SoylentNews

upstart writes:

Kelp forests in Marine Protected Areas are more resilient to marine heatwaves:

Using four decades of satellite images, University of California, Los Angeles (UCLA) researchers have looked at the impacts Marine Protected Areas (MPAs) are having on kelp forests along the coast of California.

They found that although the overall effect of MPAs on kelp forest cover was modest, the benefits became clear in the aftermath of marine heatwaves in 2014-2016, when kelp forests within MPAs were able recover more quickly, particularly in southern California.

"We found that kelp forests inside MPAs showed better recovery after a major climate disturbance compared to similar unprotected areas." Explained Emelly Ortiz-Villa, lead author of the study and a PhD researcher at UCLA Department of Geography.

"Places where fishing is restricted and important predators like lobsters and sheephead are protected saw stronger kelp regrowth. This suggests that MPAs can support ecosystem resilience to climate events like marine heatwaves."

Professor Rick Stafford, Chair of the British Ecological Society Policy Committee, who was not involved in the study said: "It's great to see these results and they clearly show that local action to protect biodiversity and ecosystem function can help prevent changes caused by global pressures such as climate change.

"However, it also demonstrates the need for effective MPAs. In this study, all the MPAs examined regulated fishing activity, and this is not the case for many sites which are designated as MPAs worldwide – including many in the UK."

fliptop writes:

While Canadians flocked to purchase gas-powered vehicles over the summer, electric vehicle sales continued to nosedive, according to new data from Statistics Canada:

Electric vehicle sales dropped 35.2 per cent in June compared to last year. Zero-emission vehicles comprised only 7.9 per cent of total new motor vehicles sold that month, with 14,090 entering the market.

Meanwhile, 177,313 new motor vehicles were sold in Canada in June, up 6.2 per cent from June 2024.

"In dollar terms, sales increased 3.1 per cent during the same period. In June 2025, there were more new motor vehicles sold in every province compared with the same period in 2024," reads the Statistics Canada data.

"Sales of new passenger cars increased 19.5 per cent in June 2025, marking the first gain in this subsector since November 2024. In June 2025, sales of new trucks (+4.3 per cent) were also higher than one year earlier."

Despite dwindling sales, the Carney government remains committed to its electric vehicle mandate of having 60 per cent of all vehicles sold be ZEVs by 2030 and 100 per cent by 2035, banning all motor vehicle sales.

Previously:

• Automobile Sales, Including EVs and Hybrids, Surge in 2024
• Mazda: Americans Want Cheap Gas Cars
• The EV Graveyard
• South Korean EV Battery Makers Reporting Big Losses as EV Demand Slows

Original Submission

Does 3I/ATLAS Generate its Own Light?

upstart writes:

Does 3I/ATLAS Generate Its Own Light?:

The best image we have so far of the new interstellar object, 3I/ATLAS, was obtained by the Hubble Space Telescope on July 21, 2025. The image shows a glow of light, likely from a coma, ahead of the motion of 3I/ATLAS towards the Sun. There is no evidence for a bright cometary tail in the opposite direction. This glow was interpreted as evaporation of dust from the Sun-facing side of 3I/ATLAS.

Figure 3 of the analysis paper (accessible here) [PDF] shows a steep surface brightness profile of the glow with a projected power-law slope of -3, which implies a three-dimensional emissivity profile with a radial power-law slope of -4. Such a slope is steeper than observed in solar system comets. Together with my brilliant colleague, Eric Keto, we realized that the observed slope of -4 is consistent with an alternative model in which the dust outflow around 3I/ATLAS is illuminated by a central source. This model naturally accounts for the steep brightness profile, since the outflow density slope of -2 is accompanied by the radial decline of the illuminating radiation flux with an additional declining slope of -2.

If 3I/ATLAS generates its own light, then it could be much smaller than expected from a model in which it reflects sunlight. The reflection model requires a diameter of up to 20 kilometers, which is untenable given that the limited reservoir of rocky material in interstellar space can only deliver such a giant rock once per 10,000 years or longer (see the calculation in my paper here).

