Transistors, the building blocks of modern electronics, are typically made of silicon. Because it’s a semiconductor, this material can control the flow of electricity in a circuit. But silicon has fundamental physical limits that restrict how compact and energy-efficient a transistor can be.

MIT researchers have now replaced silicon with a magnetic semiconductor, creating a magnetic transistor that could enable smaller, faster, and more energy-efficient circuits. The material’s magnetism strongly influences its electronic behavior, leading to more efficient control of the flow of electricity. 

The team used a novel magnetic material and an optimization process that reduces the material’s defects, which boosts the transistor’s performance.

The material’s unique magnetic properties also allow for transistors with built-in memory, which would simplify circuit design and unlock new applications for high-performance electronics.

“People have known about magnets for thousands of years, but there are very limited ways to incorporate magnetism into electronics. We have shown a new way to efficiently utilize magnetism that opens up a lot of possibilities for future applications and research,” says Chung-Tao Chou, an MIT graduate student in the departments of Electrical Engineering and Computer Science (EECS) and Physics, and co-lead author of a paper on this advance.

Chou is joined on the paper by co-lead author Eugene Park, a graduate student in the Department of Materials Science and Engineering (DMSE); Julian Klein, a DMSE research scientist; Josep Ingla-Aynes, a postdoc in the MIT Plasma Science and Fusion Center; Jagadeesh S. Moodera, a senior research scientist in the Department of Physics; and senior authors Frances Ross, TDK Professor in DMSE; and Luqiao Liu, an associate professor in EECS, and a member of the Research Laboratory of Electronics; as well as others at the University of Chemistry and Technology in Prague. The paper appears today in Physical Review Letters.

Overcoming the limits

In an electronic device, silicon semiconductor transistors act like tiny light switches that turn a circuit on and off, or amplify weak signals in a communication system. They do this using a small input voltage.

But a fundamental physical limit of silicon semiconductors prevents a transistor from operating below a certain voltage, which hinders its energy efficiency.

To make more efficient electronics, researchers have spent decades working toward magnetic transistors that utilize electron spin to control the flow of electricity. Electron spin is a fundamental property that enables electrons to behave like tiny magnets.

So far, scientists have mostly been limited to using certain magnetic materials. These lack the favorable electronic properties of semiconductors, constraining device performance.

“In this work, we combine magnetism and semiconductor physics to realize useful spintronic devices,” Liu says.

The researchers replace the silicon in the surface layer of a transistor with chromium sulfur bromide, a two-dimensional material that acts as a magnetic semiconductor.

Due to the material’s structure, researchers can switch between two magnetic states very cleanly. This makes it ideal for use in a transistor that smoothly switches between “on” and “off.”

“One of the biggest challenges we faced was finding the right material. We tried many other materials that didn’t work,” Chou says.

They discovered that changing these magnetic states modifies the material’s electronic properties, enabling low-energy operation. And unlike many other 2D materials, chromium sulfur bromide remains stable in air.

To make a transistor, the researchers pattern electrodes onto a silicon substrate, then carefully align and transfer the 2D material on top. They use tape to pick up a tiny piece of material, only a few tens of nanometers thick, and place it onto the substrate.

“A lot of researchers will use solvents or glue to do the transfer, but transistors require a very clean surface. We eliminate all those risks by simplifying this step,” Chou says.

Leveraging magnetism

This lack of contamination enables their device to outperform existing magnetic transistors. Most others can only create a weak magnetic effect, changing the flow of current by a few percent or less. Their new transistor can switch or amplify the electric current by a factor of 10.

They use an external magnetic field to change the magnetic state of the material, switching the transistor using significantly less energy than would usually be required.

The material also allows them to control the magnetic states with electric current. This is important because engineers cannot apply magnetic fields to individual transistors in an electronic device. They need to control each one electrically.

The material’s magnetic properties could also enable transistors with built-in memory, simplifying the design of logic or memory circuits.

A typical memory device has a magnetic cell to store information and a transistor to read it out. Their method can combine both into one magnetic transistor.

“Now, not only are transistors turning on and off, they are also remembering information. And because we can switch the transistor with greater magnitude, the signal is much stronger so we can read out the information faster, and in a much more reliable way,” Liu says.

Building on this demonstration, the researchers plan to further study the use of electrical current to control the device. They are also working to make their method scalable so they can fabricate arrays of transistors.

This research was supported, in part, by the Semiconductor Research Corporation, the U.S. Defense Advanced Research Projects Agency (DARPA), the U.S. National Science Foundation (NSF), the U.S. Department of Energy, the U.S. Army Research Office, and the Czech Ministry of Education, Youth, and Sports. The work was partially carried out at the MIT.nano facilities.

MIT was founded on the idea that people learn best by doing. For nearly 25 years, the Lemelson-MIT Program in the MIT School of Engineering has carried that idea into K–12 classrooms across the country, helping teachers guide students as they find real problems and invent solutions to them. 

Now the program is expanding that work into connected pathways that follow students from the early grades through college.

This month, Lemelson-MIT awarded multiyear Invention Education Pathways grants and stipends to 19 schools, districts, and colleges across three regions: Northern California, Tennessee, and Massachusetts/Connecticut. The partners include elementary, middle, and high schools, as well as a community college and two state universities, together forming routes for students to experience invention at every stage of their education.

The initiative arrives as MIT President Sally Kornbluth has launched MIT for America, a strategic effort to widen access to quality STEM education nationwide, from kindergarten to community college. One of its three focus areas, Design, Make and Innovate, aims to give students more chances to build, test, and prototype their way toward solutions, and to support teachers in bringing hands-on work into their classrooms. Invention education does exactly that.

Learning the way inventors work

In invention education, students work in teams to identify problems that matter to them and their communities, then develop prototypes of original solutions. Along the way, they apply knowledge from multiple disciplines and develop both technical skills and durable human skills such as empathy, collaboration, and communication. They also gain confidence in their own abilities.

“Understanding user needs is critical in product design,” says Michael Cima, faculty director of the Lemelson-MIT Program and a professor of materials science and engineering. “It’s the foundation everything else is built on. Invention education builds that understanding early, teaching students to ask whom they’re designing for and what is truly needed before they ever build a prototype.”

Rather than offering a one-time project or an after-school club, the pathways approach aims to give students repeated, connected invention experiences as they advance. Lemelson-MIT staff will work alongside educators in each region to assess, design, and implement learning experiences that help students apply scientific, technical, engineering, and mathematics (STEM) knowledge to real-world problems. Schools and colleges can adopt Lemelson-MIT curriculum resources selectively to jump-start or strengthen their offerings.

