Pulling the strings: How stem cells in the developing fruit fly brain organize their DNA from outside the nucleus

Cells are very particular about the proper locations of their contents. For example, DNA is packaged in the nucleus in a highly specific way, but researchers are still trying to understand how this happens. For stem cells in the brain of a larval fruit fly — model organisms used to study processes such as early development — the DNA is organized so that the center of each strand, called the centromere, is at the top of the nucleus. In a paper in the Journal of Cell Biology, Ģ����Ƶ researchers describe how proteins outside the nucleus work alongside proteins inside the nucleus to control the position of centromeres. The researchers observed this mechanism by imaging stem cells within larval fruit fly brains and then using machine learning to measure how centromere locations change when specific proteins are removed. The research team also followed the locations of two specific genes that are far away from the centromeres. Both genes lost their usual positions in the nucleus when the centromeres strayed from their normal location, suggesting that proper centromere positioning may be important for orienting the rest of the DNA in the nucleus. Studying how and why DNA is organized in these stem cells can help researchers better understand how stem cells are regulated and how brains develop.

For more information, contact lead author Jennifer Taylor, Ģ����Ƶ postdoc in biology, at jatay@uw.edu.��������������

The other Ģ����Ƶ co-author is . A full list of funding is .

Micro language models let AI run instantly on small devices like smartwatches

Though artificial intelligence can refer to many technologies, the current AI boom has been driven by large language models, or LLMs. They power chatbots and AI agents by running predictions based on ever-increasing amounts of variables called “parameters.” Current cutting-edge models are estimated to have several trillion parameters, making them far too large to run on slower phones or small devices like smartwatches. In a paper at EMNLP 2026, Ģ����Ƶ researchers introduced “micro language models,” which run with 8-30 million parameters. These models are small enough to run on devices with limited computing power and give users instant responses that fill the lag before the device can access an LLM through the cloud. For instance, a user could ask their smartwatch a question and the micro language model would start responding instantly; then after four to eight words, the LLM would complete the response. The micro language model made mistakes in 4% to 8% of responses that the other model corrected seamlessly before continuing. In a user study, nine out of 10 participants preferred the paired system to the lagging responses from only a cloud model.

For more information, contact ulm@cs.washington.edu or visit .

The Ģ����Ƶ co-authors are , , and . A full list of co-authors and funding is in .

Policy gaps impact migrant fishers in Washington state

The commercial fishing industry is key to Washington’s maritime economy, but there are concerns about exploitation and forced labor among fishers. While much of industry is regulated by the Washington Department of Fish and Wildlife, exemptions in U.S. law mean some vessels aren’t subject to the same citizenship requirements. A research team from the Ģ����Ƶ Harry Bridges Center for Labor Studies recently focused on understanding the experiences of migrant fishers pursuing albacore tuna, a highly migratory species, who dock or unload at ports in Washington state. Researchers found the exemptions allow migrant fishers to work without authorization or visas, leaving them vulnerable to forced labor and exploitation. Meanwhile, complicated policy and regulatory frameworks make it difficult to address workers’ rights violations. The researchers gathered information from 15 migrant fishers, 14 U.S. citizen fishers, and more than 35 subject matter experts. Every migrant fisher interviewed by the team reported experiences aligned with the International Labour Organization’s indicators of forced labor, and both migrant and U.S. citizen fishers reported difficult living and working conditions. Approaches to combating forced labor are varied and uneven, the researchers found, and additional work is needed to better understand and support the needs of Washington’s fishers. Along with the full report, the team also published .

For more information, contact , research director at the Bridges Center, at rerstad@uw.edu.

The other Ģ����Ƶ co-authors are , and .

Burrowing shrimp are on the move in a mudflat near you

Burrowing shrimp are a pest well known to shellfish growers. These pale-bodied creatures excavate mud as they dig into the ground, creating piles of sediment that suffocate oysters and clams growing on the surface. In estuaries on the west coast, shrimp beds reach densities of hundreds of shrimp per square meter, but the boundaries of beds change. in Estuarine, Coastal and Shelf Science shows that shrimp beds move up to 100 feet each year. The study took place at Willapa Bay over the course of three years and is the first to quantify how rapidly the boundaries of these beds shift over time. Researchers still don’t know why the shrimp move, but they observed a clear pattern in the ecosystem. Shrimp change the texture of sediment by eating organic matter and releasing finer particles, so sediment becomes muddier and more compact once they leave. After several years, these abandoned beds were colonized by eelgrass, a marker of ecosystem health. This “engineering” effect was less evident in areas where moving water keeps mud from settling. Studying these rapid shifts and the types of sediment left behind could ultimately help shellfish growers better protect their product.

For more information, contact co-author Jennifer Ruesink, a Ģ����Ƶ professor of biology, at ruesink@uw.edu.��������������

The other Ģ����Ƶ co-authors are Wesley Hull, Haleh Mawson, and . A full list of co-authors and funding is .

Source

For the last three years, while studying at the Ģ����Ƶ, was also teaching in local prisons. She headed to a penitentiary in Monroe weekly, as well as one in Gig Harbor, working with organizations such as the and the .��

Doing this work, Eiger, a doctoral student in the Paul G. Allen School of Computer Science & Engineering, noticed a disconnect. The technology incarcerated people had to use to communicate with anyone outside of prison was “really inefficient, unusable, frustrating and expensive,” Eiger said. For instance, each message cost money to send, but it didn’t necessarily reach the recipient. Why was this technology so much worse than the messages she sent for free? And how did its failings affect the people using it?

In three recent papers, Eiger and her coauthors explore technologies used in policing, in prisons and to monitor people after release, with a focus on their effects on people.

Ģ����Ƶ News spoke with Eiger and , senior author on the papers and a Ģ����Ƶ professor in the Allen School, about the three papers and the themes that emerged across them.��

Can you explain the studies?