Last night, we held the annual soccer cup match between the faculty and the students at Harvard's Institute for Theory & Computation, for which I serve as director. Although I scored 2 goals for the faculty team, the students won 3 to 2. Disappointed by the outcome, I focused on 3I/ATLAS as soon as I woke up the following morning.

First, I calculated that the luminosity of 3I/ATLAS needs to be of order 10 gigawatt. Second, I realized that the steep brightness profile around 3I/ATLAS implies that the nucleus dominates the observed light. This must hold irrespective of the origin of the light. In other words, the nucleus dominates over the emission from the glow around it.

The illumination by sunlight cannot explain the steep 1/R⁴ profile of scattered light, where R is the radial distance from the nucleus. This is because a steady dust outflow develops a 1/R² profile which scatters sunlight within the same emissivity profile. Sunlight would dominate the illumination in this model because a rocky nucleus would reflect only a small fraction of the solar intensity from a much smaller area than the 10,000-kilometer region resolved in the Hubble Space Telescope image. Another possibility for the steep brightness profile is that the scattering halo is made of icy particles that get evaporated as they move towards the Sun from the warm Sun-facing side of 3I/ATLAS. This would explain why there is no tail of these scattering particles. The required evaporation time must be of order 10 minutes but it is unclear whether this would lead to the observed 1/R⁴ brightness profile.

The simplest interpretation is that the nucleus of 3I/ATLAS produces most of the light. I calculated that the nucleus cannot be a thermal emitter with an effective surface temperature below 1000 degrees Kelvin or else its peak emission wavelength would have been longer than 3 micrometers with an exponential cutoff at shorter wavelengths, incompatible with the data. At higher effective temperatures, the required luminosity of 3I/ATLAS can be obtained from a source diameter smaller than 100 meters. A compact bright emitter would make 3I/ATLAS of comparable size to the previous interstellar objects 1I/`Oumuamua or 2I/Borisov, making more sense than the 20-kilometer size inferred in the model where it reflects sunlight.

upstart writes:

Revolutionary AI Tech Breathes Life into Virtual Companion Animals:

Researchers at UNIST have developed an innovative AI technology capable of reconstructing highly detailed three-dimensional (3D) models of companion animals from a single photograph, enabling realistic animations. This breakthrough allows users to experience lifelike digital avatars of their companion animals in virtual reality (VR), augmented reality (AR), and metaverse environments.

Led by Professor Kyungdon Joo at the Artificial Intelligence Graduate School of UNIST, the research team announced the development of DogRecon, a novel AI framework that can reconstruct an animatable 3D dog Gaussian from a single dog image.

Due to their diverse breeds, varying body shapes, and frequent occlusion of joints caused by their quadrupedal stance, reconstructing 3D models of dogs presents unique challenges. Moreover, creating accurate 3D structures from a single 2D photo is inherently difficult, often resulting in distorted or unrealistic representations.

DogRecon overcomes these challenges by utilizing breed-specific statistical models to capture variations in body shape and posture. It also employs advanced generative AI to produce multiple viewpoints, effectively reconstructing occluded areas with high fidelity. Additionally, the application of Gaussian Splatting techniques enables the model to accurately reproduce the curvilinear body contours and fur textures characteristic of dogs.

Performance evaluations using various datasets demonstrated that DogRecon can generate natural, precise 3D dog avatars comparable to those produced by existing video-based methods, but from only a single image. Unlike prior models, which often rendered dogs with unnatural postures—such as stretched bodies with bent joints, or bundled ears, tails, and fur—especially when dogs are in relaxed or crouched positions, DogRecon delivers more realistic results.

Furthermore, due to its scalable architecture, DogRecon holds significant promise for applications in text-driven animation generation, as well as AR/VR environments.