“We have learned that invention education can benefit all students, especially when students can experience learning through the invention process across multiple grade levels,” says Stephanie Couch, executive director of the Lemelson-MIT Program. “Our goal is to create opportunities for learning through invention education during the regular school day, beginning in the early grades and continuing into adulthood.”

Every student an inventor

The three regions were chosen through a nationwide search for schools and districts committed to building strong invention pathways from kindergarten through college and to increasing the number of inventors from diverse backgrounds.

That commitment rests on a simple observation from Lemelson-MIT’s work: Given the chance to invent, students of every background surprise themselves, and their teachers, with what they can do. The program’s experience, case studies, and published research show that invention education helps students, especially those underrepresented in STEM, see themselves as capable STEM learners and inventors. Students in the program’s flagship high school initiative, InvenTeams, have earned 24 U.S. patents.

The benefits extend into the workforce. “Employers benefit from graduates that have the types of capabilities they need, but have trouble finding,” Couch says. “Invention education connects students’ knowledge and passions with the technical and durable human skills employers are looking for.”

Participating institutions

• Northern California: California State University at Stanislaus; Columbia College; Columbia Union School District; Denair Unified School District; Folsom Cordova Unified School District; Stockton Unified School District; and Visions in Education University Prep Early College Academy
• Tennessee: Apison Elementary School; Battle Academy; Hixson Middle School; Normal Park Museum Magnet School; and Red Bank elementary, middle, and high schools
• Massachusetts and Connecticut: Berlin High School; Cambridge Rindge and Latin School; New Covenant School; Watertown High School; and Westfield State University

Looking ahead

Over the next three years, the Lemelson-MIT Program will work closely with educators in all three regions toward a shared goal: making invention education a regular part of the school day, rather than an occasional enrichment activity. 

Lemelson-MIT hopes that what these partners learn together will help show how invention pathways can take root in schools and colleges across the country.

Vera Grbic, communications officer at the Institute Discrimination and Harassment Response Office (IDHR), is an expert at creating clear, informative collateral. Whether she is improving the IDHR website to help people more easily find the support or resources they need or using newsletters and annual reports to explain IDHR's work and the policies that guide it, Grbic believes it is vital for members of the MIT community to know where to turn and what resources are available to them. 

Grbic earned a BA in communications, with a minor in English literature, from the University of Ottawa, and an MA in communications from Carleton University. Both institutions are in Canada's capital city of Ottawa.  

Before joining MIT in 2021, Grbic worked in a mix of federal government, nonprofit, and research organizations. After graduate school, she worked as a freelance journalist, during which she developed a skill she still relies on today: collaborating with subject-matter experts to ensure that creative communications are grounded in fact. As a senior communications officer for the National Research Council of Canada, she worked with engineers and scientists in areas such as artificial intelligence, climate research, aerospace, and physiology. She elevated their research and innovations to key audiences and helped create new partnership opportunities. In this role, she translated complex information into engaging communications that furthered the organization's goals. Grbic describes this work as the "through-line" connecting her previous experience to what she does today at MIT. 

IDHR addresses concerns related to discrimination, discriminatory harassment, and sexual misconduct by providing education, individualized supportive measures, information about available resources, and complaint resolution pathways that include both restorative and investigative processes. Grbic develops IDHR’s annual communications strategy and has spearheaded several key initiatives, including a website and annual report redesign and the launch of IDHR’s Data Dashboard. She collaborates with colleagues across MIT to gather content ideas for newsletters, data and stories for annual reports, and feedback on communications materials. She often partners with IDHR's prevention education and outreach team, which delivers training to the community. 

Grbic notes that being a communications professional means constantly learning, whether that entails researching a new topic, finding creative ways to tell a story, or meeting subject-matter experts who broaden her perspective. Accuracy is at the core of her approach. While she strives to find creative ways to explain policies and complex processes in ways that are accessible to everyone, collaboration and fact-checking are intrinsic to her work. Given the wide range of responsibilities within communications, Grbic is a jack-of-all-trades. Her experience includes creative communications, social media management, long-form storytelling, event production, report writing, video production, and website management.  

In addition to her communications responsibilities, Grbic co-organizes the annual Change-Maker Awards, which honor members of the MIT community who work diligently to prevent sexual misconduct and combat harassment. She takes great pride in helping lead an event that recognizes individuals who are making a meaningful difference in building a safer culture at MIT. Grbic describes the award winners as “true inspirations.” 

Grbic recognizes that nothing is done in a vacuum. She values the input of her team members and colleagues across MIT and is proud to incorporate their feedback into her content. Because no two projects are exactly alike, she enjoys the creativity and variety that her work offers. 

Soundbytes 

Q: What are your favorite projects that you have worked on? 

A: I was part of a team that helped develop and launch the most recent sexual misconduct surveys for students and employees. I also strategized a campaign to clarify IDHR services, which included whole-scale website improvement: rewriting many sections, adjusting navigation, improving visual design, and publishing a short video explaining our process. There was also a poster campaign about the services we offer the community. MIT is awash with important storytelling, so getting that messaging to people who need it is vital. These projects brought me pride because taking the time to understand a community's experiences can have a meaningful impact. 

Q: How would you describe the MIT community? 

A: Smart, in a well-rounded way. Everyone is an incredibly hard worker and thinks in deep, research- and experience-backed ways on how to solve complex projects.  

As an employee who has worked with students on Institute-wide initiatives, committees, and awards, I have seen how much they care about making the world a better place. The staff and faculty are gracious and bring a strong work ethic to everything they do in support of the MIT community. 

Q: Are you involved in any areas of the Institute that are not directly related to your job? 

A: I recently started participating in the MIT Women’s League, which I am excited to explore. When I was pregnant, I took prenatal classes at MIT, including a yoga class where I met people I am still in touch with years later. I also played in a volleyball league at the recreation facilities, which made me see the facilities from a different angle.  

I’ve attended many free conferences and events hosted by MIT clubs and schools. For example, a children’s clothing swap has been organized for years, which also provides an opportunity to connect with other families in the community. I also prioritize keeping up with current topics and have attended two AI conferences organized by the Institute.  

While these opportunities may feel secondary to the workplace, they are important. I’m glad I am in a space that helps feed my physical health and has opportunities for broader learning. 

Millions of people rely on the cloud every day, yet researchers working to secure and improve the cloud are often locked out of the technology it runs on. Jeffery Lim is trying to change that.

