In a groundbreaking development, researchers from the University of Indiana Bloomington (UIB) have introduced a new biohybrid computer named “Brainoware,” which merges a “brain organoid” with traditional artificial intelligence (AI). This innovative system demonstrated its potential by achieving 78% accuracy in a speech recognition task, highlighting the possibilities of enhancing computing capabilities with elements of human biology.

The rationale behind this endeavor stems from the remarkable energy efficiency of the human brain. While supercomputers require 20 megawatts of power for processing over a quintillion calculations per second, the human brain accomplishes equivalent tasks with a mere 20 watts. Inspired by this efficiency, scientists explored the idea of augmenting computers by integrating them with three-dimensional clusters of lab-grown human brain cells, known as brain organoids.

The UIB team grew a brain organoid from stem cells and placed the tissue on a plate covered in thousands of electrodes. Traditional computing hardware was then employed to deliver electrical pulses to the organoid, recording its subsequent neural activity. To illustrate the potential of this system, the team converted 240 recordings of Japanese speakers uttering vowel sounds into electrical pulses. An AI was trained to predict the speaker based on the neural responses of the brain organoid to the electrical stimulation.

Throughout the training process, Brainoware’s predictions steadily improved, reaching a remarkable 78% accuracy after just two days. However, it is worth noting that Brainoware proved less accurate in speech recognition compared to a conventional computing system running an AI. Additionally, sustaining the organoid’s viability necessitated a CO2 incubator and other energy-intensive resources.

In essence, Brainoware may not outperform existing technologies, but it serves as a crucial stepping stone towards the development of more advanced biocomputing systems. Lead researcher Feng Guo emphasized that this is a proof-of-concept to demonstrate the feasibility of leveraging the biological neural network within brain organoids for computing purposes. The study opens the door to further exploration and advancements in the promising field of biohybrid computing.

By Impact Lab