This research was led by first author Gyeongsu Cho, with contributions from Changwoo Kang (UNIST) and Donghyeon Soon (DGIST). Gyeongsu Cho remarked, "With over a quarter of households owning pets, expanding 3D reconstruction technology—traditionally focused on humans—to include companion animals has been a goal," adding, "DogRecon offers a tool that enables anyone to create and animate a digital version of their companion animals."

Professor Joo added, "This study represents a meaningful step forward by integrating generative AI with 3D reconstruction techniques to produce realistic models of companion animals." He further added, "We look forward to expanding this approach to include other animals and personalized avatars in the future."

Journal Reference: Gyeongsu Cho, Changwoo Kang, Donghyeon Soon, and Kyungdon Joo, "DogRecon: Canine Prior-Guided Animatable 3D Gaussian Dog Reconstruction From A Single Image," IJCV, (2025) https://doi.org/10.1007/s11263-025-02485-5

Original Submission

upstart writes:

Astronomers used the James Webb Space Telescope to investigate whether the most distant star identified in the universe is, in fact, a star cluster:

The most distant star ever discovered may have been misclassified: Instead of being a single star, the object — nicknamed Earendel from the Old English word for "morning star" — may be a star cluster, a group of stars that are bound together by gravity and formed from the same cloud of gas and dust, new research suggests.

Discovered by the Hubble Space Telescope in 2022, Earendel was thought to be a star that formed merely 900 million years after the Big Bang, when the universe was only 7% of its current age.

Now, in a study published July 31 in The Astrophysical Journal, astronomers used the James Webb Space Telescope (JWST) to take a fresh look at Earendel. They wanted to explore the possibility that Earendel might not be a single star or a binary system as previously thought, but rather a compact star cluster.

They found that Earendel's spectral features match those of globular clusters — a type of star cluster — found in the local universe.

"What's reassuring about this work is that if Earendel really is a star cluster, it isn't unexpected!" Massimo Pascale, an astronomy doctoral student at the University of California, Berkeley, and lead author of the study, told Live Science in an email. "[This] work finds that Earendel seems fairly consistent with how we expect globular clusters we see in the local universe would have looked in the first billion years of the universe."

Earendel, located in the Sunrise Arc galaxy 12.9 billion light-years from us, was discovered through a phenomenon known as gravitational lensing, a phenomenon predicted by Einstein's theory of general relativity in which massive objects bend the light that passes by them. A massive galaxy cluster located between Earth and Earendel is so large that it distorts the fabric of space-time, creating a magnifying effect that allowed astronomers to observe Earendel's light, which would otherwise be too faint to detect. Studies indicate that the star appears at least 4,000 times larger due to this gravitational lensing effect.

[...] After Earendel's discovery in 2022, researchers analyzed the object using data from JWST's Near Infrared Imager (NIRCam). By examining its brightness and size, they concluded that Earendel could be a massive star more than twice as hot as the sun and roughly a million times more luminous than our star. In the color of Earendel, astronomers also found a hint of the presence of a cooler companion star.

"After some recent work showed that indeed Earendel could (but is not necessarily) be much larger than previously thought, I was convinced it was worthwhile to explore the star cluster scenario," Pascale said..

upstart writes:

The University of Utah published an article about the 8,000-year history of Great Salt Lake and its watershed:

Photo of Great Salt Lake, taken in 2020, shows how the rail causeway built in 1959 has divided the lake into bodies with much different chemistries. On the right is lake's North Arm, which has no tributaries other than what flows through an opening in the causeway from the South Arm and consequently has much higher salinity. The red tint comes from halophilic bacteria and archaea that thrive there.

Over the past 8,000 years, Utah's Great Salt Lake has been sensitive to changes in climate and water inflow. Now, new sediment isotope data indicate that human activity over the past 200 years has pushed the lake into a biogeochemical state not seen for at least 2,000 years.

A University of Utah geoscientist applied isotope analysis to sediments recovered from the lake's bed to characterize changes to the lake and its surrounding watershed back to the time the lake took its current shape from the vast freshwater Lake Bonneville that once covered much of northern Utah.

