In a groundbreaking experiment, researchers at Indiana University Bloomington have demonstrated a basic form of speech recognition using organoids—clusters of human brain cells connected to a computer. The potential advantage of such systems is their anticipated lower energy consumption for AI tasks compared to traditional silicon chips. Feng Guo, leading the study, acknowledges that this is a proof-of-concept and emphasizes the considerable distance yet to be covered.

Organoids: Mini-Brains and Brainoware System: Brain organoids, resembling mini-brains, are created by growing nerve cells from stem cells under specific conditions. The resulting organoids, a few millimeters wide and comprising up to 100 million nerve cells, are placed on a microelectrode array, forming the Brainoware system. This system allows both the transmission of electrical signals to the organoid and the detection of nerve cell firing.

Speech Recognition Task: For the speech recognition task, the organoids were trained to identify an individual’s voice among 240 audio clips featuring eight people pronouncing Japanese vowel sounds. Initial accuracy ranged from 30 to 40 percent, improving to 70 to 80 percent after two days of adaptive learning sessions, demonstrating the organoids’ capacity for unsupervised learning.

Challenges with Conventional AI and Biocomputing Potential: Conventional AI faces challenges related to high energy consumption and limitations of silicon chips. Guo’s team, like others exploring biocomputing, aims to leverage living nerve cells to overcome these challenges. While some experts remain cautious, recognizing deep learning’s current superiority, they acknowledge the potential for biocomputing to contribute to AI advancements.

Limitations and Future Prospects: Despite the positive strides in adaptive learning, Brainoware’s simplified task focuses solely on identifying speakers, not deciphering speech content. Additionally, the organoids can only be maintained for one or two months, posing a challenge that Guo’s team is actively addressing. Enhancing the computational power of organoids for AI computing requires overcoming these limitations.

In conclusion, the Brainoware experiment signifies a pioneering step toward integrating living nerve cells into AI systems, exploring the possibilities of more energy-efficient and biologically inspired computing methods.

By Impact Lab