Hybrid Computer Combines Lab-Grown Brain Tissue With Electronic Hardware, Performs Voice Recognition and Simple Math

The human brain is a very complex and powerful organ that is unimaginably impressive in all the processes that it performs.

So far, no computer created by mankind has even come close. The key to the brain's capability is the efficiency of the neurons to serve as both a processor and memory device, in contrast to the physically separated units in modern computers.

Simulating Brain Activities

There have been many attempts to make computers more brain-like. To date, our best effort to mimic the activity of the brain in an artificial system barely scratched the surface. In 2013, Riken's K Computer tried to simulate the biological functions of the brain. Using 82,944 processors and a petabyte of main memory, the researchers took 40 minutes to simulate one second of brain activity. This involves 1.73 billion neurons connected by 10.4 trillion synapses, around just ont to two percent of the brain.

Over the recent years, scientists and engineers have been trying to design hardware and algorithms that mimic the structure and function of the brain. Known as neuromorphic computing, this process is energy-intensive, in addition to the fact that training artificial neural networks is time-consuming.

In the quest to create artificial minds, experts try to create cyborg computers where the capabilities of the human brain are blended with the power of electronics.

Brainoware System

In a recent breakthrough, researchers have created a hybrid computer using human brain tissue grown in the laboratory. This invention joins the growing field of biological computing with the potential to advance neuroscience research models of the brain.

Led by engineer Feng Guo of Indiana University Bloomington, the team integrated real, actual, human brain tissue with electronics to produce the Brainoware system. It is slightly less accurate than a pure hardware computer that runs on artificial intelligence, but it demonstrates a significant step in computer architecture.

Brainoware uses brain organoids, or clusters of human cells that mimic organ tissue. The researchers created the organoids from stem cells which have the ability to develop into different types of cells such as neurons similar to those found in the human brain. The scientists aim to establish a connection between artificial intelligence and organoids, since both systems rely on signal transmission through interconnected nodes that form a neural network.

To develop the Brainoware system, a single organoid was placed on a plate containing thousands of electrodes which connect the brain to electric circuits. The desired input information was converted into a pattern of electric pulses which were delivered to the organoid. Then the response of the brain tissue is recorded by a sensor and analyzed by a machine-learning algorithm which decodes the relevant information.

To test the capabilities of Brainoware system, Feng and his colleagues employed voice recognition task. The system was trained with 240 voice recordings of eight individuals and the audio was converted into electric signals delivered to the organoid. They found that the mini-brain showed different reactions to each voice and generated distinct patterns of neural activity. The AI learned to interpret the responses and identify the speaker, achieving 78% accuracy rate after the training.

The combination of organoids and computers could allow scientists to harness the speed and energy efficiency of the human brain for AI applications. In the future, the research team plans to adapt brain organoids for more complex biological functions.