"Lakes are great integrators. They're a point of focus for water, for sediments, and also for carbon and nutrients," said Gabriel Bowen, a professor and chairman of the Department of Geology & Geophysics. "We can go to lakes like this and look at their sediments and they tell us a lot about the surrounding landscape."

Sedimentary records provide context for ongoing changes in terminal saline lakes, which support fragile, yet vital ecosystems, and may help define targets for their management, according to Bowen's new study, published last month in Geophysical Research Letters.

This research helps fill critical gaps in the lake's geological and hydrological records, coming at a time when the drought-depleted level of the terminal body has been hovering near its historic low.

hubie writes:

The future of typography is uncertain:

Monotype is keen for you to know what AI might do in typography. As one of the largest type design companies in the world, Monotype owns Helvetica, Futura, and Gill Sans — among 250,000 other fonts. In the typography giant's 2025 Re:Vision trends report, published in February, Monotype devotes an entire chapter to how AI will result in a reactive typography that will "leverage emotional and psychological data" to tailor itself to the reader. It might bring text into focus when you look at it and soften when your gaze drifts. It could shift typefaces depending on the time of day and light level. It could even adapt to reading speeds and emphasize the important portions of online text for greater engagement. AI, the report suggests, will make type accessible through "intelligent agents and chatbots" and let anyone generate typography regardless of training or design proficiency. How that will be deployed isn't certain, possibly as part of proprietarily trained apps. Indeed, how any of this will work remains nebulous.

Monotype isn't alone in this kind of speculation. Typographers are keeping a close eye on AI as designers start to adopt tools like Midjourney for ideation and Replit for coding, and explore the potential of GPTs in their workflow. All over the art and design space, creatives are joining the ongoing gold rush to find the use case of AI in type design. This search continues both speculatively and, in some places, adversarially as creatives push back against the idea that creativity itself is the bottleneck that we need to optimize out of the process.

That idea of optimization echoes where we were a hundred years ago. In the early 20th century, creatives came together to debate the implications of rapid industrialization in Europe on art and typography at the Deutscher Werkbund (German alliance of craftspeople). Some of those artists rejected the idea of mass production and what it offered artists, while others went all in, leading to the founding of the Bauhaus.

The latter posed multiple vague questions on what the industrialization of typography might mean, with few real ideas of how those questions might be answered. Will typography remain on the page or will it take advantage of advances in radio to be both text and sound? Could we develop a universal typeface that is applicable to any and all contexts? In the end, those experiments amounted to little and the questions were closed, and the real advances were in the efficiency of both manufacturing and the design process. Monotype might be reopening those old questions, but it is still realistic about AI in the near future.

upstart writes:

NASA Challenge Winners Cook Up New Industry Developments - NASA:

NASA invests in technologies that have the potential to revolutionize space exploration, including the way astronauts live in space. Through the Deep Space Food Challenge, NASA, in partnership with CSA (Canadian Space Agency), sought novel food production systems that could provide long-duration human space exploration missions with safe, nutritious, and tasty food. Three winners selected last summer are now taking their technology to new heights – figuratively and literally – through commercial partnerships.

Interstellar Lab of Merritt Island, Florida, won the challenge's $750,000 grand prize for its food production system NuCLEUS (Nutritional Closed-Loop Eco-Unit System), by demonstrating an autonomous operation growing microgreens, vegetables, and mushrooms, as well as sustaining insects for use in an astronaut's diet. To address the requirements of the NASA challenge, NuCLEUS includes an irrigation system that sustains crop growth with minimal human intervention. This end-to-end system supplies fresh ingredients to support astronauts' health and happiness, with an eye toward what the future of dining on deep space missions to Mars and the Moon may look like.