One such technology, single-root input/output virtualization (SR-IOV), lets major cloud providers like Amazon Web Services boost the performance of their systems. Cloud providers combine SR-IOV with custom hardware and software to deliver faster cloud services than traditional setups allow.

Lim is a computer engineer at MIT Lincoln Laboratory. While pursuing his master's degree, he noticed that academic research proposing smarter or safer cloud infrastructure did not account for SR-IOV. Very few open-source hardware platforms existed that would let researchers work directly with SR-IOV as they developed new software or models to improve the cloud.

So, Lim filled that gap with his master's thesis, building a hardware platform compatible with SR-IOV that captures key features of the infrastructure used by cloud providers. The result is a research and development platform that allows researchers to build, test, and evaluate realistic cloud designs.

"As a society, everyone uses the cloud now," Lim says. "I hope that people will use this platform to build systems more akin to how cloud providers operate, and, more broadly, that researchers everywhere can take advantage of SR-IOV for the performance benefits it offers."

Lim presented a paper at the IEEE High-Performance Extreme Computing Conference in September, and he plans to release this work open source. In recognition of the hardware's potential impact, the Boston University College of Engineering awarded Lim its 2026 Master's Societal Impact Award.

Winding path to cybersecurity

Growing up, Lim was fascinated with building computers. That fascination set him on a winding path to the Secure Resilient Systems and Technology Group at Lincoln Laboratory.

After studying electrical and computer engineering at the University of Colorado at Boulder, he spent several years in industry working on field-programmable gate arrays (FPGAs) — integrated circuits that can be repeatedly reprogrammed for different functions. This background led to his 2018 hire at the laboratory, where he developed FPGAs for small radars that detect the breathing of humans trapped in rubble and for large radars that track satellites and orbital debris from atop Millstone Hill. He next turned to investigating radiation-induced faults in integrated circuits aboard space platforms and ways to protect the chips.

This focus on safeguarding hardware turned his attention to cybersecurity, an area he had long found interest in but never fully explored. So, Lim transferred to his current group in 2022 to pursue this interest.

"I don't think there's another place in the world where I could say, 'Hey, I want to try something new,' and actually have that opportunity," Lim says. "Lincoln Laboratory does that really well, giving people the chance to transition into a different field of engineering or area of expertise."

To deepen his expertise, he applied and was accepted in 2024 to the Lincoln Scholars Program, which allows technical staff to pursue full-time graduate studies at the master's or doctoral level. Scholars remain employed and apply their academic work to national security needs.

As he pursued his master's degree at Boston University, Lim worked closely with his Lincoln Laboratory mentor Benjamin Nahill, who helped Lim keep his research tied to the laboratory's mission and provided a sounding board for ideas. "I'd call him my rubber ducky," Lim says, a nod to the classic software engineering practice of "rubber duck debugging" — reasoning things out through speaking aloud to a rubber duck.

"We talked about his thesis concept for quite a while, but the implementation suddenly became real toward the end," says Nahill. "He transformed what began as an idea for a research tool to an incredibly performant implementation. Having worked with Jeffery on projects in our group, I shouldn't have been surprised, but it was most impressive."

Missing piece for cloud research

When most people say "the cloud," they mean cloud services provided by companies such as Amazon, Microsoft, and Google. Virtualization is the invisible process that makes those services possible. Users uploading photos or streaming movies are interacting with virtual machines, running on servers in data centers.

In the past decade, SR-IOV has changed the way traditional virtualization works. Traditionally, a software component known as the hypervisor acts as a middleman between the virtual machines and the server's physical hardware. Its main role is to distribute computing resources as needed, and provide functions (such as networking) to the virtual machines.

In contrast, SR-IOV enables virtual machines to access a dedicated slice of the hardware device, reducing the hypervisor overhead. This capability is often found on specialized devices called Smart Network Interface Cards (SmartNICs). Major cloud platforms, such as Amazon Web Services’ Nitro System and Microsoft’s Azure Accelerated Networking, use SmartNICs with SR-IOV to provide high-speed networking to many virtual machines.

"This was the missing research capability that I set out to build: an open-source SmartNIC platform compatible with SR-IOV," Lim says. Because SmartNICs can be built on FPGAs, Lim had a familiar starting point. In fact, he found a couple of platforms that supported SR-IOV; however, they lacked features that would make them practical tools for researchers to utilize. He spent a couple of months building those required components.

The first component is a switching fabric. Often, cloud customers' applications or programs may be split across multiple virtual machines, which may be distributed across multiple servers. Therefore, the cloud system needs a way to determine whether a piece of data needs to stay local or leave the server. Lim's switching fabric handles that traffic-directing job.

The second component is a network function engine. This small, specialized processor is built to run security rules, like firewalls, that keep one customer's data safely separated from another's. To build it, Lim used Extended Berkeley Packet Filter, a system that allows custom code to inspect network traffic, approving or blocking data as needed. Rather than starting from scratch, he adapted the design from doctoral research conducted by Zaid Tahir, who studied under their shared advisor, Professor Martin Herbordt of Boston University.

Herbordt described Lim's efforts as "combining vision with heroic system-building, single-handedly putting together the hardware and software necessary to prove the concept."

"I'm very much a hands-on hardware guy, so, yes, I ended up building two servers at home to test my designs," Lim says. He also used the Mass Open Cloud’s Open Cloud Testbed, a test bed meant for testing cloud platforms, to gather metrics and validate his system.

Impact beyond the platform

Throughout the project, Lim maintained focus on applicability and transparency. "My biggest worry was, is this going to be relevant? Is this something that can actually be utilized?" Lim reflects.  

Earning the Societal Impact Award points to yes. The award recognizes projects that address pressing global challenges through innovative applications of technology. Lim's platform opens the door to evaluating a cloud trust model that remains largely opaque, even as many sectors of society, from health care to banking, increasingly rely on it.

"Right now, we have limited visibility into how major cloud providers implement and evolve their infrastructure," Lim says. "If researchers can catch up to how the modern cloud operates, they might develop new approaches that improve security or performance."

As part of his role as a Lincoln Scholar, Lim is also exploring how his work connects to Lincoln Laboratory's national security mission. He sees several potential avenues.

One program in his group is evaluating cyber-hardening techniques for hypervisors, as the cloud takes on an expanding role in defense systems. "Introducing Lim's SR-IOV-compatible platform to this work would help bring our research closer to how cloud environments operate, demonstrating the impact of our techniques in realistic environments," says Nathan Burow, the program's principal investigator. Another program is examining what it takes to stand up high-performance data centers, work that could draw directly on Lim's research into the technologies SR-IOV requires.

