MIT Media Lab

Incarcerated people in US prisons are particularly susceptible to increased summer temperatures, such as the extreme heat waves of this past summer. In a new study, MIT researchers including Media Lab and MIT AeroAstro alum Ufuoma Ovienmhada and Professor Danielle Wood combine high-spatial-resolution air temperature data with census data and information from the US Department of Homeland Security to map summertime heat exposure in prisons across the United States. The study, published in the American Geophysical Union’s GeoHealth journal, provides policymakers and community leaders with ways to estimate, and take steps to address, a prison population’s heat risk, which could worsen with climate change. “This isn’t a problem because of climate change. It’s becoming a worse problem because of climate change,” Dr. Ovienmhada says. https://lnkd.in/e-ZBmf8E

As climate-related disasters become more frequent, the need for actionable climate intelligence has never been greater. Earth Observation (EO) technologies, including advanced satellite sensors, algorithms, and AI tools, are helping to improve our understanding of our changing environment. In a new white paper prepared in cooperation with the World Economic Forum, Media Lab Director Dava Newman and Research Engineer Minoo Rathnasabapathy examine key EO technology trends that are unlocking new climate insights.

In ComputerWeekly.com, Media Lab researcher Pat Pataranutaporn and other experts discuss the potential consequences of developing and using autonomous weapons systems. "The most dangerous AI isn’t the Terminator-type,” Dr. Pataranutaporn says. “Because its evil intent is obvious. The real danger lies in AI that appears friendly but subtly manipulates our behavior in ways that we can’t anticipate."

Unlock the potential of computer vision through the physics of lighting. On September 25, explore solutions with MIT Media Lab alum and Ubicept founder Tristan Swedish. ?? Join the conversation with MIT Horizon: https://bit.ly/3N291q5 #OnlineLearning #OnlineEducation #LifelongLearning #ProfessionalDevelopment #Training #Computer #Technology #Physics

Using custom, 3D-printed glass bricks manufactured by Media Lab spinoff Evenline, MIT researchers have created interlocking glass bricks strong enough to be used as construction materials. This glass masonry could be reconfigured and reused many times over as recyclable bricks for building facades and internal walls. “Glass as a structural material kind of breaks people’s brains a little bit,” says Michael Stern, a former Media Labber and founder of Evenline. “We’re showing this is an opportunity to push the limits of what’s been done in architecture.” The work is published in Glass Structures & Engineering. Authors: Daniel Massimino (MIT Department of Mechanical Engineering (MechE)); @Ethan Townsend (Evenline); Charlotte Folinus (MIT Department of Mechanical Engineering); Michael Stern (Evenline); Professor Kaitlyn Becker (MIT Department of Mechanical Engineering) https://lnkd.in/dgXFmjvD

In a recent paper published in Acta Astronautica, MIT researchers investigate the use of VR environments for local, small-scale geological analysis in lunar and planetary surface exploration missions. Users explored real environments from Svalbard, Norway, represented in both VR and desktop applications. The researchers’ comparative study found that the VR application provided users with a better sense of scale, improved ability to contextualize geological features, and reduced workload compared to the desktop application. However, it showed no significant improvement in the identification of high-level features. In the paper, the authors discuss what factors may have influenced these results and consider the implications for future VR development in geological exploration and astronaut training. Authors: Cody Paige (MIT Media Lab); Don D. Haddad (MIT Media Lab); Trent Piercy (MIT EECS); Jessica Todd (MIT AeroAstro and MIT-Woods Hole Oceanographic Institution Joint Program in Oceanography/Applied Ocean Science & Engineering); Ferrous Ward (MIT AeroAstro); Ariel Ekblaw (Aurelia Institute); Director Dava Newman (MIT Media Lab)

The Cambridge Science Festival starts today! The Media Lab is thrilled to be part of this annual celebration of creativity, innovation, art, science, and technology. Explore the panel discussions, workshops, and other activities featuring Media Lab researchers—all of these events are free, but many require registration: https://lnkd.in/gVb5cRGT

LLMs?make inconsistent decisions about calling the police when analyzing home surveillance videos, a study shows. The “move-fast, break-things” practice of widely deploying generative AI models “deserves much more thought since it could be quite harmful,”?Ashia Wilson?says. https://lnkd.in/epDxTTby

Curious about applying to the Media Lab? Register for the Virtual Open House for Prospective Students on September 30! You’ll have opportunities to hear from staff, faculty, and students; learn about the academic program and the research groups; and ask questions. https://lnkd.in/gMhfgptg

In a new study, researchers from MIT and Penn State University, including Media Lab alum Dana Calacci, find that large language models make inconsistent decisions about whether to call the police when analyzing surveillance video. “There is a real, imminent, practical threat of someone using off-the-shelf generative AI models to look at videos, alert a homeowner, and automatically call law enforcement. We wanted to understand how risky that was,” Dr. Calacci says. The researchers found that models often disagreed with one another over whether to call the police for the same video, and some were less likely to flag videos for police intervention in neighborhoods where most residents are white. “This is surprising,” they write, “because neighborhood characteristics are not provided in text prompts and not explicit in the video content.” The study grew out of a dataset that Dr. Calacci built in 2020, which contains thousands of Amazon Ring home surveillance videos. These videos were posted by Ring users to the public-facing social media site Ring Neighbors, a platform connected with the Ring system. It will be presented at the Association for the Advancement of Artificial Intelligence (AAAI) Conference on AI, Ethics, and Society. Authors: Shomik J. (MIT Institute for Data, Systems, and Society (IDSS)); Dana Calacci (Penn State University); Ashia Wilson (MIT EECS; MIT Schwarzman College of Computing)