Since the close of the challenge, Interstellar Lab has partnered with aerospace company Vast to integrate a spinoff of NuCLEUS, called Eden 1.0, on Haven-1, a planned commercial space station. Eden 1.0 is a plant growth unit designed to conduct research on plants in a microgravity environment using functions directly stemming from NuCLEUS.

"The NASA Deep Space Food Challenge was a pivotal catalyst for Interstellar Lab, driving us to refine our NuCLEUS system and directly shaping the development of Eden 1.0, setting the stage for breakthroughs in plant growth research to sustain life both in space and on Earth," said Barbara Belvisi, founder and CEO of Interstellar Lab.

Team SATED (Safe Appliance, Tidy, Efficient & Delicious) of Boulder, Colorado, earned a $250,000 second prize for its namesake appliance, which creates an artificial gravitational force that presses food ingredients against its heated inner surface for cooking. The technology was developed by Jim Sears, who entered the contest as a one-person team and has since founded the small business SATED Space LLC.

At the challenge finale event, the technology was introduced to the team of world-renowned chef and restaurant owner, José Andrés. The SATED technology is undergoing testing with the José Andrés Group, which could add to existing space food recipes that include lemon cake, pizza, and quiche. The SATED team also is exploring partnerships to expand the list of ingredients compatible with the appliance, such as synthetic cooking oils safe for space.

upstart writes:

Radio Waves Can Strengthen Sense of Smell - Neuroscience News:

Our sense of smell is more important than we often realize. It helps us enjoy food, detect danger like smoke or gas leaks, and even affects memory and emotion.

Many people — especially after COVID-19, aging, or brain injury — suffer from a loss of smell. However, there are very few effective treatments, and those that exist often use strong scents or medicines that cause discomfort in patients.

In a study published this week in APL Bioengineering, by AIP Publishing, researchers from Hanyang University and Kwangwoon University in South Korea introduced a simple and painless way to improve our sense of smell using radio waves.

Unlike traditional aroma-based therapy, which indirectly treats smell loss by exposing the patient to chemicals, radio waves can directly target the part of our brain responsible for smell, without causing pain.

"The method is completely noninvasive — no surgery or chemicals needed — and safe, as it does not overheat the skin or cause discomfort," author Yonwoong Jang said.

In the study, the team asked volunteers with a healthy sense of smell to sit while a small radio antenna was placed near, but not touching, their forehead. For five minutes, this antenna gently sent out radio waves to reach the smell-related nerves deep in the brain.

Before and after the short treatment, the authors tested how well the patient could smell very faint odors, like diluted alcohol or fruit scents, using pen-shaped odor dispensers called Sniffin' Sticks. They also recorded the patients' brain signals to see how active their smell nerves were.

The team found that their method improved subjects' sense of smell for over a week after just one treatment.

canopic jug writes:

The Creative Commons has become an unofficial UNESCO NGO partner. UNESCO is the part of UN which advances international cooperation in the fields of education, science, culture, and communication.

This new, formal status is an important recognition of the synergies between our two organizations and of our shared commitment to openness as a means to benefit everyone worldwide. As an official NGO partner, Creative Commons (CC) will now have the opportunity to contribute to UNESCO’s program and to interact with other official partner NGOs with common goals. In particular, we look forward to:

• Participating in UNESCO meetings and consultations on various subjects core to CC's mission. This will give us a seat at the table to advocate for the communities we serve and share our expertise on openness, the commons, and access to knowledge.
• Participating in UNESCO's governing bodies in an observer capacity. This will enable us to deliver official statements on matters within our sphere of expertise and contribute to determining UNESCO's policies and main lines of work, including its programs and budget.
• Taking part in consultations about UNESCO's strategy and program and being involved in UNESCO's programming cycle. This will give us opportunities to communicate our views and suggestions on proposals by the Director-General.

