Ä¢¹½ÊÓÆµ

Skip to content

Technology

Since 2020, the Ä¢¹½ÊÓÆµâ€™s Center for an Informed Public has published “rapid researchâ€� during elections that aims to cut through the noise and confusion of online rumors that can undermine trust in election processes and voting procedures. These rumors are now so pervasive that CIP researchers have designed an interactive Rigged Election Bingo game where you can learn more about common election integrity tropes.  This year the CIP’s work is focused on:  Ensuring people have timely and accurate…

Ä¢¹½ÊÓÆµ researchers gathered eight kids on campus to explore how they played with AI toys. The kids initially were curious, asking introductory questions, such as “What is your name?â€� But after communication breakdowns, the kids turned to antagonizing the toys, calling them “uglyâ€� and “evilâ€� and joking about throwing them in the ocean.

Hossein Naghavi, an assistant professor of electrical and computer engineering at the Ä¢¹½ÊÓÆµ, explores new applications for an underused segment of the electromagnetic spectrum known as the “terahertz band.â€� Naghavi recently received a grant from the Department of Energy to create powerful yet lightweight sensors using terahertz waves, which could aid emergency responders, autonomous vehicles, security systems and more.

AI systems such as Google’s Gemini Storybook now let parents or teachers conjure illustrated, personalized kids stories. Ä¢¹½ÊÓÆµ researchers found that when six leading AI models made stories about talking animals, 57% of characters were either gender neutral or ungendered, 41% were male, and just 2% were female.

A Ä¢¹½ÊÓÆµ team studied seven popular agentic AI browsers and found that four create ways for malicious actors to bypass a fundamental cybersecurity protocol called the “same-origin policy,â€� which makes websites open in a browser unable to interact with each other’s information. Researchers ran a successful proof-of-concept cyberattack on one browser.

A Ä¢¹½ÊÓÆµ team built an AI tool that helps users turn research papers into 45-second videos. A researcher uploads a paper to the tool, which uses Google Gemini to write a short script explaining the paper. The researcher can then iteratively edit the transcript and resulting video clip.

A Ä¢¹½ÊÓÆµ-led research team created GovScape, an efficient search system for PDFs from the End of Term Web Archive. Users can look up exact keywords, like “FAFSA,â€� or use semantic search, which finds documents on a topic even if the exact search terms don’t appear on the page. Because researchers used highly efficient artificial intelligence models, processing the 10 million PDFs hosted online during Donald Trump’s first term costs less than $1,500, or about $1 per 47,000 pages.

Ä¢¹½ÊÓÆµ researchers developed an artificial intelligence system that automatically estimates the environmental impacts of making different electronic devices. The system takes only a minute to run and achieves estimates with accuracy similar to human experts’.

Ä¢¹½ÊÓÆµ researchers created BikeButler, a demo web app that lets users find personalized bike routes in Seattle. Cyclists plug in their destination and origin — just like in other mapping apps — and can then toggle sliders for eight attributes to create personalized route options.

Ä¢¹½ÊÓÆµ researchers developed a system called VueBuds that uses tiny cameras in off-the-shelf wireless earbuds to allow users to talk with an AI model about the scene in front of them. For instance, a user might look at a Korean food package and say, “Hey VueBuds, translate this for me.â€� They’d then hear an AI voice say, “The visible text translates to ‘Cold Noodles’ in English.

A team led by Ä¢¹½ÊÓÆµ researchers has created DopFone, a system that uses an off-the-shelf smartphone’s existing speaker and microphone to accurately estimate fetal heart rate. The phone mimics a Doppler ultrasound, emitting a tone and listening for the subtle variations in its echo caused by fetal heart beats. A machine learning model then estimates the heart rate.

A Ä¢¹½ÊÓÆµ and Ai2 research team built OpenScholar, an open-source AI model designed specifically to synthesize current scientific research. In tests, OpenScholar cited sources as accurately as human experts, and 16 scientists preferred its response to those written by subject experts 51% of the time.

Ä¢¹½ÊÓÆµ researchers analyzed the checkout data from the last 20 years of the 93 authors included in the post-1945 volume of “The Norton Anthology of American Literature,â€� which is assigned in U.S. English classes more than any other anthology. Sci-fi was especially popular.

New research from the Ä¢¹½ÊÓÆµ and the Toyota Research Institute explores how drivers trade off between cognitive tasks, driving and using the vehicle’s touch screen. Researchers placed participants in a driving simulator and had them complete memory tests while interacting with the simulator’s touch screen.

A new Ä¢¹½ÊÓÆµ study suggests that AI could learn cultural values by observing human behavior, similar to how children learn by watching and interacting with adults. Researchers had AI systems observe two cultural groups playing a video game. The AI systems were able to learn each group’s degree of altruism and apply this to a new situation.

Researchers at the Ä¢¹½ÊÓÆµ have developed AI-powered headphones that automatically isolate conversation partners in a noisy soundscape. The system employs AI models that listen for the back-and-forth pattern of conversation and mute voices that aren’t following the established rhythm.

A web-based method was shown to mitigate political polarization on X by nudging antidemocratic and extremely negative partisan posts lower in a user’s feed. The tool, which is independent of the platform, has the potential to give users more say over what they see on social media.

In a new Ä¢¹½ÊÓÆµ study, 528 participants worked with simulated AI systems to select job candidates. The researchers simulated different levels of racial biases for resumes from white, Black, Hispanic and Asian men. Without suggestions, participants’ choices exhibited little bias. But when provided with recommendations, participants mirrored the AI’s biases.

The Royal Swedish Academy of Sciences on Tuesday awarded the 2025 Nobel Prize in Physics jointly to John Clarke, Michel H. Devoret and John M. Martinis, “for the discovery of macroscopic quantum mechanical tunneling and energy quantization in an electric circuit.â€� Clarke, a professor emeritus of physics at the University of California, Berkeley, collaborates with the Axion Dark Matter Experiment at the Ä¢¹½ÊÓÆµ.

New research led by the Ä¢¹½ÊÓÆµ provides clear evidence that highly walkable areas lead to significantly more walking. Authors compared the steps per day of 5,424 people who moved one or more times among 1,609 U.S. cities. Across all relocations, when the Walk Score rose or fell more than 48 points, average steps increased or decreased by about 1,100 per day.

Ä¢¹½ÊÓÆµ researchers recruited self-identifying Democrats and Republicans to make political decisions with help from three versions of ChatGPT: a base model, one with liberal bias and one with conservative bias. Democrats and Republicans were both likelier to lean in the direction of the biased chatbot they were talking with than those participants who interacted with the base model.

Ä¢¹½ÊÓÆµ researchers developed the game AI Puzzlers to show kids an area where AI systems still typically and blatantly fail: solving certain reasoning puzzles. In the game, users get a chance to solve puzzles by completing patterns of colored blocks. They can then ask various AI chatbots to solve and have the systems explain their solutions — which they nearly always fail to do accurately.