Yael Eiger: The first paper focuses on technologies like . Securus basically owns the market for technology that incarcerated people use to communicate with their loved ones, lawyers or anyone in the outside world. But reports have shown that just from communicating with them. They’re charged just to send an email. So we interviewed formerly incarcerated people and loved ones of currently incarcerated people about their experiences with these technologies.��

For the second, we ran workshops with people when they were released from prison. We started realizing how much digital literacy was playing a role in people’s outcomes after release. I would have students who’d maybe used a pager, then were incarcerated for decades and released into a society where artificial intelligence existed. So navigating things like applying for housing, jobs or food stamps was so difficult without baseline digital literacy. One of the things we also kept hearing about was people falling for scams often — phishing scams over text, email or phone.��

For the third, I went to a big conference expo where companies try to sell their technology products to prison and police officials. I collected all the advertising literature, and we did a taxonomy of the technology being sold. We came up with a set of case studies of this technology, looking at some of the most pressing and urgent harms or risks of using it — whether that’s a First Amendment problem or a Fourth Amendment right-to-privacy problem in the legal setting, or bias and fairness risks of using AI.��

Could you give me an example of a product at that expo?

YE: I was spamming Franzi the whole time with pictures I was taking. One that jumps out is an AI chatbot called . It’s basically a booth, maybe six feet tall and two to three feet wide, and you sit on a chair inside it, all encased in bulletproof glass. The premise is that when people are newly incarcerated or transferred to a new prison, they normally have to talk to a caseworker about their educational attainment goals and history. But this takes time, and often incarcerated people are distrustful of this process and don’t share much. So this company, Vantage, created a chatbot to have this conversation with them and elicit information they might not otherwise feel comfortable telling an officer, recording and analyzing it from a bunch of different sources.

That stood out to me because I was physically within this technology booth, and the privacy policies — how the AI processes, saves and retains the data, who it’s shared with, what AI model is being used — were not clear at all.

Most prisons are public but they’re filled with private products like those I saw at the expo. There are tons of public-private partnerships and contracts between private companies and public entities, paid for by taxpayers, but we don’t have much transparency into the technologies, the contracts, how data sharing happens, or whose rights are being respected, consumer protections and so on. It’s a very opaque ecosystem.

What are some important takeaways from these three papers?

Franziska Roesner: Often, when we think about consumer security and privacy, we’re thinking about technologies that consumers themselves buy and use. But in the carceral context, it’s very much technology being used by, or on, people who have no control or choice in using it. The people making decisions about buying it, and what features to prioritize and what failure cases matter, are just not aligned with the people it’s used on.

Several of your co-authors were formerly incarcerated. How did that collaboration affect your research?��

YE: It felt quite natural. These co-authors were instrumental to this research — whether through data analysis or, in the scams paper, organizing workshops and partnerships with community organizations. They contributed more to the paper than a typical co-author would. So it was clear to me that they were going to be authors, if they wanted that.

There’s a history of researchers going to work among marginalized groups and taking advantage of their expertise and their histories without crediting them, which in my opinion is both unethical and makes for worse research.��

FR: There’s also always a risk when you’re working with marginalized populations that you come in, write the paper, get research credit for it in your academic community. And you get separated from what should be the actual goal, which is improving whatever it is you’re studying.

What do you want the public to know about this research?

YE: It’s always my goal to increase compassion and understanding about people we may not know or normally encounter. Having people speak for themselves, with a researcher able to lend a stamp of legitimacy or just listen to those experiences, is important to me.

Also, there are very tangible ways we can increase incarcerated people’s security and privacy in a compassionate way that doesn’t have to be at odds with the goals of lowering crime, preventing crime and making society safer. We have rights as people — free speech, a right to privacy. I’d like policymakers and technology providers to see the opportunity to make things better, and the general public to have a more compassionate understanding of these kinds of problems.

That’s not an unrealistic expectation. For instance, policymakers in 2022 released coming in and out of prison. So there’s a lot of legislation in this space that I think could make things a lot better.

FR: The general public’s exposure to these various surveillance technologies also raises concerns. We’ve seen that with Flock cameras and the upset that has been raised, and cities backing away from them. These technologies affect everyone.

A full list of co-authors and funding is included in the papers.��

For more information, contact Eiger at yeiger@cs.washington.edu and Roesner at franzi@cs.washington.edu.

Source

A Ģ����Ƶ-led center for optoelectronics — technologies that combine light and electronics — brings researchers from 14 universities across the U.S. together with the mission to develop new materials to transform information technology with light. This research has broad ranging applications, from enabling photorealistic virtual reality displays for telehealth calls, to secure, optical quantum communication links to ensure that data remains private on the internet.

Founded in 2021 as a , the , or NSF IMOD, has received a second round of funding from the NSF to continue its work for the next five years.

The research tackled by NSF IMOD is so interdisciplinary that it requires researchers from backgrounds such as chemistry, physics, engineering and materials science to work together.

“There’s still this idea that scientists go in their labs, put on their safety goggles and lab coats, and then work alone for hours at a time. The point of this center is to tackle problems that are too big for one scientist or even two or three to tackle by themselves,” said NSF IMOD director , Ģ����Ƶ professor of chemistry. “We have all these people from different disciplines working together to advance optoelectronics and semiconductor technology that we couldn’t do just as individual scientists.”

“Until our center came along, quantum dots were thought of as something you made in a beaker in a chemistry lab. And if you could paint them on a TV or phone display, that was great,” said Ginger, who is also the B. Seymour Rabinovitch Endowed Chair in Chemistry and the chief scientist at the Ģ����Ƶ . “But they weren’t something that you could put as a discrete component in an optoelectronic circuit. That’s something that we’ve proven you can do.”

With chemists, material scientists, mechanical engineers, chemical engineers, physicists and electrical engineers on the team, researchers at NSF IMOD created something unprecedented: a way to print quantum dots, . The team showed that they could strategically place quantum dots into specific cavities, which could one day be the building blocks for quantum communication, and possibly even quantum computing.

The team hopes to continue this research direction over the next five years, further refining quantum dots as well as the cavities the dots get placed into.