Previously:
(2022) New UNESCO Flagship Report Calls for Reinventing Education

Original Submission

fliptop writes:

Commerce Secretary Howard Lutnick delivered major news on Friday, confirming that the United States has finalized an investment deal with Intel, securing a 10% ownership stake in the semiconductor powerhouse. This development marks a significant step in bolstering America's position in global technology amid ongoing concerns about supply chain vulnerabilities and competition from abroad:

The agreement stems from negotiations tied to the 2022 CHIPS and Science Act, which aimed to revitalize domestic chip production. Under the terms, the U.S. gains a nonvoting equity position in Intel in return for federal funding support.

While specific financial details remain under wraps, the move aligns with efforts to ensure taxpayer dollars yield tangible returns for national interests. Intel, for its part, has committed billions to constructing advanced manufacturing facilities in Ohio, with full operations expected by 2030. This follows an $8 billion grant finalized last fall to accelerate those projects.

[...] Critics from the left may decry increased government involvement in private enterprise, but proponents argue it's essential for safeguarding national security in an era of geopolitical tensions. As Lutnick noted, the pact benefits both Intel and the public, positioning the U.S. to lead in semiconductors—a sector vital for everything from consumer electronics to defense systems.

This deal could set a precedent for future public-private partnerships, ensuring that American ingenuity drives global progress while keeping strategic assets firmly under domestic control. With operations ramping up in the coming years, the long-term impacts on the economy and technology landscape will be worth watching closely.

Intel press release. Also at Politico, Newsweek and NBC News.

Previously: Trump Administration Considering US Government Purchase of Stake in Intel

Original Submission

upstart writes:

'Quiet cracking' is spreading in offices: Half of workers are at breaking point, and it's costing companies $438 billion in productivity loss:

• "Quiet cracking" is the new workplace phenomenon sweeping offices. As AI looms over jobs and promotions stall, workers' mental health is quietly fraying. For employers, it has resulted in a staggering $438 billion loss in global productivity in the past year alone. But not all hope is lost. A career expert tells Fortune there are ways for managers and employees to course-correct.

Workers are down in the dumps about a lack of career growth opportunities and emptying offices as companies slash staffers to make way for AI, all while being put under constant pressure to do more with less.

Scared of speaking out and putting their neck on the line in a dire job climate, staff are silently but massively disengaging with their employers: Welcome to "quiet cracking."

The latest workplace phenomenon sees staff showing up and doing their job but mentally and emotionally struggling. About 54% of employees report feeling unhappy at work, with the frequency ranging from occasionally to constantly, according to a 2025 report from TalentLMS.

"The telltale signs of quiet cracking are very similar to burnout. You may notice yourself lacking motivation and enthusiasm for your work, and you may be feeling useless, or even angry and irritable," Martin Poduška, editor in chief and career writer for Kickresume, tells Fortune. "These are all common indicators of quiet cracking, and they gradually get worse over time."

Unlike "quiet quitting," this decline in productivity from workers isn't intentional. Instead, it's caused by feeling worn down and unappreciated by their employers. And oftentimes, as with burnout, they don't even register it creeping up on them until it's too late. But feeling unable to quit in protest because of the current job market, it's left them ultimately stuck and unhappy in their roles.

A fleet of unhappy workers may sound easy to spot, but the problem is sneaking up on workplaces without much course correction.

Last year, the proportion of engaged employees globally dropped from 23% to 21%—a similar dip in enthusiasm seen during the COVID-19 lockdown—costing the world economy about $438 billion in lost productivity, according to a 2025 report from Gallup.

Quiet cracking isn't only creating a bad culture for employees to work in, but the trend is also hitting businesses hard. It's imperative that bosses seize the moment to develop an engagement strategy before the problem festers into a ticking time bomb. And employees can also make adjustments to better advocate for their own career happiness.

"It isn't obvious when quiet cracking happens," Poduška explains. "You may be starting to quietly crack right now, but you wouldn't know as this type of burnout takes some time for others, and even you, to notice."

upstart writes:

Turning the lights back on:

South Australia experienced a state-wide blackout in 2016 due to a severe storm that damaged critical electricity transmission infrastructure and left 850,000 customers without power. Most electricity supplies were restored within eight hours, but it was a major event and prompted a multi-agency response involving emergency services and the Australian Defence Force.

