid intelligence, and it sits at the intersection of neuroscience, computer science, and synthetic biology. The premise: use living brain cells as the substrate for computation, harnessing the energy efficiency and pattern-recognition capabilities of biological neural networks.
Why Neurons?
The human brain performs roughly 10^15 operations per second while consuming about 20 watts of power. The most efficient AI chips today require thousands of watts to approach comparable performance. This gap — roughly five orders of magnitude in energy efficiency — is what makes biological computing attractive to researchers.
"We're not trying to build a brain," says Dr. Thomas Hartung, who leads the Johns Hopkins organoid intelligence lab. "We're trying to borrow some of its tricks. Specifically, its ability to learn from very little data and its extraordinary energy efficiency." Hartung's lab published a roadmap for organoid intelligence in Frontiers in Science in 2023, outlining the research agenda and ethical framework for the field.
How It Works
The team grows brain organoids — small clusters of neurons derived from human stem cells — on multi-electrode arrays that can both stimulate the cells and record their electrical activity. By sending carefully designed electrical signals into the organoid and recording the outputs, researchers can study whether the biological network can perform specific computational tasks.
Cortical Labs' published research demonstrated that organoids could improve performance on a simplified Pong task — a result that has been discussed extensively in the neuroscience community. The interpretation of what "learning" means in this context, and whether it implies anything about consciousness, remains actively debated.
The Road to Practical Biocomputing
The challenges are formidable. Organoids are fragile, require constant biological maintenance, and currently survive for only a few months. Scaling from a few thousand neurons to the billions required for complex computation remains an unsolved problem. The field is at an early stage, and claims about near-term practical applications should be evaluated critically.
DARPA has funded research in this area, which reflects genuine scientific interest from the defence research community — though DARPA funding indicates research potential, not commercial readiness.
The Ethical Dimensions
Organoid intelligence raises profound ethical questions that the research community is actively grappling with. The most fundamental: could a sufficiently complex organoid be sentient? Current organoids are far too simple to raise this concern — they contain thousands of neurons, compared to the 86 billion in a human brain. But the field's trajectory points toward more complex systems, and establishing ethical frameworks before they become necessary is a priority.
The Johns Hopkins team has published an ethical framework for organoid intelligence research that addresses questions of consent (the stem cells used to create organoids are derived from human donors), welfare (how to assess and minimise any potential suffering), and governance (who should oversee research in this area). The framework is a model for responsible development of a technology that could have profound implications.
Energy Efficiency: The Core Opportunity
The most compelling near-term application of organoid intelligence is not general computation but specific tasks where biological neural networks have demonstrated advantages over silicon: pattern recognition, anomaly detection, and learning from limited data. These are tasks where the brain's architecture — evolved over millions of years for exactly these problems — may be genuinely superior to current AI approaches.
If organoid-based systems can perform these tasks at a fraction of the energy cost of silicon AI, the applications are significant: edge computing devices that can run sophisticated AI without large batteries, medical diagnostic systems that can operate in resource-limited settings, and environmental monitoring systems that can be deployed at scale without significant power infrastructure.
Sources & Further Reading
- Frontiers in Science — Hartung et al., "Organoid intelligence (OI): the new frontier in biocomputing and intelligence-in-a-dish" (2023)
- Neuron — Cortical Labs' peer-reviewed Pong paper (2022)
- Johns Hopkins Hub — official university coverage of the organoid intelligence research programme
- DARPA — biological computing research programmes and funding
- Nature — neural engineering research including organoid-electrode interface studies
The Investment Landscape
Despite being at an early research stage, organoid intelligence has attracted venture capital interest. Cortical Labs raised $10 million in Series A funding in 2023. Several stealth-mode startups are working on proprietary organoid computing platforms. The field is small but growing, and the potential applications — from drug discovery to edge AI — are attracting serious scientific and commercial attention. Whether organoid computing becomes a practical technology or remains a fascinating research curiosity will depend on solving the fundamental engineering challenges of scaling, stability, and programmability.