Researchers have developed technology to enable artificial intelligence that could function like the human brain.

The team of scientists from the Optoelectronics Research Centre (ORC) at the University of Southampton worked with colleagues from the Centre for Disruptive Photonic Technologies at Singapore's Nanyang Technological University (NTU).

Together, they have developed optical microfibers drawn from specialty glasses called chalcogenides, which are based on sulfur and are sensitive to light. The light-induced effects of chalcogenides allow the fibers to be switched on or off — a property the researchers say can be used in next-generation computing applications.

The optical pulses emitted by these microfibers, made specifically from gallium lanthanum oxysulfphide, mimic real synapses. Synapses pass along electrochemical signals from one brain cell to another by carrying information from one microfiber to another.

The researchers say they have developed a range of artificial equivalents to various biological brain functions — including neural resting state and the electrical changes in the synapse when stimulated.

In the AI brain, the researchers say the changing properties of the chalcogenides will act as the changing chemical activity in the brain. Each microfiber is stimulated by light to change properties. That allows for the light to switch, which is equivalent to the firing of a nerve cell.  

"This work implies that 'cognitive' photonic devices and networks can be effectively used to develop non-Boolean computing and decision-making paradigms that mimic brain functionalities and signal protocols, to overcome bandwidth and power bottlenecks of traditional data processing," said Cesare Soci, lead researcher and professor at NTU.

To develop at least a prototype of an artificially intelligent machine that can process information using these "photonic synapses," researchers will first need to develop a physical structure to combine these microfibers into networks of artificial synapses that can send information from one end to another.

Research into high-speed computing and AI has led to breakthroughs in software and hardware that mimic brain functions and signal protocols, effectively improving the efficiency of computers. However, current computers are still massively less efficient than the human brain, which takes only five seconds and a few calories to process what takes a computer 500 seconds and 1.4 milliwatts of power.

"Since the dawn of the computer age, scientists have sought ways to mimic the behavior of the human brain, replacing neurons and our nervous system with electronic switches and memory," said co-author Dan Hewak of the ORC.

"Now instead of electrons, light and optical fibers also show promise in achieving a brain-like computer. The cognitive functionality of central neurons underlies the adaptable nature and information processing capability of our brains."  

The new research was published in the Advanced Optical Materials journal. 

ⓒ 2021 All rights reserved. Do not reproduce without permission.