Toggle light / dark theme

In recent years, these technological limitations have become far more pressing. Deep neural networks have radically expanded the limits of artificial intelligence—but they have also created a monstrous demand for computational resources, and these resources present an enormous financial and environmental burden. Training GPT-3, a text predictor so accurate that it easily tricks people into thinking its words were written by a human, costs $4.6 million and emits a sobering volume of carbon dioxide—as much as 1,300 cars, according to Boahen.

With the free time afforded by the pandemic, Boahen, who is faculty affiliate at the Wu Tsai Neurosciences Institute at Stanford and the Stanford Institute for Human-Centered AI (HAI), applied himself single mindedly to this problem. “Every 10 years, I realize some blind spot that I have or some dogma that I’ve accepted,” he says. “I call it ‘raising my consciousness.’”

This time around, raising his consciousness meant looking toward dendrites, the spindly protrusions that neurons use to detect signals, for a completely novel way of thinking about computer chips. And, as he writes in Nature, he thinks he’s figured out how to make chips so efficient that the enormous GPT-3 language prediction neural network could one day be run on a cell phone. Just as Feynman posited the “quantum supremacy” of quantum computers over traditional computers, Boahen wants to work toward a “neural supremacy.”

Researchers at the Indian Institute of Science (IISc) have developed a brain-inspired analog computing platform capable of storing and processing data in an astonishing 16,500 conductance states within a molecular film. Published today in the journal Nature, this breakthrough represents a huge step forward over traditional digital computers in which data storage and processing are limited to just two states.

Such a platform could potentially bring complex AI tasks, like training Large Language Models (LLMs), to personal devices like laptops and smartphones, thus taking us closer to democratizing the development of AI tools. These developments are currently restricted to resource-heavy data centers, due to a lack of energy-efficient hardware. With silicon electronics nearing saturation, designing brain-inspired accelerators that can work alongside silicon chips to deliver faster, more efficient AI is also becoming crucial.

“Neuromorphic computing has had its fair share of unsolved challenges for over a decade,” explains Sreetosh Goswami, Assistant Professor at the Centre for Nano Science and Engineering (CeNSE), IISc, who led the research team. “With this discovery, we have almost nailed the perfect system—a rare feat.”

Researchers say they’ve built and tested a ‘structural battery’ that packs a device or EV’s chassis with energy, saving a ton of weight. It could unlock smartphones as thin as credit cards, laptops at half the weight and a 70% boost to EV range.

EVs rely heavily – pun intended – on large lithium-ion batteries to cover long distances. Researchers at Chalmers University of Technology wondered if they could build a battery that doubles as the load-bearing material holding the car together, and shed some weight.

As part of their work on what they call ‘massless energy storage,’ the research team in Sweden has developed a battery made of a carbon fiber composite. It promises similar stiffness to aluminum, while also being capable of storing a fair bit of energy – enough to be used commercially.

Back in August 2021, LA-based Portl launched a 7-ft-tall hologram projection box for life-like remote communications. Now renamed Proto, the company has revealed that its Epic technology is allowing cancer patients to consult life-size virtual specialists.

Proto was founded in 2018 by David Nussbaum, who took his experience working on huge holograms for arena gigs, movie premieres and fashion shows to produce a hologram in a box called the Epic. The idea is to plonk the machine in a venue, university, boardroom, medical facility and so on, and allow folks to chat with a life-like 3D hologram of a person who might be thousands of miles away.

So instead of a tiny image on a smartphone screen, the viewer essentially gets to interact with someone as if they’re actually in the room for a more natural communications experience. LED lighting inside the box helps with shadows and reflections for added realism, the front of the unit is touch-enabled, microphones and speakers are cooked in, and there are AI-powered cameras onboard too.

A novel acoustic attack named ‘PIXHELL’ can leak secrets from air-gapped and audio-gapped systems, and without requiring speakers, through the LCD monitors they connect to.

In a PIXHELL attack, malware modulates the pixel patterns on LCD screens to induce noise in the frequency range of 0–22 kHz, carrying encoded signals within those acoustic waves that can be captured by nearby devices such as smartphones.

The researchers’ tests showed that data exfiltration is possible at a maximum distance of 2 meters (6.5 ft), achieving a data rate of 20 bits per second (bps).

A hot potato: For almost as long as we’ve had smartphones, there has been the belief that they surreptitiously listen to our spoken conversations to serve us targeted ads; most people have experienced seeing an ad on Facebook for something they were recently talking about. It’s always been claimed that this type of privacy invasion doesn’t happen. However, a marketing agency, whose clients included Facebook and Google, has admitted to using an “Active Listening” feature that eavesdrops on conversations via phone mics to gather data.

A pitch deck from Cox Media Group (CMG), seen by 404 Media, states that the marketing firm uses its AI-powered Active Listening software to capture real-time data by listening to phone users’ conversations. The slide adds that advertising clients can pair the gathered voice data with behavioral data to target in-market consumers.