Or his path may lead to something new. The laboratory allocates internal technical investment funding — supported by the Office of the Under Secretary of War for Research and Development — for staff-proposed projects addressing emerging needs. Lim is considering a proposal to bring SR-IOV to the embedded-systems level, say, to drones or satellites.

"We saw how cloud providers moved from traditional virtualization to SR-IOV to maximize performance while maintaining isolation," he says. "I wonder if we can do the same at the embedded level, where power and compute constraints are even tighter. I'd love to complete that transition. I learned a lot from the cloud. Now, can we apply those lessons somewhere else?"

Utilizing waste carbon at scale, assessing public trust in artificial intelligence and data centers, and building a quantum toolkit for critical applications were among the topics energy experts explored at the MIT Energy Initiative (MITEI)’s annual research conference on Sept. 22-23.

The complexity of the energy transition demands “a truly interdisciplinary and international approach,” said Paula Hammond, Institute Professor and dean of the MIT School of Engineering, kicking off the conference, which was themed “Transitioning energy at a pivotal moment.” 

More than 200 people attended Energizing@MIT: MITEI’s Annual Research Conference in person and online, including faculty, scientists, and researchers; representatives of global energy firms; nonprofit organizations; and academic institutions.

With the rapid proliferation of artificial intelligence and data centers, the challenge of the energy transition grows ever greater, requiring “researchers who can pursue fundamental questions about materials, chemistry, and energy systems … engineers who can design new technologies and determine how they might function at scale … and experts in economics, policy, human behavior, and sociology who can help us understand how those technologies will interact with communities and with the world around us,” said Hammond. 

From keynote speakers to panelists, seed grant innovators to startup entrepreneurs, the conference featured exemplars in all these categories.

Deploying today’s technologies

Eric Toone, chief technical officer and managing partner at Breakthrough Energy, set the tone for day one with a keynote address that was clear-eyed in his assessment. “If energy is prosperity,” he told the audience, “at least today, prosperity is emissions.” Yet, Toone was optimistic when he heralded examples of Breakthrough Energy-funded ventures, from iron-air batteries to vacuum-insulated glass windows, which can reduce the carbon costs of energy. 

“We can absolutely develop new technologies that provide the exact same service, but at zero carbon,” he said.

The conference made Toone’s claim tangible as it spanned the full arc of decarbonization — from deploying what we know today to uncovering what may be possible tomorrow. 

There are “projects that are actually being developed, companies that are putting steel in the ground now, electrons onto the grid to deliver capacity now, or in the very, very near term,” said John Parsons, deputy director for research at the MIT Center for Energy and Environmental Policy Research. Parsons moderated a panel on the revival of nuclear fission power. 

Highlights included Crane Clean Energy Center’s restoration of the Three Mile Island Unit 1 nuclear plant, set to restart in mid-2027; work by Électricité de France (EDF) to extend and modernize France’s 57-reactor nuclear fleet, adapting plants to climate impacts; and GE Vernova’s development of mature light-water reactor technology, with a new plant under construction in Ontario.

Carbon utilization offers another path to mitigating carbon dioxide (CO2) emissions. “The land and ocean are passively taking up almost 50 percent of anthropogenic CO2 emissions,” said Carolyn Ruppel, MITEI’s deputy director for science and technology, who moderated the panel. “The idea is to find uses for some of the rest of it.” The session showcased research exploring a host of technologies for making products with captured CO2.

Nicola Ferralis, a research scientist in the MIT Department of Materials Science and Engineering (DMSE), highlighted the importance of large-scale uses for carbon waste, such as structural carbon for vehicles and buildings. “We need to find applications that are actually able to match the supply by mass,” he said. 

Admir Masic, an associate professor in the MIT Department of Civil and Environmental Engineering and DMSE, described self-heating concrete panels used in real-world conditions in Sapporo, Japan, as well as new concrete materials that can store energy.

“The pace of the transition to a renewable-energy economy will strongly rely on the availability of scalable, bulk energy-storage solutions,” said Masic.

Kripa K. Varanasi, a professor in the MIT Department of Mechanical Engineering, is working on ways to make technologies such as emissions-plume capture at power plants economically viable. 

In a panel devoted to startups with links to MITEI or MITEI member companies, speakers described moving novel energy technologies from research to commercialization: Alfredo Rivela of Adaion has developed an AI tool for electrical utilities that optimizes their use of the grid; Eddie Obropta of Raptor Maps trains drones to inspect solar farms; Michele Guarnieri’s HiBot has created a robot that detects issues in power lines; and Michael Wang of Lithios is conducting field pilots that use electrochemistry to separate lithium from low-grade brines.

Successful ventures, the speakers agreed, begin with a clear customer problem that can be validated by industry users and that offers a path to scaled deployment. For industry users, “innovation is not a corporate virtue,” said Steve Hellman, managing partner at Alliance Partners. “It’s an existential requirement.”

Early-stage research and basic science

The conference also offered a bounty of promising early research on technologies and approaches for cutting carbon emissions. A panel on MITEI’s Seed Innovation Fund Program projects introduced MIT scientists who are separating and recycling rare earth elements; designing algorithms to decide which power plants should run to meet electricity demand; and modeling the impacts on human and natural systems of development and decarbonization.

A poster session revealed the next generation of energy researchers in such fields as chemistry, architecture, aeronautics and astronautics, and computer science. Graduate students are developing a wide array of technologies for carbon mitigation and new energy systems, such as a steam electrolysis process for hydrogen production; models of end-to-end decarbonization in buildings; and designs for hydrogen- and methane-fueled aircraft.

The conference highlighted the importance of fundamental science in feeding the pipeline of new energy technologies. Wednesday’s keynote speaker, Danna Freedman, the F. G. Keyes Professor of Chemistry at MIT and inaugural director of Quantum@MIT, traced scientists’ evolving capacity to harness quantum computing as a tool for solving a range of problems, including those in energy.

“What we want to do at Q-MIT is solve the hard problem, and the hard problem isn’t developing these ridiculously intricate and beautiful quantum tools,” said Freedman. “It is connecting the quantum tools with the applications that will change the world.”

A companion panel explored one such application: quantum sensing, which leverages fundamental physics to capture a nanoscale view of materials and their properties. For instance, Lukasz Zielinski, scientific advisor to the Integrated Sensing and Measurement Program at SLB, uses a lidar camera with single-photon detectors to identify methane plumes. Moderator Paola Cappellaro, the Ford Professor of Engineering in the MIT Department of Nuclear Science and Engineering, described her work using diamond-based sensors for detecting tiny magnetic fields from battery current, potentially helping measure battery conditions without disturbing the battery electrically. Such quantum sensors, Cappellaro said, can be “practical and impactful.” 

