Unlock the Editor’s Digest for free
Roula Khalaf, Editor of the FT, selects her favourite stories in this weekly newsletter.
Scientists have combined brain-like tissue with electronic hardware to create a speech-recognition and calculation system, advancing research into the creation of high-powered biological computers.
The work boosts efforts to build so-called neuromorphic computing devices powered by human brain cells that have the potential to learn faster and be more energy-efficient than traditional silicon-based machines. A breakthrough in this emerging field could improve artificial intelligence applications, enabling advances in fields such as medical science and treatment, researchers say.
“Brain-inspired computing hardware aims to emulate the structure and working principles of the brain and could be used to address current limitations in AI technologies,” wrote the authors of a paper on the hybrid creation published in Nature Electronics on Monday
Researchers from Indiana University Bloomington, the universities of Florida and Cincinnati, and Cincinnati Children’s Hospital Medical Center built the “Brainoware” system using a so-called brain organoid — a small three-dimensional neural structure grown from human stem cells.
The team connected computer hardware to send electrical stimulation to the organoid and read the neural activity it produced in response. The system recognised Japanese vowel sounds and predicted a mathematical map.
In a language test, the scientists gave Brainoware the task of distinguishing between eight different male Japanese speakers recorded on a total of 240 audio clips. The system got significantly better after training of its underlying algorithm, improving from about 51 per cent accuracy to approximately 78 per cent.
In a maths test, the researchers tried to get the system to predict a Hénon map, a representation of chaotic activity. Here, Brainoware proved slightly less accurate than silicon-based neural networks — but its training time was more than 90 per cent lower.
The research comes after the launch in February by an international team, led by scientists from Johns Hopkins University in the US, of a detailed road map towards “organoid intelligence”. The plan promises discoveries in computing, neuroscience and other areas of medical research.
While researchers admit that general-use biological computers may be decades away from realisation, the Brainoware experiments highlight the field’s potential.
Biological computing research is likely to generate “foundational insights into the mechanisms of learning, neural development and the cognitive implications of neurodegenerative diseases”, wrote a trio of Johns Hopkins scientists in a commentary also published on Monday in Nature Electronics.
“In the next few years, increasingly complex neural systems that can interact with increasingly complex artificial environments are likely to emerge,” said the three researchers, who were not involved in the Brainoware paper.
The advances also raises ethical questions about creating brain-like “intelligence in a dish” with the potential even to acquire basic consciousness.
“As the sophistication of these organoid systems increases, it is critical for the community to examine the myriad of neuroethical issues that surround biocomputing systems incorporating human neural tissue,” the researchers added.
Read the full article here