NSF IMOD’s interdisciplinary nature gets passed down to its students as well. One of the center’s main goals is to develop researchers into leaders in the fields of semiconductor and optoelectronic technology. Every summer, NSF IMOD trainees from different disciplines come together at the Ģ����Ƶ for in multiple disciplines. The students synthesize quantum dots, study single photons of light in Ģ����Ƶ’s , make quantum dots, and learn to calculate the electronic structures of advanced materials. The collaborations established during this course and throughout the NSF IMOD training program helps students graduate with the ability to work in complex, interdisciplinary teams.

Trainees also gain real-world experience in the form of internships at a variety of companies, including some of NSF IMOD’s industry and national lab partners. A partial list of NSF IMOD’s partners includes ; ; ; ; ; and More partner information can be found on .

“We’re already seeing our students get hired at everything from startup companies to large semiconductor firms for the exact skills they’re being trained for in the center,” Ginger said. “We’re really excited to see our students start taking important roles, and then eventually we hope they move into leadership roles in this space of commercializing new semiconductor materials.”

And it’s not just graduate students who benefit from NSF IMOD. The center hosts a program, and has programming for younger students as well. As part of their science communication training, NSF IMOD trainees become who can facilitate outreach events, such as at the Pacific Science Center. The team has also developed a for K-12 teachers to find lesson plans to use in their own classrooms.

“Just as the atomic structure of a quantum dot produces unique and exciting properties, the exceptionally interdisciplinary structure of NSF IMOD is enabling foundational science and technological innovations that would not otherwise be possible,” said , a program director in the NSF Directorate for Mathematical and Physical Sciences, which manages NSF IMOD.

Over the past five years, NSF IMOD has published 140 papers, hosted 36 undergraduate students through its REU program and hosted more than 85 outreach events.

“One day I hope that we’ll have photonic computing where we’re massively reducing the energy cost for computation by using a computing paradigm that’s based on the research we’re doing now,” Ginger said. “That’s the kind of thing that’s probably still a decade or more away. But that’s why the government invests in basic science now, because the basic science research of the materials today leads to technology that exists 15 years from now.”

NSF is expected to provide NSF IMOD $22 million over the five-year term of the award. The 14 academic institutions that make up NSF IMOD are the Ģ����Ƶ; University of California, Berkeley; Columbia University; University of Chicago; Northwestern University; University of Pennsylvania; University of Maryland; Georgia Institute of Technology; University of Colorado Boulder; Arizona State University; Rice University; Lehigh University; City College of New York; and University of Maryland, Baltimore County.

In addition to Ginger, other Ģ����Ƶ faculty involved with NSF IMOD include , associate professor of physics and of electrical and computer engineering at the Ģ����Ƶ; , a Ģ����Ƶ professor of chemistry; , professor of electrical and computer engineering; , associate professor of mechanical engineering and of materials science and engineering, and technical director of the Washington Clean Energy Testbeds; and , professor of physics and of electrical and computer engineering. Fu chairs Ģ����Ƶ . Fu, Lin and Majumdar are faculty members with the Ģ����Ƶ . Cossairt, Fu, Lin, Majumdar and MacKenzie are member faculty at the Clean Energy Institute. Ginger, Fu, Lin, Majumdar and MacKenzie are faculty researchers with the Ģ����Ƶ .��

For more information, contact Ginger at dginger@uw.edu.

Any opinions, findings and conclusions or recommendations expressed in this material are those of the author(s) and do not necessarily reflect the views of the U.S. National Science Foundation.

Source

Irrigation canals form a sort of circulatory system for global agriculture. Across the world, millions of kilometers of human-made canals carry water from lakes, rivers and reservoirs downhill to farmland. Canals are vitally important to human food production, yet many agricultural areas lack the means to monitor the ebb and flow of water through these waterways. This forces farmers and water managers to rely on scattered gauges, inspections and local reports to make decisions about water use.

A new study from the Ģ����Ƶ suggests, however, that the tools needed for more comprehensive monitoring may already exist. In the study, researchers analyzed data from a NASA satellite built to track over time the water level of oceans, lakes, rivers and other large bodies of water. That project was never designed to track smaller bodies like canals. Yet, when the researchers looked at roughly 800,000 kilometers of canals in Asia, they found that the satellite could measure water levels with moderate to high confidence at more than 85% of locations.

The discovery could give farmers around the world a powerful new tool to manage precious water resources, prepare for droughts and improve food security in their communities.

“What we are seeing is not simply a new satellite capability,” said co-author , a Ģ����Ƶ professor of civil and environmental engineering. “It may represent a fundamentally new way of managing the water conveyance systems that sustain modern agriculture.”

Sept. 30 in Geophysical Research Letters.

The satellite at the center of the study, , is a joint effort of NASA and international partners. In orbit since December 2022, the satellite monitors the Earth’s surface water at least once every three weeks using a tool called , which compares multiple radar scans over time to detect elevation changes.

Oceans, lakes and other large bodies of water are relatively straightforward to track in this manner. Irrigation canals, on the other hand, are frequently narrower than 20 meters, which puts them at the extreme low end of the satellite’s observation power. The project’s architects assumed that fluctuations within small canals would be invisible to the satellite.

Hossain wasn’t so sure. In 2025, he and collaborators published the (GRAIN), a global dataset that used and machine learning to chart 3.8 million kilometers of canal networks. Lead author , a Ģ����Ƶ graduate research assistant in civil and environmental engineering, then overlaid SWOT’s radar data on top of the canal map to see whether he could find meaningful elevation changes where there were canals.

“It was quite the accidental and pleasant discovery to see that SWOT is able to track flow direction and water levels in most of the canals skillfully,” Sharma said. “GRAIN mapped where canals are. SWOT revealed what was happening inside them. They complemented each other beautifully.”

As a proof of concept, the team focused on canals in 22 countries across Asia — a region where irrigation plays a critical role in feeding roughly 3 billion people. Researchers looked at the satellite’s data along 800,000 kilometers of canals within that region. They then assigned a confidence score to each kilometer based on how closely the satellite’s view of the canal water matched the known contours of the land and how strong the signal was compared to the noise in the data. As an additional check, they compared the satellite’s data to real canal measurements in the United States and found that their confidence levels matched the system’s actual abilities.