[...] Historically, Australia has been heavily reliant on gas and coal generator units for system restart after a blackout, but those units are quickly reaching their end-of-life.

The grid has also changed significantly in the last decade alone, and today's electricity network looks very different, with large commercial wind and solar farms making up a higher percentage of Australia's generation mix every year.

Sorrell's work looks at how power systems can be restarted using large-scale, grid-forming batteries storing power from wind and solar sources as the primary restart source. While he recognises restarting the grid is not something most renewable plants were intentionally designed for in the first place, he remains confident in their ability.

"We're 100 per cent moving in a direction where large-scale batteries are going to feature prominently, if not be the primary black starter of the grid after major blackouts," Sorrell said.

During the South Australian blackout, severe weather damaged powerlines and subsequently nearly all wind turbines across the state shut down in quick succession. This was caused by a protection setting unknown to operators. Losing the turbines caused a massive energy imbalance, and with far too much load for the generation available the system collapsed. Within seconds, the whole state lost power.

"It's not because it's wrong for those protection devices to be there. They're there for very good reasons," Sorrell said.

"What the problem tends to be, and what was the case in South Australia, was that despite being compliant with existing standards, these particular settings were not present in the models that the manufacturers provided."

This meant the equipment was not being correctly represented, either in technical standards or in the simulation models that power system operators need, especially in understanding extreme circumstances.

Sorrell said there has since been a concerted effort across the industry to implement new standards in modelling so that they accurately represent the equipment in the field and their performance.

"Australia is a world-leader for setting modelling and performance standards," he said.

In his latest System Restoration and Black Start DOCX (15 MB) report, Sorrell used these next generation computer models and simulations to explore how large-scale batteries, wind and solar can actively participate in system restart.

jelizondo writes:

The Stanford Report has an interesting article on a brain interface:

Neurosurgery Assistant Professor Frank Willett, PhD, and his teammates are using brain-computer interfaces, or BCIs, to help people whose paralysis renders them unable to speak clearly.

The brain's motor cortex contains regions that control movement – including the muscular movements that produce speech. A BCI uses tiny arrays of microelectrodes (each array is smaller than a baby aspirin), surgically implanted in the brain's surface layer, to record neural activity patterns directly from the brain. These signals are then fed via a cable hookup to a computer algorithm that translates them into actions such as speech or computer cursor movement.

To decode the neural activity picked up by the arrays into words the patient wants to say, the researchers use machine learning to train the computer to recognize repeatable patterns of neural activity associated with each "phoneme" – the tiniest units of speech – then stitch the phonemes into sentences.

Willett and his colleagues have previously demonstrated that, when people with paralysis try to make speaking or handwriting movements (even though they cannot, because their throat, lip, tongue and cheek muscles or the nerve connections to them are too weak), a BCI can pick up the resulting brain signals and translate them into words with high accuracy.

Recently, the scientists took another important step: They investigated brain signals related to "inner speech," or language-based but silent, unuttered thought.

Willett is the senior author, and postdoctoral scholar Erin Kunz, PhD, and graduate student Benyamin Meschede-Krasa are the co-lead authors of a new study about this exploration, published Aug. 14 in Cell.

Willett said:"Inner speech (also called 'inner monologue' or self-talk) is the imagination of speech in your mind – imagining the sounds of speech, the feeling of speaking, or both. We wanted to know whether a BCI could work based only on neural activity evoked by imagined speech, as opposed to attempts to physically produce speech. For people with paralysis, attempting to speak can be slow and fatiguing, and if the paralysis is partial, it can produce distracting sounds and breath control difficulties."