Data centers and public trust

One critical challenge facing energy researchers and industry emerged in sessions throughout the two days: the explosive growth of AI and the energy-hungry data centers that fuel this technology. As Deepjyoti Deka, a MITEI principal research scientist, put it, “the big test for AI right now is whether AI can reduce energy usage, improve efficiency, as well as bring in innovation faster than the rate at which it is going to consume electricity to get trained and deployed.” 

“New ‘hyperscale’ facilities built for artificial intelligence can use more power than entire cities,” said Steve Clemmer, director of energy research in the Climate and Energy Program at the Union of Concerned Scientists, sharing a recent analysis of data center proliferation. With new facilities being 10 to even 10,000 times larger than data centers built just a decade ago, he said, wholesale electricity costs will rise, and “increased reliance on fossil fuels will result in tens of billions of dollars in health damages, and trillions in climate damages.” He noted that decision-makers can mitigate these harmful impacts through clean energy development, strong ratepayer protections, and other safeguards.

Given mounting opposition to data centers, Kathryn Kline Tyndall, associate director of energy at the Bipartisan Policy Center, recommended that industry proponents provide early, accessible information about power, water, quality-of-life, land-use, environmental, and economic impacts, tailored to each community in which they hope to set up shop.

Larry Susskind, the Ford Professor of Urban and Environmental Planning in the MIT Department of Urban Studies and Planning, recommended that developers go further, and demonstrate credible benefits to communities. To achieve this, Susskind suggests a facilitated negotiation process between developers and local groups whose object is a community benefits agreement — a deal that might involve altering the data center design, and providing workforce development, infrastructure investment, and clean energy partnerships.

Ultimately, developers must accept that they can’t rush community acceptance; they have to earn it by giving residents a meaningful role in shaping projects and sharing their benefits. 

“You’re going to have to proceed at the speed of trust,” said Susskind.

Creating better-performing materials comes down to having more control over their structure at the smallest possible scale. One particularly effective way to accomplish that is using nanoparticles, which have highly customizable shapes and sizes and have already shown promise in applications across energy, medicine, and electronics.

But the process of turning nanoparticles into useful new materials has been hampered by the difficulty of controlling the number and direction of the links between each particle and its neighbors. Researchers have tried changing the shape of particles or precisely aligning the rigid connector chains between them, but both of those methods limit the material structures nanoparticles are able to form and are difficult to scale.

A team including researchers from MIT is taking a different approach. The researchers developed a tunable way to use super-flexible connector molecules that regulate the packing of nanoparticle networks by intentionally deforming based on their surrounding environment. In a paper published today in Nature Materials, the researchers showed how connector flexibility and electric charges can be used to tune the network’s structure, and they used the technique to create new kinds of nanoparticle-based materials.

The approach could offer a more versatile way to create high-performing materials for photonics, chemical catalysis, medicine, agriculture, and more.

“People are interested in nanoparticle superlattices because of the promise of building designer materials where you have more say in what types of structures they can actually build into,” says Rob Macfarlane, the Richard P. Simmons (1953) Professor in Metallurgy in MIT’s Department of Materials Science and Engineering. “This is a new approach in nanoparticle assembly that hasn’t been investigated before. It’s exciting to have a new tool that is so counterintuitive. It’s the exact opposite of what the dogma has been for 20 years: that you need to make connector ligands rigid, and any deviation causes assembly to fail. We showed if you just crank it all the way to the opposite extreme, it works just as well.”

Joining Macfarlane on the paper are MIT PhD candidate Nicholas Sbalbi; University of Delaware doctoral student Stephen Kronenberger; MIT undergraduate Hana Shinzawa; Matthew Ye PhD ’26; and former University of Delaware professor Arthi Jayaraman.

More flexible structures

A foundational truth in materials science is that a material’s structure governs its function.

“That means if you know what your building blocks are and how they’re arranged, you can explain what properties materials have,” Macfarlane says. “Controlling that is how you make materials that have new properties.”

Nanoparticles give scientists a flexible building block to make materials with desired properties, and individual nanoparticles are already used in applications as diverse as drug delivery, energy storage, computing, and sensing. But assembling them into larger structures has proven to be a challenge because of how tightly they tend to pack together.

To overcome that challenge, researchers make nanoparticles different shapes, which changes how they connect. They can also make only specific parts of the particles’ surfaces sticky so they form different patterns.

“The challenge with both of those approaches is they’re difficult to do synthetically as we make the particles,” Macfarlane explains. “The methods for defining stickiness that have been developed are always bespoke: They work for one material, at one size scale, in one condition. It’s good for studying, but you can’t do much with it.”

Both methods also require connector molecules, called ligands, at the surface of the nanoparticles to be rigid so they stay at the specific angles needed to maintain the formation researchers want, which further limits their functionality.

For their new study, the researchers wanted to turn that approach on its head.

“We thought, ‘Let’s do the exact opposite: make the bonds so flexible it doesn’t matter what direction they’re initially pointing.’ The ligands can then reorganize as the particles come together, and that reorganization itself limits how many neighbors each particle can bind,” Macfarlane says.

Macfarlane’s group has spent years creating ways to make different kinds of particles from grafts of polymers. The nanoparticles in their new system have flexible ligands with electrically charged tips, some positively charged and others negatively charged. The opposites attract, so the ligands bend to connect with opposite charges.

“The idea is, instead of defining a specific direction the bonds are going to occur in with complicated synthesis methods, we just said, ‘The bonds are going to be as flexible as they want to be, and they’ll spontaneously go to whatever valency is going to maximize bond stability,’” Macfarlane says.

In experiments led by Sbalbi, the researchers showed they could control the connections by changing the nanoparticle sizes and ligand chain lengths. They also used the technique to assemble the nanoparticles into previously unseen crystal structures.

“Without having to predesign where the particles connect or what their shapes are, we can control the number of bonds each particle has,” Macfarlane explains. “The amount of flexibility of the polymer dictates how many bonds it has. It allows us to lower the number of bonds each particle has and make lattices we have never seen before.”

A new path toward customizable materials

Macfarlane believes the system is a step toward getting custom nanoparticle-based materials out of the research lab and into the world.

“We in the nanoparticle assembly community talk about having completely independent control over what your particle is and what assembly it goes into,” Macfarlane says. “In theory, that should be the case. In reality, it absolutely has not been.”

More broadly, the researchers say their approach establishes ligand flexibility as a new path to making material structures that can be customized at scale.