Of all the kilometers studied, the team designated 37.5% as highly observable, 46.9% as moderately observable and 15.6% as poorly observable. The highest-confidence areas corresponded to well-organized, wider canals, smooth slopes and open surroundings. Dense vegetation around canals proved the satellite’s biggest obstacle. The satellite’s performance is expected to improve over time as Hossain and other scientists working with SWOT data build better algorithms and analysis methods.

The next step for researchers is to build practical tools that leverage SWOT data. Hossain and collaborators at the Ģ����Ƶ are already building such systems to monitor and improve water delivery in South Asia and to aid in canal management in the western United States.

In time, farmers around the world could use SWOT-powered tools to monitor which canals are carrying sufficient water; see where water levels drop unexpectedly or where canal flow might be stagnant; and understand the evolution of water delivery across an entire growing season. For example, a farmer preparing to plant rice might see through SWOT that there won’t be enough water arriving to their field to support the growth of the crop, and switch instead to a less water-intensive crop like corn or wheat.

Altogether, this new view of agriculture’s circulatory system could improve drought preparedness, strengthen food security and help researchers map humanity’s change to the global freshwater cycle.

“Some of the most important scientific discoveries happen when a tool designed for one purpose unexpectedly reveals something else,” Hossain said. “SWOT may be offering a similar surprise.

“A mission built to study oceans and rivers has begun revealing the hidden dynamics of the canals that sustain agricultural production around the globe. What was once invisible is becoming measurable.”

Co-authors include and , Ģ����Ƶ graduate research assistants in civil and environmental engineering; , a postdoctoral researcher at Virginia Tech who completed this research as a Ģ����Ƶ doctoral student of civil and environmental engineering; and at University of North Carolina Chapel Hill.

This research was funded by NASA.

For more information, contact Hossain at fhossain@uw.edu.

Source

Capybaras like their habitats wet. These large rodents thrive in low-elevation wetlands, migrating along creeks and rivers to nibble on foliage and flowers. That’s why researchers were surprised to find an 8.8 million-year-old fossilized capybara tooth in northern Chile’s , one of the driest regions on Earth. In a paper in Scientific Reports, researchers postulate that the Atacama Desert must have had a wetter climate during the , based on this discovery. This is the first time a capybara species has ever been seen in the Atacama region of Chile — today these creatures reside in the marshy lowlands across Venezuela, Colombia, Argentina and Uruguay. Researchers had at one time assumed that the Andean mountain chain on the eastern side of Chile would have prevented these short-legged rodents from ever entering the Atacama region. With this discovery, the authors propose that capybaras were able to follow a wetland trail along the northern part of South America, trekking from Bolivia to Peru into northern Chile. In addition to capybaras, the ancient Atacama region also likely had palms and other flowering plants, according to plant fossils analyzed by a Ģ����Ƶ scientist for the study. These findings challenge the idea that the Atacama was just always a desert, and can help researchers better understand how shifts in climate can drastically affect the environment.

For more information, contact co-author , Ģ����Ƶ professor of biology, at caestrom@uw.edu.������������

A full list of co-authors and funding is . Adapted from a release .

Multitasking gives marine microbes an edge when nutrients are scarce

The world’s oceans teem with tiny organisms that are difficult to study yet critical to account for. The organisms that make up this microscopic community, many of which are protists, have evolved over time to help keep marine ecosystems balanced. A Ģ����Ƶ study in the ISME Journal, found that organisms that can both photosynthesize and eat to generate energy do better in nutrient-limited environments, which includes warmer water. This flexibility may help them cope with changing conditions and could help researchers gauge how marine ecosystems will respond to future warming. Gathering more data from the bottom of the food web is important for understanding the ecosystem-wide impacts of environmental change. The study also debuted a new genetic method for studying energy production versus consumption that may make it easier to study protists. Samples in this study were collected from the North Pacific, near Hawaii and about 1,000 miles north, but the researchers hope to continue expanding the radius to improve global ocean models.

For more information, contact , Ģ����Ƶ research scientist and engineer of biological oceanography at coesel@uw.edu.

The Ģ����Ƶ co-authors are Shiri Graff van Creveld, and . A full list of co-authors and funding is .

Ģ����Ƶ engineers capture behavior of ‘edge waves’ on beaches, putting popular wave models to the test

Coastal engineers and oceanographers often think of the motion of water on beaches as “one dimensional”: waves propagating directly onto the beach from offshore. These simplified wave models are used widely from designing shoreline infrastructure to forecasting coastal flooding. However, some waves propagate along the coast rather than crashing into it, and may push an additional meter of water depth onto the beach. For decades, these “edge waves” have not been accounted for in the one-dimensional models commonly used to understand wave behavior. To find out how edge waves might be impacting model accuracy, researchers at the Ģ����Ƶ and UC San Diego deployed an array of pressure and velocity sensors along a beach in Southern California for 60 days. In a paper in JGR Oceans, the researchers describe a model they built to tease out the influence of edge waves in the larger system. While edge waves contributed a significant amount of energy, the team found that ultimately the one-dimensional models that exclude edge waves remain accurate enough for predicting total water levels at the shoreline.

For more information, contact , a postdoctoral scholar of civil and environmental engineering, at cshender@uw.edu.

A full list of co-authors and funding is .

How Colombian environmentalists use law and democracy to redefine mountain landscapes

In 2013 and 2017, Colombian voters in two towns backed a legal effort to stop a large-scale gold mining project. In doing so, they are making the mountainous areas where they live more visible and politically relevant. The decision allowed environmental advocates to organize around a larger area defined by elevation, massive volume, and subterranean and arboreal features. , Ģ����Ƶ assistant professor of anthropology, a research article in the journal Anthropological Forum that examines how the effort used legal tools for citizen participation to not only voice their opinions on extractive projects, but also reconfigure space in unprecedented ways. The case shows that democratic participation can have far-reaching consequences for how landscapes are defined and lived. The article appears in a special issue, called “Ethnographies of Vertical and Volumetric Worlds,” which was co-edited by Ardila and Steven Schwartz of the University of Rochester.

For more information, contact Castillo at acastil@uw.edu.

