Imagine a future where the devices we rely on every day — from smartphones to smart home appliances — are powered by technology that uses a fraction of the energy consumed by today's computer chips.
A new circuit designed by researchers at the University of Pennsylvania could turn that vision into reality, according to MIT Technology Review.
So, what is this game-changing innovation?
It's a simple assembly of 32 resistors that can be trained to sort data, much like the machine learning models that underpin today's artificial intelligence. But unlike the power-hungry GPUs those models run on, this humble circuit learns in a radically different — and much more efficient — way.
The key lies in how the circuit encodes information. Rather than rigid 1s and 0s, it uses variable voltages — an approach known as analog computing. Analog fell out of favor decades ago, but researchers have long known it offers superior energy efficiency to digital at scale.
Aatmesh Shrivastava, an electrical engineer at Northeastern University, affirms this advantage of analog, stating: "The power efficiency benefits are not up for debate," per MIT Technology Review.
Support pets in need with these special-edition memory foam shoes BOBS from Skechers has helped over 2 million shelter pets around the world — and the charity program just announced this year’s Paws for a Cause design-winning sneakers. These "hound huggers" and "kitten kicks" sneakers are machine washable and equipped with memory foam insoles. Plus, they were designed by passionate students who were inspired by their very own rescue pets. BOBS from Skechers is also committed to donating half a million dollars to the Best Friends Animal Society this year to help every dog and cat experience the safety and support of a loving home. |
The Penn team estimates their design could be 10 times more efficient than cutting-edge AI chips.
Mimicking the human brain, the resistors in the circuit adjust their values based on external feedback, without any top-down instructions. This neuromorphic approach eliminates the need for separate processor and memory components, which suck up significant power in conventional computers.
In trials, the circuit has already proved it can classify data with 95% accuracy and perform fundamental machine-learning operations. With further development to refine the design and scale it up, this unassuming prototype could mark a turning point in the race to create faster, greener computing.
According to lead researcher Sam Dillavou, the technology has piqued the interest of several companies. "People are most interested in the energy-efficiency angle," he noted, per MIT Technology Review.
TCD Picks » Quince Spotlight
💡These best-sellers from Quince deliver affordable, sustainable luxury for all
While a commercial neuromorphic device based on the Penn team's work may still be years away, their outside-the-box thinking shines a light on the path to a more sustainable computing future — one that could make our daily digital demands a little easier on the planet.
Join our free newsletter for weekly updates on the coolest innovations improving our lives and saving our planet.
Cool Picks