"We studied four people with severe speech and motor impairments who had microelectrode arrays placed in motor areas of their brain. We found that inner speech evoked clear and robust patterns of activity in these brain regions. These patterns appeared to be a similar, but smaller, version of the activity patterns evoked by attempted speech. We found that we could decode these signals well enough to demonstrate a proof of principle, although still not as well as we could with attempted speech. This gives us hope that future systems could restore fluent, rapid, and comfortable speech to people with paralysis via inner speech alone."

"The existence of inner speech in motor regions of the brain raises the possibility that it could accidentally 'leak out'; in other words, a BCI could end up decoding something the user intended only to think, not to say aloud. While this might cause errors in current BCI systems designed to decode attempted speech, BCIs do not yet have the resolution and fidelity needed to accurately decode rapid, unconstrained inner speech, so this would probably just result in garbled output. Nevertheless, we're proactively addressing the possibility of accidental inner speech decoding, and we've come up with several promising solutions."

"For current-generation BCIs, which are designed to decode neural activity evoked by attempts to physically produce speech, we demonstrated in our study a new way to train the BCI to more effectively ignore inner speech, preventing it from accidentally being picked up by the BCI. For next-generation BCIs that are intended to decode inner speech directly – which could enable higher speeds and greater comfort – we demonstrated a password-protection system that prevents any inner speech from being decoded unless the user first imagines the password (for example, a rare phrase that wouldn't otherwise be accidentally imagined, such as "Orange you glad I didn't say banana"). Both of these methods were extremely effective at preventing unintended inner speech from leaking out."

"Improved hardware will enable more neurons to be recorded and will be fully implantable and wireless, increasing BCIs' accuracy, reliability, and ease of use. Several companies are working on the hardware part, which we expect to become available within the next few years. To improve the accuracy of inner speech decoding, we are also interested in exploring brain regions outside of the motor cortex, which might contain higher-fidelity information about imagined speech – for example, regions traditionally associated with language or with hearing."

Once the system works, we could have forcible installation with no password to make you "spill the beans"...

Original Submission

hubie writes:

They're cheap and grew up with AI ... so you're firing them why?

Amazon Web Services CEO Matt Garman has suggested firing junior workers because AI can do their jobs is "the dumbest thing I've ever heard."

Garman made that remark in conversation [YouTube 51:35 -- JE] with AI investor Matthew Berman, during which he talked up AWS's Kiro AI-assisted coding tool and said he's encountered business leaders who think AI tools "can replace all of our junior people in our company."

That notion led to the "dumbest thing I've ever heard" quote, followed by a justification that junior staff are "probably the least expensive employees you have" and also the most engaged with AI tools.

"How's that going to work when ten years in the future you have no one that has learned anything," he asked. "My view is you absolutely want to keep hiring kids out of college and teaching them the right ways to go build software and decompose problems and think about it, just as much as you ever have."

Naturally he thinks AI – and Kiro, natch – can help with that education.

Garman is also not keen on another idea about AI – measuring its value by what percentage of code it contributes at an organization.

"It's a silly metric," he said, because while organizations can use AI to write "infinitely more lines of code" it could be bad code.

"Often times fewer lines of code is way better than more lines of code," he observed. "So I'm never really sure why that's the exciting metric that people like to brag about."

That said, he's seen data that suggests over 80 percent of AWS's developers use AI in some way.

"Sometimes it's writing unit tests, sometimes it's helping write documentation, sometimes it's writing code, sometimes it's kind of an agentic workflow" in which developers collaborate with AI agents.

Garman said usage of AI tools by AWS developers increases every week.

The CEO also offered some career advice for the AI age, suggesting that kids these days need to learn how to learn – and not just learn specific skills.

"I think the skills that should be emphasized are how do you think for yourself? How do you develop critical reasoning for solving problems? How do you develop creativity? How do you develop a learning mindset that you're going to go learn to do the next thing?"

Garman thinks that approach is necessary because technological development is now so rapid it's no longer sensible to expect that studying narrow skills can sustain a career for 30 years. He wants educators to instead teach "how do you think and how do you decompose problems", and thinks kids who acquire those skills will thrive.

Original Submission