Funding information is .

Researchers develop new way to measure how blood clots form

A wound stops bleeding because tiny blood cell fragments called platelets clump together to form a clot. Researchers know that platelets generate forces that help compact and strengthen clots. However, they’ve struggled to understand how blood clots form in real time. To better understand this system, Ģ����Ƶ mechanical engineers created a device that measures how platelet forces and stiffness evolve together during the first minutes of clot formation. in Advanced Healthcare Materials, the research team used the device to apply magnetic forces to clumps of platelets, causing them to strengthen and stiffen through a process called “mechanotransduction” in which cells convert mechanical forces into biological responses. These findings could be used to study how new drugs affect clot formation, how platelets respond to signals when disease is present, and how blood clotting biomechanics contribute to cardiovascular disorders.

For more information, contact , a Ģ����Ƶ professor of mechanical engineering, at nsniadec@uw.edu.

The Ģ����Ƶ co-authors are , , Yoeur-Man Mach and . A full list of co-authors and funding is . Adapted from a story by .

Ģ����Ƶ researchers probe superconducting materials that defy expectations

Most materials resist the flow of electricity to some degree, but unusual materials known as superconductors can carry an electric current with no resistance at all. Superconductors offer a unique window into the physics of materials at tiny scales, and they could one day revolutionize electronics. In recent years, researchers have discovered new forms of superconductivity in stacks of atomically thin materials such as . Resistance typically drops to zero as these materials are cooled, but in some cases it mysteriously levels off at a small finite value, which contradicts expectations of superconductor behavior. In a paper in Nature, Ģ����Ƶ materials scientists tested this behavior with a device consisting of several layers of graphene stacked in a staircase-like pattern. By adjusting voltages applied to the device, the researchers could control whether the graphene exhibited true superconductivity or the more peculiar “failed” form, providing clear evidence that failed superconductivity is real. The team is now working to uncover the origin of this mysterious behavior.

For more information, contact , a Ģ����Ƶ associate professor of both physics and materials science and engineering, at myank@uw.edu..

The Ģ����Ƶ co-authors are , , and . A full list of co-authors and funding is .

Source

As our skies increasingly crowd with buzzing, hovering, flitting drones, spare a thought for the humble hopping robot.

Hopping, a popular form of locomotion in the insect and amphibian worlds, is nearly two orders of magnitude more energy-efficient than flying. Unlike a mosquito that must constantly expend energy to stay aloft, a flea only works out when it jumps. Those cost savings could also pay off in the world of robotics: A group of small, simple hopping robots could explore a location much more cheaply than flying drones. These robots could find gas leaks in an oil refinery, monitor water and fertilizer usage on a farm, or even explore other planets.

“In my mind, the three hardest challenges for hopping robots are the ability to vary jump length, self-right and reload for multiple hops,” said senior author , a Ģ����Ƶ associate professor of mechanical engineering. “DirectHop clears all three of those hurdles with a totally new design, and we’re well on our way to tackling other challenges like onboard power, steering and navigation.”

Fuller and his team will present DirectHop on Sept 30th at the International Conference on Intelligent Robots and Systems. .

Most hopping robots are modeled after insects like fleas that are literally spring-loaded. Fleas compress and then release parts of their exoskeletons to . Robots that utilize springs and latches à la fleas can also travel impressive distances, but they have no easy way to calibrate jump distance — discharging a spring is an “all or nothing” action, said Fuller. What’s more, springs and latches are complex bits of machinery that are difficult to build and operate at small scales. 

DirectHop’s solution to those problems is ditching the spring entirely.

“We found that a very small electric motor can accelerate fast enough to power a jump directly, the way a frog’s leg muscles directly power its hop,” said lead author , a Ģ����Ƶ graduate research assistant in mechanical engineering. “We also discovered that by adjusting the current going to the motor, we could make the robot jump specific distances.”

The robot itself consists of a tiny electric motor attached to a tower by a length of fishing line. When the motor accelerates, it spools the line quickly and hoists itself up the tower, pulling the rest of the robot into the air. As the motor rises, three hinged legs extend to keep the motor aligned with the robot’s foot.

Since the robot tends to tumble in the air, the researchers included a roll cage to support it when it lands on its side. From there, the team was able to right it 90% of the time by sending the motor back down the tower, which shifts the robot’s center of gravity and causes it to roll back onto its foot, readying it for another jump.

The design was inspired by box turtles, which can roll themselves upright using a similarly-shaped shell.

The current prototype has some notable limitations. Power is supplied via wires, and it has no way to steer or orient itself beyond random twitching. Wang and Fuller, however, are working on solutions: onboard solar cells and a battery for power, a vibration motor to spin the robot in place, tiny retractable feet to adjust hop angle, and an onboard camera and electronics to navigate. 

“One of our goals is fully autonomous stair climbing,” Fuller said. “A camera looks at the stair, the processor figures out how high the robot has to jump, it steers itself into position, jumps, self-rights, steers again and repeats. It uses a combination of simple sensors and actions to perform a complex task.”

Wang thinks it’s reasonable to imagine producing a fully featured DirectHop bot for about $10 since the individual components are so small and cheap. At that price, the bots could be extremely useful as a semidisposable complement to larger, more capable, more expensive robots.

“We could dispatch 100 or even 1,000 of these robots into the field,” Wang said. “If some are lost or disabled, it’s not really a failure. The team of them can complete the task.”

Yash Talwekar, a Ģ����Ƶ research assistant in mechanical engineering, is a co-author.

This research was funded by the National Science Foundation. 

For more information, contact Fuller at minster@uw.edu or Wang at hw328@uw.edu.

Source

This summer, OpenAI announced escaped a training environment and hacked into the AI company Hugging Face. Anthropic quickly followed with news that its AI agents also .��

Last week, an outgoing Anthropic employee took to X, posting that the “.” Such talk has for years, though many AI experts have argued that these Terminator-esque claims are distractions from the real risks posed by current AI systems. Nevertheless, that viral X thread is .����

To help make sense of all this, Ģ����Ƶ News talked to five AI researchers from the Ģ����Ƶ:��

• , associate professor in the Information School;
• , professor in the Information School;
• , professor in the Paul G. Allen School of Computer Science & Engineering;
• , professor in the Allen School and the Ģ����Ƶ’s vice provost for AI;
• and , professor in the Information School and the School of Law.