“Nanoparticle superlattices offer a way to create designer materials where your building blocks have all sort sorts of unique properties,” Macfarlane says. “But just densely packing these nanoparticles together limits your applications. This offers a way to get less dense formations of nanoparticles, and that’s where a lot of the interesting applications are.”

The work was supported, in part, by the National Science Foundation, the Department of Energy, the U.S. Army and the U.S. Navy.

Regina Barzilay, MIT’s School of Engineering Distinguished Professor of AI and Health, and James Collins, MIT’s Termeer Professor of Medical Engineering and Science, are the recipients of the 2026 Pritzker Prize for AI in Science Research Excellence. 

The award recognizes their joint work in “pioneering AI-enabled antibiotic discovery, integrating deep learning, bioengineering, and experimental validation to identify novel antibacterial candidates against drug-resistant pathogens.” 

Established in 2024 by the University of Chicago in partnership with Caltech, the Pritzker Prize for AI in Science Research Excellence recognizes contributions that jointly advance AI and the natural sciences or engineering.

Barzilay and Collins, who are faculty co-leads of the Abdul Latif Jameel Clinic for Machine Learning in Health (Jameel Clinic) and affiliates of the MIT Schwarzman College of Computing and the MIT School of Engineering, published their first breakthrough paper together in 2020 with then-MIT postdoc and first author Jonathan Stokes (who worked in the Collins lab from 2017 to 2021) in Cell. The paper described the first artificial intelligence-discovered antibiotic, called halicin, named for the fictional AI system HAL 9000 in Stanley Kubrick’s 1968 film 2001: A Space Odyssey.

A broad-spectrum antibiotic that was originally studied as a potential diabetes drug, halicin was identified by a deep learning model trained on 2,500 molecules. Upon testing, the research team found that halicin was able to kill antibiotic-resistant bacteria (i.e., “superbugs”) such as Clostridium difficile, Acinetobacter baumannii, and Mycobacterium tuberculosis. A discovery process that once took years was completed in just a few weeks. Amidst the growing alarm around the global antimicrobial resistance crisis, the results demonstrated that AI could prove to be an effective tool for rapidly identifying new and effective antibiotics.

Following halicin, Barzilay and Collins worked with Stokes (now an assistant professor at McMaster University) to use machine learning to identify abaucin in 2023, a narrow-spectrum antibiotic demonstrating effectiveness against A. baumannii, an antibiotic-resistant bacteria commonly found on hospital doorknobs and in the wounds of U.S. soldiers stationed in Iraq and Afghanistan. Notably, abaucin’s “narrow spectrum” efficacy against a single species of treatment-resistant bacteria is its main advantage, minimizing the likelihood of bacteria developing resistance and harming beneficial bacteria in the gut.

“Historically, antibiotic discovery has depended on laborious, trial-and-error strategies,” says Collins. “This process is slow, costly, and unpredictable, often yielding few viable antibiotic candidates. With generative AI, we're now seeing a shift from AI being used as a tool for discovery to AI being used as a design tool. This shift opens new frontiers in antibiotic discovery that could bring us the next generation of lifesaving drugs.”

In Barzilay’s view, “despite significant progress in molecular modeling, we are still in early stages of drug discovery. With AI, we can only automate parts of the pipeline, rather than totally transforming the discovery process. To make it happen, we need to work really hard to build realistic AI models of human biology.”

Fish that are cultivated in pools or ponds are an increasingly important source of food around the world. Unfortunately, disease outbreaks can cause substantial losses in aquaculture. 

Fish farmers use vaccines to prevent disease outbreaks just like regular farmers, but vaccinating a fish is a lot harder than vaccinating a cow. The most effective way to vaccinate fish today uses needles, but there are several downsides to this process, both for the fish and the farmers. 

Now a team of researchers at MIT and the U.S. Department of Agriculture has developed a potentially better way to deliver vaccines. The approach uses water-resistant, food-grade microneedles that can lightly penetrate and remain attached to fish skin before decomposing. In a new paper published in ACS Nano, the researchers described their system and showed it could be used to deliver a range of cargos, such as bacterial vaccines.

“To feed the world we need more fish, which are a major source of protein and nutrition,” says Benedetto Marelli, a professor in the Department of Civil and Environmental Engineering. “Aquaculture is a very effective way of producing a large amount of fish. Our approach could give farmers a more precise tool to maintain fish health. Farmers are really concerned about the health care of these fish, so providing them with better tools is really important.”

Joining Marelli on the paper are Meng Li, an MIT postdoc and first author on the paper; Jad Raad MEng ’24; Yunteng Cao PhD ’22, a postdoc at Yale University; Yue Hu, a former MIT postdoc; Colleen Wolfe MEng ’23; Doyoon Kim, a former MIT postdoc; Bradley Frank, a current MIT postdoc; and Gregory Wiens, a researcher at the U.S. Department of Agriculture’s National Center for Cool and Cold Water Aquaculture.

Better vaccine delivery

Aquaculture holds enormous potential to address food insecurity, reduce energy consumption, and decarbonize global food systems. In 2021, for the first time, farmed aquatic animals exceeded wild-caught aquatic animals in global food production by weight.

“The world’s population is going to grow by almost 2 billion people by 2050,” Marelli says. “Fish provide access to essential nutrition. They are high in protein and healthy fat, which are difficult to make, so the world is trying to produce more fish. But there are many challenges associated with fisheries.”

To prevent losses to disease, farmers use a range of automated and manual vaccination strategies, each with their own drawbacks. Some treatments can go into fish food, but those might not reach every fish. Another method is putting the treatment directly in a pool and immersing the fish in the treated water. That approach requires far higher quantities of treatment and, again, isn’t 100 percent effective.

The most effective way to vaccinate fish is with metal needles. The process requires farmers to sedate the fish, then get them out of the water and onto a conveyor belt, where workers hurry to vaccinate the fish and get them back in the water to minimize harm.

“Workers inject fish one by one,” Li explains. “It’s high intensity, there’s a risk of cross contamination of blood-borne diseases, and workers may accidentally jab themselves because they are working so quickly. So it’s labor intensive, costly, stressful for fish, and dangerous for workers.”

Marelli’s lab has been developing silk microneedles to deliver nutrients to plants for years. The silk has been classified as safe for human consumption in food coatings and can degrade in the body or in the environment over time.

A few years ago, Marelli learned about the challenges with vaccinating fish at a conference in Southeast Asia, where aquaculture is an especially important part of the aquatic food supply. His lab received support from the Abdul Latif Jameel Water and Food Systems Lab (J-WAFS) to explore adapting the silk platform he’d developed for underwater use for fish. But the early silk designs would quickly dissolve when exposed to water, releasing the vaccine immediately and failing to penetrate fish skin.

