Toggle light / dark theme

Large language models (LLMs) are increasingly automating tasks like translation, text classification and customer service. But tapping into an LLM’s power typically requires users to send their requests to a centralized server—a process that’s expensive, energy-intensive and often slow.

Now, researchers have introduced a technique for compressing an LLM’s reams of data, which could increase privacy, save energy and lower costs. Their findings are published on the arXiv preprint server.

The new algorithm, developed by engineers at Princeton and Stanford Engineering, works by trimming redundancies and reducing the precision of an LLM’s layers of information. This type of leaner LLM could be stored and accessed locally on a device like a phone or laptop and could provide performance nearly as accurate and nuanced as an uncompressed version.

That ordinary smartphone in your pocket could be a powerful tool for investigating outer space. In a new study, researchers at Google and CU Boulder have transformed millions of Android phones across the globe into a fleet of nimble scientific instruments—generating one of the most detailed maps to date of the uppermost layer of Earth’s atmosphere.

The group’s findings, published Nov. 13 in the journal Nature, might help to improve the accuracy of GPS technology worldwide several-fold. The research was led by Brian Williams of Google Research and included Jade Morton, professor in the Ann and H.J. Smead Department of Aerospace Engineering Sciences at CU Boulder.

“These phones can literally fit in your palm,” Morton said. “But through crowdsourcing, we can use them to change the way we understand the space environment.”

Scientists at UCL and the IIT-Istituto Italiano di Tecnologia (Italian Institute of Technology) have created a temporary tattoo with light-emitting technology used in TV and smartphone screens, paving the way for a new type of “smart tattoo” with a range of potential uses.

The technology, which uses organic light-emitting diodes (OLEDs), is applied in the same way as water transfer tattoos. That is, the OLEDs are fabricated onto temporary tattoo paper and transferred to a new surface by being pressed on to it and dabbed with water.

The researchers, who described the process in a new paper in the journal Advanced Electronic Materials, say it could be combined with other tattoo electronics to, for instance emit light when an athlete is dehydrated, or when we need to get out of the sun to avoid sunburn. OLEDs could be tattooed on packaging or fruit to signal when a product has passed its expiry date or will soon become inedible, or used for fashion in the form of glowing tattoos.

Cybersecurity firm McAfee says it’s identified a “significant global increase” in a smartphone hack that triggers extortion, harassment and drains bank accounts.

McAfee says hackers are increasingly deploying a group of malicious financial apps containing “SpyLoan” malware, posing a serious threat with more than eight million active installations around the world.

The apps are on Google Play and use the names, logos, brand colors and interfaces of well-known financial institutions.

Siddarth Kara’s bestseller, “Cobalt Red: How the Blood of Congo Powers Our Lives,” focuses on problems surrounding the sourcing of cobalt, a critical component of lithium-ion batteries that power many technologies central to modern life, from mobile phones and pacemakers to electric vehicles.

“Perhaps many of us have read how are vital for energy storage technologies,” says Eric Schelter, the Hirschmann-Makineni Professor of Chemistry at the University of Pennsylvania. “But how materials that make up such batteries are sourced can be concerning and problematic, both ethically and environmentally.”

Schelter says that mining in the Democratic Republic of Congo, which supplies about 70% of the world’s cobalt, raises concerns due to environmental degradation and unsafe working conditions, and that large-scale mining disrupts ecosystems and can contaminate , leaving lasting environmental damage. In addition, he notes that a looming cobalt shortage threatens to strain as demand for battery technologies continues to grow.

Running massive AI models locally on smartphones or laptops may be possible after a new compression algorithm trims down their size — meaning your data never leaves your device. The catch is that it might drain your battery in an hour.

Light-emitting diodes (LEDs), semiconductor-based devices that emit light when an electric current flows through them, are key building blocks of numerous electronic devices. LEDs are used to light up smartphone, computer, and TV displays, as well as light sources for indoor and outdoor environments.

Past studies consistently observed a decline in the performance and efficiency of LED devices based on two-dimensional (2D) materials at high current densities. This loss of efficiency at high current densities has been linked to high levels of interaction between excitons, which cause a process known as exciton-exciton annihilation (EEA).

Essentially, the properties of some 2D materials prompt excitons to strongly interact with each other, causing excitons to “deactivate” one another. This results in a significant waste of energy that could otherwise contribute to the lighting of LEDs.

A plan to use millions of smartphones to map out real-time variations in Earth’s ionosphere has been tested by researchers in the US. Developed by Brian Williams and colleagues at Google Research in California, the system could improve the accuracy of global navigation satellite systems (GNSSs) such as GPS and provide new insights into the ionosphere.

\r \r.

A GNSS uses a network of satellites to broadcast radio signals to ground-based receivers. Each receiver calculates its position based on the arrival times of signals from several satellites. These signals first pass through Earth’s ionosphere, which is a layer of weakly-ionized plasma about 50–1500 km above Earth’s surface. As a GNSS signal travels through the ionosphere, it interacts with free electrons and this slows down the signals slightly – an effect that depends on the frequency of the signal.