IBM has declared that their prototype "brain-like" chip could enhance the energy efficiency of artificial intelligence (AI).
Worries have been expressed about the pollutants related to warehouses heavily populated with computers that are running AI programs.
IBM claimed its prototype might lead to AI chips for smartphones that are more economizing and require less battery power.
The components that work similarly to neuronal connections in the human brain are what make it efficient, the company said.
Scientist Thanos Vasilopoulos from IBM's research lab in Zurich, Switzerland, pointed out that the human brain is capable of producing impressive results while utilizing minimal energy in comparison to regular computers.
He explained to the BBC that, because of its higher energy efficiency, "it's possible to utilize large and more complex workloads in power-limited or battery-run situations", such as in cars, cellular phones, and cameras.
He also noted that cloud suppliers can leverage these chips to bring down their energy costs and their environmental impact.
Most chips are digital, utilizing 0s and 1s to store information, however, the new chip introduces memristors [memory resistors], which are analogue and can hold a selection of numbers.
You can consider the contrast between digital and analogue similar to the distinction between a light switch and a dimmer switch.
The human brain is analog, functioning similarly to the operation of memristors which are analogous to the way synapses work.
Prof Ferrante Neri of the University of Surrey indicates that memristors belong to a group of computing systems which imitate the way the brain works, referred to as nature-inspired computing.
A memristor has the potential to "retain" its electric history in much the same way as a synapse does in a biological system.
He stated that interconnected memristors could form a network that is analogous to a biological brain.
He was cautiously hopeful concerning the future of chips utilizing this technology: "These improvements indicate that we may be on the brink of seeing the arrival of chips similar to the brain in the not-too-distant future."
He cautioned that designing a memristor-based computer is no small feat and predicted issues with cost and production processes that would be encountered with mass acceptance.
The new chip is made more energy efficient through the utilization of these components, while also including digital elements.
This renders the chip simpler to incorporate into existing AI systems.
The majority of phones now possess AI chips in order to provide assistance with tasks such as processing images. An example of this is the neural engine that comes as part of iPhone's chip.
IBM anticipates in the future that chips in phones and cars could be more efficient, potentially delivering longer battery life and opening the door to fresh applications.
Eventually, chips similar to IBM's prototype could be instrumental in conserving energy if they were to replace those found in the clusters of computers used to power highly efficient AI systems.
They could reduce the amount of water necessary to cool the energy-intensive data centres. Data centres require colossal amounts of electricity to keep running - a big facility will use as much electricity as an average-sized town.
James Davenport, a Professor of IT at the University of Bath, remarked upon IBM's findings as "potentially interesting", but cautioned that the chip was not a "straightforward" way to address the issue, instead acting as "a potential initial step".
Comments