What if the next leap in computing isn’t in the transistors of computer chips but in the neurons of artificial brains? What if intelligence itself could be grown, not just programmed? 

Though the concept of using neurons to carry out computations may seem like something out of science fiction, its reality is closer than you might expect. In fact, it is the central idea driving one of the most provocative scientific developments of our time: organoid intelligence (OI). 

At its core, organoid intelligence challenges our assumptions about computation. After all, brain organoids—miniature self-organizing structures that mimic the developing human brain—appear almost alien in comparison to traditional silicon-based systems. However, brain organoids grown from induced pluripotent stem cells (iPSCs) can self-organize into complex, three-dimensional architectures that could offer something silicon-based systems could not: efficiency and adaptability. 

The human brain is a marvel of efficiency. It allows us to process vast amounts of information using just 20 watts of power, a feat no supercomputer can hope to match. Currently, researchers at Johns Hopkins University are attempting to replicate this efficiency with organoids. Early experiments done by scientists in Australia have already shown that these highly efficient organoids can learn simple tasks like playing Pong.

Given its advantages over silicon-based computing, there is an immense allure to organoid intelligence. Research published by Nature forecasts that by 2030, information and communications technologies could represent more than 20% of the world’s global electricity demand. With a paradigm shift to organoid intelligence, energy consumption could drastically decrease. Furthermore, given the natural parallel processing and adaptive learning present in brain organoids, their neural connections could allow for novel strategies in processing complex data. 

The significance of organoid intelligence goes beyond computing alone. Brain organoids may offer a human-relevant model to study neurodevelopmental disorders like Autism Spectrum Disorder (ASD). By using these organoids, researchers have managed to conduct developmental neurotoxicity testing, where they rapidly screen libraries of potential drugs to pinpoint candidates that improve neural connectivity and function.

Despite its potential, organoid intelligence still faces significant hurdles. Scaling up brain organoids to match the computational capacity of even a fraction of the human brain remains a monumental task. Current organoids are limited by their size, being typically no larger than 500 micrometers, which is roughly the same thickness as your fingernail. This limitation is due to their lack of blood vessels to deliver nutrients effectively. Researchers are currently exploring solutions like 3D bioprinting vascular networks or integrating microfluidic systems to sustain larger cultures over time. Further challenges lie in interfacing these biological systems with traditional computers. While advances in microelectrode arrays and optogenetics have made it possible to communicate with organoids, decoding their complex neural activity is, well, complex. How do we ensure that these systems are reliable enough for real-world applications?

Then there are the ethical questions. With ever-developing organoid complexity, could a brain organoid ever become conscious? Even if consciousness remains out of reach, what responsibilities do we have toward living systems designed for computation?

Despite all the obstacles and all the questions, it remains clear that organoid intelligence is going to transform the way we imagine computation. In that messy, fleshy boundary between the machine and human, we may find that the future of intelligence may lie not in the cold logic of silicon but in the warm, pulsating networks of artificial life.