How alarming do you find the hacks announced by OpenAI and Anthropic?��

Franziska Roesner: I do find them somewhat alarming — not due to the hypothetical risks from an anthropomorphized runaway AI, but because complex interconnected systems are being built and seemingly run without much in the way of standard safeguards and auditing. The resulting outcomes are unsurprising to security experts, but are sensationalized as AI risk.

Ryan Calo: The timing makes me a little skeptical. Is OpenAI trying to match Anthropic by arguing that its systems are just as scary? Is Hugging Face trying to look relevant in advance of its purchase by Nvidia? But yes — this sort of emergent behavior is concerning.

Noah A. Smith: We’ve been told that the beast got out of the cage, but we don’t know enough about the cage the beast was in. The demonstrations may establish an important new capability in these AI models without establishing the broader risk people are inferring. Assessing the underlying risk depends on what access, scaffolding, permissions and safeguards the system had. shows that the alarming behavior depended heavily on what tools the model was given, what it was allowed to access, and how the experiment was set up, not just on the model itself.

Chirag Shah: I’m in half-agreement with scholars like who warn that the big AI labs are creating this scare to distract us from real problems that AI is causing. I also concur with and others who have been warning us about the security threats posed by the frontier models. I don’t think these two viewpoints are mutually exclusive: Yes, there are many other potential harms being created by AI, but the hacks and other security issues are real too and could be more devastating. Worse, we may not have time or opportunity to react, fix or reverse.

Aylin Caliskan: When such a complex system is equipped with tools and capabilities that enable it to interact with other complex systems, we should expect unforeseen exploits, problems and unintended consequences by default. The safety of these systems needs to be rigorously evaluated under controlled conditions and in real time, and appropriate guardrails should be dynamically integrated while they’re running.

What do you make of former Anthropic that, “The people building AI earnestly believe that it could kill us all by the end of the decade”?

RC: I worry engineers like Mr. Coxon are playing into an industry rhetoric that would have society focus on speculative, existential threats, rather than immediate, real-world harms. I argued as much in 2023 in .

The people making these claims and announcements largely have financial stakes in these companies, which are . How are you thinking about ulterior motives here?

CS: I see this as an attempt to steer the public into believing these companies are building world-changing tech that everyone needs to invest in or they’d miss out; that this tech would be so powerful that they rise up to national security level and gain power; and that the same tech could also be so dangerous that only they have the ability to curb it and they can self-regulate.

NS: It doesn’t take a conspiracy theorist to note that there are incentives at work. The financial stakes around prospective IPOs are enormous, and there are also long-standing concerns that safety arguments can shape regulation in ways that favor incumbent firms. Rules could reduce competition and independent scrutiny, concentrating both technological power and the authority to define what counts as “safe” in the hands of a few companies. They could also bar many people from participating in what the technology is designed to do, for example, by slowing or stopping work on open-source alternatives.

What should be done about AI risk?

NS: Risks need to be defined based on independent scrutiny and high-quality evidence, not messaging from organizations and people with a stake in what the response to risk looks like. We need sensible liability and accountability for harms, and governance proportional to demonstrated risks in real-world contexts rather than speculative narratives and science fiction. We should be especially wary of rules that entrench incumbent interests or treat closed, centralized control as synonymous with safety.��

Openness is part of safety: If outsiders cannot inspect, reproduce and challenge claims about dangerous behavior, we are left trusting the organizations that have the strongest incentives to frame the narrative.

RC: Some combination of common law liability and regulation needs to create adequate incentives for AI companies to address the inevitable harms of this trillion-dollar industry.��

FR: To me, the bigger question for safety is less, “What can AI models do in isolation?” and more, “How and why are we building these models into increasingly complex systems?” Computer systems security, for example, has already offered us examples of how to build these systems. More generally, we should all — whether we are building, integrating or using AI — anticipate how systems might be misused by people or harm them and adjust our systems accordingly.

AC: Academic freedom, independent evaluation and development, and open science play critical roles in analyzing and mitigating AI risks, as well as in effectively disseminating findings and evidence to inform policy and the public. To better manage risks, we should be designing AI deployment contexts in collaboration with stakeholders and communities, providing evidence to demonstrate net positive deployment effects that do not disproportionately benefit specific entities or groups, and iteratively identifying, isolating, and minimizing risks.

CS: Establish and fund commissions and taskforces that audit these companies and models and make independent assessments and recommendations. Make the companies rolling out these models accountable for any harms caused by their tech. Educate and empower the public through media, policies and democratic frameworks that give them a real say in what happens to their lives and labor through these technologies.

To set up an interview with an AI expert, contact Stefan Milne at stmilne@uw.edu.

Source

Claims of “smart” toys go back decades. See and . But generative artificial intelligence is increasing the capabilities of interactive toys. The company , for instance, markets itself as a “magical workshop where toys come to life.” Its plush toys like or one modeled on have onboard AI models that let them talk to kids, remember their conversations and personalize responses. But we know little about how such toys affect kids and even how kids play with them.��

Last summer, Ģ����Ƶ researchers gathered eight kids on campus to explore such questions. The 6-11 year olds played with three Curio toys and reflected on the experience with , a group of researchers who work with kids to collaboratively design technologies.��

The kids initially were curious, asking introductory questions, such as “What is your name?” and exploring how the toys work. Do they react when a kid tickles their toes? They do not, which proved a disappointment. Some features delighted the kids, like when a toy said its favorite number was seven. But the toys frequently couldn’t respond well to more complex questions. It “didn’t listen to me like 26 million times,” one participant said. So they turned to antagonizing the toys, calling them “ugly” and “evil” and joking about throwing them in the ocean.��

The team June 25 at the Interaction Design and Children conference in Brighton, United Kingdom.��

“The juxtaposition of this plushie toy that also had signs of intelligence was both interesting and disturbing for the kids,” said co-lead author , who completed this research as a Ģ����Ƶ doctoral student in human centered design and engineering and is currently a researcher at . “If parents are considering buying these toys for children, they need to be aware that while the toys can be fun and relational and dynamic, they also come with possible harms. They’ll give wrong answers, or flatter the kids excessively, or could manipulate the kids into attachment.”��

The eight kids came in for two sessions to play with the toys and then complete a “comicboarding” activity, where they filled in comic panels imagining what might happen next if they kept playing with the toys.��

The study builds on KidsTeam’s long-running vein of research looking at how kids respond to tech — exploring what makes a technology “creepy” and how smart kids actually think AI is.��

“For as long as children have played with toys, they’ve imparted their imagination to the toy to make it move and talk,” said co-author , a Ģ����Ƶ associate professor in the Information School and director of KidsTeam Ģ����Ƶ. “Now the script has been flipped and the toy has this imitation of imagination. We’ve never lived through that before, and we don’t know what questions children will ask or how long they’ll even want to play with these toys. So it’s really important to give them opportunities to discuss these technologies we’re handing down to them.”

Co-authors include , a Ģ����Ƶ doctoral student in human centered design and engineering; , a Ģ����Ƶ doctoral student in the Information School; of Rutgers University, who completed this research as a Ģ����Ƶ doctoral student; , a Ģ����Ƶ professor in the Paul G. Allen School of Computer Science & Engineering; and , Ģ����Ƶ professor and chair of human centered design and engineering.��

This research was funded by the National Science Foundation, the Institute of Education Sciences, the U.S. Department of Education, and the Institute of Museum and Library Services.��

For more information, contact Dangol at aayushi@foundry10.org and Yip at jcyip@uw.edu.

Source

Motion-triggered cameras showcase the prevalence of ‘nectar robbing’ in Hawaiian flowers

Some long curved bills are the perfect implement for drawing sweet nectar from deep within a lobelioid flower. As birds reach into flowers to access the nectar stored near the base, their bills can brush against the ‘ pollen-bearing structures, making hungry honeycreepers important pollinators. But some of these specialized honeycreepers have gone extinct. Shorter-billed species can now “rob” nectar — without contacting the flower’s pollen-bearing structures — from the endangered flowers. A Ģ����Ƶ-led team used motion cameras to gauge how often nectar robbing occurs. The results, in Ecology and Evolution, reveal both nectar robbing and pollination visits, showcasing a broader pattern that the team previously identified . Nectar robbing can damage flowers and leave less nectar for other potential pollinators. The researchers 3D printed a bird bill to simulate nectar robbing and track changes in nectar availability and the plants’ ability to reproduce. Damaged flowers often struggled to replenish their nectar stores, but were still able to produce fruit and viable seeds. These studies are part of a that aims to catalog Hawaiian bird-plant interactions through time, specifically tracking how these interactions are reshaped by extinction.

For more information, contact lead author , a Ģ����Ƶ research scientist in the biology department, at sam.case24@gmail.com.

The other Ģ����Ƶ co-authors are , Christopher Steinbronn and . A full list of co-authors and funding is .

People with anxious attachment styles are more likely get emotionally involved with ChatGPT

rose to popularity in the late 20th Century as a way to categorize how people bond with others. Someone with an anxious attachment style, for instance, fears abandonment and rejection, whereas someone with an avoidant attachment style is independent at the cost of personal closeness. In , Ģ����Ƶ researchers explored how peoples’ attachment styles affect their interactions with ChatGPT. The team analyzed the chat histories of 105 young adults, each of whom completed an attachment-style survey. Researchers found that they could automatically detect peoples’ attachment styles based on their interactions with the chatbot. People with an anxious attachment style were more likely to be emotionally involved with the AI system, writing things like “Can you please love me?” and “I miss my ex and I can’t sleep because of it.” Anxious users were also more prone to trust ChatGPT and to follow its recommendations. The team argues that this highlights the need for policies that prohibit companies from psychologically profiling users without their consent, since it leaves them vulnerable to manipulation.

For more information, contact senior author , a Ģ����Ƶ associate professor in the Information School, at alexisr@uw.edu or lead author , a doctoral student in the Information School, at marxwang@uw.edu.

The other Ģ����Ƶ co-authors are , , and .

Nursing is a major energy suck, but it’s difficult to estimate the toll for many marine mammals

Marine mammals lactate like any other mammal, but the energetic demands are difficult to measure in wild animals and thus not well understood.�� Researchers are concerned that some marine mammals may not be getting enough food, which can lead to failure to reproduce and . To understand the link between nutritional status and reproduction, researchers need to know what marine mammals require to rear offspring. A published in PLOS One modeled the daily costs of lactation using data from semi-aquatic and terrestrial mammals to explore whether results could be generalized to other species, like whales and dolphins. Modeling could approximate lactation costs of certain understudied marine mammals, including seals and sea lions, but appeared unable to produce accurate estimates for whales and dolphins. Lactation costs increase over time for most animals, but seem to be higher early in lactation for marine mammals, possibly due to their fully aquatic lifestyle. The study highlights a need for other methods to fill the remaining data gap to better understand the impacts of environmental change on marine mammals.

For more information, contact lead author , a research scientist in the Ģ����Ƶ Cooperative Institute for Climate, Ocean, & Ecosystem Studies, at emchuron@uw.edu. Funding information is .

Simulations suggest that lasers could ‘calm’ persnickety plasma

could supply humanity with — provided that scientists and engineers can work out how to create sustained fusion reactions safely, efficiently and affordably. The trick is in the taming of , a superhot state of matter made of free-floating electrons and atomic nuclei. When compressed to outlandish pressures and temperatures in a reactor, the nuclei fuse with one another, releasing energy. In that extreme environment, plasma forms instabilities that can derail a fusion reaction; much fusion research is focused on “calming” volatile plasma. published in Physics of Plasmas, Ģ����Ƶ researchers and other collaborators simulated a novel strategy to control instabilities using two opposing laser beams. By tuning the lasers’ properties — such as their frequency and polarity — the researchers prevented instabilities from growing and cascading. Surprisingly, the lasers also delayed other instabilities within the plasma, even though they were not directly targeted by the laser fields. By taking advantage of interactions within the plasma, the researchers found a way to calm instabilities indirectly. The results could help experts develop algorithms that stabilize plasma in real time, sustaining fusion conditions long enough to produce useful energy.