Li, a biomedical engineer, started to work on the project in Marelli’s lab beginning in 2022. In 2024, she received a fellowship from the USDA, which provided funding as well as important connections with collaborators at the agency and beyond. For example, the U.S. Geological Survey gave the researchers access to a DNA vaccine for rainbow trout.

Marelli credits Li with finding a way around silk’s tendency to soften when exposed to water. The system described in the new paper uses a thin layer of shellac, an insect-secreted resin used in food coating and pills, to make silk microneedles water-resistant by providing a moisture barrier that remains intact before fish injection.

“The material choice is critical because we’re dealing with food, even though we expect the microneedle to fall out of the fish or to be degraded in their body,” Li says. “Our criteria were safety for fish and humans alike.”

The researchers showed the microneedles can store a range of vaccines, and they demonstrated different versions of the microneedles that lasted longer in the water or diffused vaccines more quickly.

“In some cases, you may want the vaccine released within 15 minutes, but in others, the release can last 10 or even 24 hours,” Li says. “That could be better for other therapeutic uses beyond vaccines.”

The microneedles were also able to preserve half of the DNA plasmids’ structure when stored for four weeks in dry conditions.

“That stability is important, particularly for the last mile of delivery, because aquaculture is mostly done in rural areas,” Marelli says. “This allows farmers to not need to rely on the infrastructure to maintain the cold chain.”

Protecting fish

Just like terrestrial animal vaccines, fish vaccines are most effective when administered early in life before disease exposure. That means the fish that farmers want to inject are small, further complicating vaccination. The researchers believe their approach could offer a better way to vaccinate small fish, and they tested it in rainbow trout.

“In trout aquaculture, one of the biggest challenges we face is preventing disease early in a fish’s life, because traditional needle‑based vaccination just doesn’t work well until the fish reach about 30 grams, which leaves the smallest juveniles unprotected during a critical growth period,” says Wiens. “What our research shows is food‑grade silk fibroin microneedles can deliver vaccines underwater to fish as small as 1 to 2 grams. We designed these single‑use microneedles to pierce wet skin, stay attached while the fish swim, and release either bacterin or DNA vaccines in a controlled way.”

The researchers say there is still work to be done before their device can be seamlessly used at scale in fish farms. But they believe this paper is an important first step.

“We have all the components to believe this technology could be developed further,” Marelli says. “There are so many challenges with aquaculture and fish vaccination right now. We need to support farmers to operate safely, improve working conditions, protect their business, make vaccination more accessible, and improve the living conditions of farmed fish and ultimately the quality of the food we eat. This new technology could be a much more effective and safer approach.”

The work was supported, in part, by MIT’s Abdul Latif Jameel Water and Food Systems Lab and the U.S. Department of Agriculture.

Five MIT faculty members and two additional alumni were recently named to the 2026 cohort of AI2050 Fellows. The honor is announced annually by Schmidt Sciences, Eric and Wendy Schmidt’s philanthropic initiative that aims to accelerate scientific innovation. 

Conceived and co-chaired by Eric Schmidt and James Manyika, AI2050 is a philanthropic initiative aimed at helping to solve hard problems in artificial intelligence. Within their research, each fellow will contend with the central motivating question of AI2050: “It’s 2050. AI has turned out to be hugely beneficial to society. What happened? What are the most important problems we solved and the opportunities and possibilities we realized to ensure this outcome?”

AI2050 Senior Fellow

Sendhil Mullainathan is the Peter de Florez Professor at MIT with dual appointments in the departments of Economics and Electrical Engineering and Computer Science who is also a principal investigator at the MIT Laboratory for Information and Decision Systems. His current research is at the intersection of algorithms and people, with a central focus on launching The Bike Shop@MIT, a new research center that works toward more human-centered AI through scalable bicycles for the mind, and innovative new gears for machine learning algorithms. Mullainathan is the co-founder of ideas42, a nonprofit to apply behavioral science; MIT's Abdul Latif Jameel Poverty Action Lab, which promotes the use of randomized control trials in development; and Dandelion Health, which combines proprietary AI tools with real-world patient data. He's the co-author of "Scarcity: Why Having Too Little Means So Much."

Among other awards, he is a recipient of the MacArthur “Genius Grant”; has been designated a “Young Global Leader” by the World Economic Forum; was labeled a “Top 100 Thinker” by Foreign Policy Magazine; and was named to the “Smart List: 50 people who will change the world” by Wired Magazine (U.K.). He serves on the board of the MacArthur Foundation, is affiliated with the NBER and BREAD, and is a member of the American Academy of Arts and Sciences. His AI2050 project focuses on the translation of patterns detectable by AI into testable theories.

AI2050 Early Career Fellows

Alex Damian is an assistant professor in the Department of Mathematics, holding an MIT Schwarzman College of Computing shared position with the Department of Electrical Engineering and Computer Science, who is also a principal investigator at the MIT Laboratory for Information and Decision Systems. His research focuses on the mathematical foundations of deep learning, with an emphasis on optimization dynamics and representation learning. He was previously a research fellow at the Kempner Institute at Harvard University. He received his PhD in applied and computational mathematics from Princeton University and his BS in mathematics from Duke University. His AI2050 project focuses on effective AI model training.

Bailey Flanigan is a Theodore T. Miller (1922) Career Development Assistant Professor at MIT, holding an MIT Schwarzman College of Computing shared position with the departments of Electrical Engineering and Computer Science and Political Science, and is also a principal investigator at the MIT Laboratory for Information and Decision Systems. She received her PhD in computer science from Carnegie Mellon University in 2024. In 2024-25, she was a Harvard Data Science Initiative Fellow, hosted at the Harvard Ash Center for Democratic Innovation. Her research combines tools from across disciplines (including social choice theory, game theory, algorithms, statistics, and survey methods) to advance political methodology and strengthen public participation in democracy. She is specifically interested in sampling algorithms, opinion measurement/preference elicitation, and the design of democratic innovations like deliberative minipublics and participatory budgeting. Her AI2050 project focuses on gathering structured public input that lawmakers can use when making complex decisions.

Yunha Hwang is the Samuel A. Goldblitch Career Development Assistant Professor at MIT, holding an MIT Schwarzman College of Computing shared position with the departments of Electrical Engineering and Computer Science and Biology. She is also a principal investigator at the MIT Laboratory for Information and Decision Systems. Hwang serves as the co-founder and chief scientist at Tatta Bio, a scientific nonprofit dedicated to advancing genomic AI for biological discovery. She received her PhD in biology from Harvard University and BS in computer science from Stanford University. Her research interests span machine learning for microbial evolution, biomanufacturing, and open science. Her AI 2050 project focuses on determining biological enzyme functions. 

