Skip Navigation
Johns Hopkins Bloomberg School of Public HealthCAAT

Organoid Intelligence

Organoid Intelligence illustration
The human brain is unmatched by modern information technology: In 2013, the world’s fourth largest computer took 40 minutes to model one second of 1% of a human’s brain activity[1]. The storage capacity of a single human brain has been estimated at 2,500 TB[2]. To give some perspective on energy requirements, it would take 34 massive coal-powered plants generating 500 megawatts each to equal the power demands that US based data centers would require to hold this much data (Masanet et al., 2016). Brains, on the other hand, have been estimated to be about 100,000 times more energy-efficient than computers, and could meet the same data storage capacity with only 1600 kilowatts of energy, illustrating the immense implications of brain-directed computing for energy conservation.

Besides massive increase in energy efficiency, it is becoming more and more clear that the human brain outperforms man-made computers on many levels such as decision taking on incomplete datasets. For this project, we are proposing to learn from and utilize the inherent computing capacity of human brain organoids; to harness the computing power from human brain organoids.

The last six years have seen a revolution in brain cell culture moving away from traditional flat monolayer cultures to more organ-like three-dimensional (3D) organized cultures. In 2016, the PI’s group developed the first standardized and mass-produced brain organoids from human induced pluripotent stem cells (Pamies et al., 2017). These brain organoids are spontaneously electrophysiologically active and were the first human brain model to show myelination of axons, which makes electrical conductivity about 100 times more efficient. Cell density is about 1,000fold higher than in 3D boosting synapse formation. Each of these brain organoids is made up of about 30-40 thousand cells and is thus roughly one three millionth the size of the human brain (theoretically equating to 800 MB of the brain’s memory storage). Our team is currently producing thousands of these per week in a small incubator and is fully prepared to scale this effort.

To date, the term biological computing has been mainly used to describe the use of biochemical data storage systems (i.e. using DNA as a medium for digital information storage). Our vision, however, is to expand this definition to include brain-directed computing (Organoid Intelligence, O.I.); to scale the current brain-organoid model from ~30,000 neural cells to ~10 million (1 gram (g) of cultured biomatter), and use biofeedback to systematically train it with increasingly complex sensory inputs and output opportunities. In essence, we will move to O.I. by interfacing the brain organoid with computers, sensors and machine interfaces to facilitate supervised and unsupervised learning. Ultimately, this field of study could enable many complex functions of software-based artificial intelligence (A.I.) to be performed more efficiently, faster, and at a fraction of the energy costs.