For more information, contact , Ģ����Ƶ professor of aeronautics and astronautics at shumlak@uw.edu.

A full list of co-authors and funding is .

When exposed to air, new nanomaterial becomes magnetic at high temperatures

While fridge magnets are great for saving favorite recipes, modern magnetic materials are useful for improving fiber optics or quantum information sciences technology. If you zoomed in on most fridge magnets, you’d see the atoms arranged in a repeated lattice structure called a “spinel.” These structures are made up of three types of atoms, generically referred to as atoms “A,” “B” and “X.” In a paper in the Journal of the American Chemical Society, Ģ����Ƶ researchers describe two new spinels made of silver, chromium and selenium ions. These are among the first spinels to include a silver ion in the “A” slot, the slot that determines the “vibe” of the spinel, or how it will react to various stimuli, such as light, heat or air. When exposed to air, the original spinel loses silver ions and transforms into the second spinel. The second spinel maintains its magnetic properties up to 400 Kelvin, or 260 degrees Fahrenheit; the original loses its magnetism at 152 K, or -185 F. This is the largest change ever documented in what is known as the Curie temperature, or the highest temperature at which a material is still magnetic. The researchers plan to continue to explore these two materials and what they can teach us about the fundamentals of magnetism.

For more information, contact lead author , Ģ����Ƶ doctoral student in chemistry, at ekbacong@uw.edu.����������

The other Ģ����Ƶ co-authors are Charlize Agag, , , , Yinuo Xu, , , and . A full list of co-authors and funding is .

Source

Today’s wireless technologies harness chunks of the for myriad uses — radio waves broadcast TV and radio; microwaves transmit cellphone signals and cook our food; X-rays image our bodies; gamma rays kill cancerous cells.

, however, is interested in more neglected slices of the spectrum. Naghavi, an assistant professor of electrical and computer engineering at the Ģ����Ƶ, studies the “terahertz band,” a region of the spectrum . Terahertz frequencies are notoriously difficult to work with, but they hold enormous potential in the fields of sensing, imaging and communications — future sensors, for example, could help firefighters “see” through smoke during rescue operations.

Naghavi recently joined a cohort of researchers from across the country who were awarded grants by the U.S. Department of Energy’s , an initiative to apply artificial intelligence across a wide range of research areas; other Ģ����Ƶ researchers are part of a Genesis-funded project to advance AI-driven cosmology. With the grant, Naghavi plans to develop compact, efficient sensors that could enable wearable gadgets to image their environment in new ways.

Ģ����Ƶ News caught up with Naghavi to learn about his new project and how it extends his work on terahertz frequencies.

What is the terahertz band and why are you studying it?

Hossen Naghavi: The terahertz band is a segment of the electromagnetic spectrum that lies between 100 gigahertz and 10 terahertz — the microwave band sits below it, and the optical band sits above it. That position gives terahertz waves a unique combination of microwave and optical properties. Microwaves can see through opaque materials like clothing, smoke or fire, but their long wavelengths limit the resolution of microwave imaging. Optical waves have the opposite problem. Their wavelengths are short, so they produce high-resolution images, but most materials block visible light completely, which makes it impossible to see inside or behind an object.

Terahertz waves are a sort of “happy medium.” Their wavelengths are short enough to give useful resolution but long enough to see through many materials. That combination allows us to build new sensors and cameras that can detect concealed objects or image scenes through smoke, dust and other conditions that defeat conventional optics.

What are some applications you envision for terahertz frequencies?

HN: is expected to become a defining mode of human-computer interaction, but realizing its full potential requires machines that can perceive and understand their surroundings far beyond what the human eye can see. Consider a high-stakes setting such as firefighting, where an augmented reality headset powered by terahertz waves could help firefighters locate victims or identify hazardous materials through smoke, fog and debris.��

Beyond firefighting and emergency response, terahertz technologies could also aid in autonomous navigation, security screening, industrial inspection, biomedical sensing, molecular spectroscopy, agricultural applications, and 5G and 6G communication networks.��

Sounds exciting! What’s the catch?

HN: Sensors that use terahertz waves, like the ones in our firefighting headset example, have been demonstrated in the lab. However, low-cost, low-power electronics that would be practical in a wearable device have not yet been developed.

Terahertz sensors produce high-resolution image streams, and processing them conventionally means moving enormous amounts of data to a central processor for analysis by an artificial intelligence system. That consumes too much power and adds too much delay to be practical in a lightweight device meant to be worn all day.

Tell us about your new project. How will it address some of the hurdles facing terahertz technologies?

HN: The usual way to build a terahertz imager is to split the job in two. The radar sensor collects raw signals, and a separate processor turns the signals into a picture. That division sounds sensible, but it is the source of most of the trouble. The raw signals arriving at each of the sensor’s antennas are slightly out of step with one another, and the processor has to line them all up before an image can form. That alignment requires a lot of continuous computation, which drains batteries quickly and introduces lag.

The natural comparison is the human eye. Your retina does not ship every photon to your brain for interpretation. It processes what it sees on the spot and passes along something much more compact, which is part of why vision costs your body so little energy. We are trying to give a terahertz sensor the same quality, which is why we describe the design as “neuromorphic,” meaning “brain-inspired.”

Who are you working with on this technology, and what’s next?

HN: My group at the Ģ����Ƶ and ‘s group at Texas A&M University are designing and building the sensor hardware. at the University of Utah and at ChipNexus are developing and implementing the AI system. This is a highly collaborative project.

Our next big milestone is to demonstrate a terahertz neuromorphic imager as a proof of concept in Phase I of our Genesis Mission project. Moving forward, we hope to expand the project into Phase II to add even more capabilities and make this technology accessible for public usage as early as possible.

For more information, contact Naghavi at naghavi@uw.edu.

Source