The developer of a prototype “brain-like” processor has stated that it might make artificial intelligence (AI) more energy efficient.

Emissions connected with warehouses full of computers powering AI systems have sparked concerns.

According to IBM, this prototype could lead to more efficient, lower-power AI circuits for smartphones.

Its efficiency is due to components that function similarly to synapses in human brains, according to the company.

In comparison to typical computers, “the human brain is capable of achieving remarkable performance while consuming little power,” according to Thanos Vasilopoulos, a scientist at IBM’s Zurich research lab.

According to him, greater energy efficiency means “larger and more complex workloads could be executed in low power or battery-constrained environments,” such as autos, mobile phones, and cameras.

“Additionally, cloud providers will be able to use these chips to reduce their energy costs and carbon footprint,” he added.

‘From digital to analog’
Most chips are digital, which means they store information as 0s and 1s, but the new chip employs analogue memristors [memory resistors] that can store a range of values.

The distinction between digital and analogue is analogous to the distinction between a light switch and a dimmer switch.

The human brain is analog, and memristors function similarly to synapses in the brain.

Prof Ferrante Neri of the University of Surrey explains that memristors are a type of nature-inspired computing that replicates brain function.

A memristor, like a synapse in a biological system, may “remember” its electric history.

“Interconnected memristors can form a network resembling a biological brain,” he explained.

He was cautiously hopeful about the future of chips based on this technique, saying, “These advancements suggest that we may be on the cusp of witnessing the emergence of brain-like chips in the near future.”

He did warn, though, that constructing a memristor-based computer is not an easy operation, and that there will be a number of barriers ahead for general adoption, such as material prices and manufacturing difficulties.

The use of these components makes the new chip more energy efficient, but it also has digital aspects.

This makes it easy to integrate the chip into existing AI systems.

Many phones now include AI chips to assist with tasks such as photo processing. The iPhone, for example, contains a “neural engine” processor.

IBM thinks that in the future, semiconductors in phones and cars will be more efficient, resulting in longer battery life and new applications.

Chips like IBM’s prototype may eventually help save a lot of energy if they replaced the chips in the banks of computers running strong AI systems.

They could also reduce the amount of water required to cool the power-hungry datacenters. Data centers require massive quantities of electricity to operate – a large facility will consume as much electricity as a medium-sized town.

James Davenport, professor of information technology at the University of Bath, described IBM’s discovery as “potentially interesting,” but cautioned that the chip was not a “easy to use” solution to the problem, but rather “a possible first step.”