Manish Raghavan is the Drew Houston (2005) Career Development Professor at the MIT Sloan School of Management, holding an MIT Schwarzman College of Computing shared position with the Department of Electrical Engineering and Computer Science. He is also a principal investigator at the MIT Laboratory for Information and Decision Systems. Before that, he was a postdoc at the Harvard Center for Research on Computation and Society. Raghavan received his bachelor’s degree in electrical engineering and computer science from the University of California at Berkeley, and PhD from Cornell University. His research interests lie in the application of computational techniques to domains of social concern, including algorithmic fairness and behavioral economics, with a particular focus on the use of algorithmic tools in the hiring pipeline. His AI2050 project focuses on mitigating negative effects, and strengthening positive aspects, of widespread AI reliance. 

MIT alumni Irene Chen PhD '22 and Allison Z. Koenecke '14 as well as former postdoc Juan Carlos Perdomo were also named AI2050 Early Career Fellows.

The Applied Research Training in Evaluation, Methods, and Impact Studies (ARTEMIS) program, launched by J-PAL North America, equips early-career researchers with high-level technical training and practical skills to design randomized evaluations (also referred to as a randomized controlled trials, or RCTs) on issues of poverty in the U.S. context. Fully funded and highly rigorous, the program's inaugural cohort of 20 participants — including PhD students through pre-tenure faculty — is receiving training, mentorship, and proposal development support through December. 

ARTEMIS is the newest offering within J-PAL North America’s Economics Pathways program, which broadens and strengthens the pipeline of economics researchers by providing support to aspiring scholars at every stage — from undergraduate interns to research staff and, now, academic researchers preparing to launch their first randomized evaluations.

ARTEMIS programming commenced this past June with the Summer Institute — an intensive weeklong course at MIT Endicott House in Dedham, Massachusetts, that focused on delivering technical content and building participants' practical skills and confidence to design and pursue funding for rigorous evaluations to address poverty in the United States. During the Summer Institute, participants had the opportunity to learn from researchers in the J-PAL network, including Gerald Daniels, associate professor of economics at Howard University, who served as a lecturer for the program. “ARTEMIS is an amazing program that bridges the gap for researchers who haven't previously had the opportunity to work with randomized evaluations,” notes Daniels. “It takes rigorous methodologies, like randomized evaluations, and makes them approachable, helping scholars translate their research ideas into high-impact evaluation designs." 

The inaugural cohort brought together early-career researchers from institutions across the United States, each tackling a research question addressing an issue related to poverty that is well-suited for a randomized evaluation. Their interests spanned the breadth of J-PAL’s work, touching on education, climate, health and labor markets, racial equity, and social protection — reflecting the wide range of policy questions that rigorous, evidence-based research can help answer. 

“I really appreciated being a part of the ARTEMIS Institute this summer,” says Felipe Juan, an ARTEMIS inaugural cohort participant. “I initially thought the randomized evaluation methodology would be out of reach for me, but the summer training helped to break down the practical steps of conducting an RCT and identify creative ways to approach experimental designs.” 

Bolstered by the intensive Summer Institute training, ARTEMIS participants are now engaging in six months of continued study design and proposal development support, including proposal webinars with J-PAL grants staff and office hours with senior technical staff who will advise on study design and proposal challenges. ARTEMIS participants will also participate in a research design workshop, where participants present their early-stage RCT designs and receive feedback from the broader J-PAL research network. Finally, participants will receive introductions to researchers in the J-PAL network with shared research interests to forge connections that may lead to future collaboration and opportunities. Through this sustained mentorship model, ARTEMIS is designed to help participants bridge the gap between developing a promising research idea and launching a funded, policy-relevant randomized evaluation. 

For early-career researchers, this hands-on, sustained support has already proven to be transformative. “Participating in ARTEMIS Summer Institute was an incredibly valuable experience. The combination of lectures by experts in the design and implementation of randomized controlled trials, hands-on workshops, and thoughtful feedback helped me sharpen my research questions and think more carefully about research design, implementation, statistical power, and meaningful stakeholder engagement,” says Hoyoung Yoo, an ARTEMIS 2026 participant. “I also greatly appreciated the opportunity to learn from J-PAL faculty and staff and exchange ideas with an inspiring group of fellow researchers. I left the training with a clearer path toward transforming an early-stage idea into a feasible, rigorous, and policy-relevant evaluation.” 

For many, the program’s value also extends beyond technical training to the community and long-term relationships it builds. “J-PAL North America’s support has continued well beyond the Summer Institute, which has been especially valuable, and the connections I made have already started shaping how I am approaching the work,” notes Tessie Krishna, an ARTEMIS 2026 participant. “I left with a clearer proposal, a wider network, and a real sense of belonging in the J-PAL community. I am grateful to the J-PAL team for creating something so thoughtful and welcoming.” 

By building the technical capacity and confidence of the next generation of researchers, ARTEMIS is advancing J-PAL’s global mission to reduce poverty by ensuring that policy is informed by rigorous, scientific evidence. Every individual trained through ARTEMIS is joining a growing network of researchers working to answer the most critical questions of today. 

ARTEMIS is made possible by grants from Arnold Ventures, Russell Sage Foundation, and the Alfred P. Sloan Foundation. Visit the J-PAL website for more information about ARTEMIS and the Economics Pathways program.

Global warming, the observed increase in the Earth’s average global surface temperature since preindustrial (1850-1900) times, currently stands at around 1.63 degrees Celsius. While this number is expected to continue to rise, local temperatures will not go up uniformly around the world. In regions where warming trends are most pronounced, local adaptation and resilience measures will be especially critical to minimizing risks to ecosystems, agricultural productivity, the built environment, human health, and economic well-being. 

To that end, a new open-access study in the journal Scientific Data identifies the highest-risk regions for warming within the United States. Using data from well-known, high-resolution climate and weather datasets (MERRA2 and ERA5) covering the years 1980-2022, researchers at the MIT Center for Sustainability Science and Strategy (CS3) report global warming-attributed trends in monthly-averaged maximum daily temperature for every U.S. state. 

The study finds that Alaska, Utah, New Mexico, Colorado, and Nevada have experienced the strongest average warming trends; the lowest occurred in Hawaii, Montana, North Dakota, South Dakota, and Arkansas. Across the country, warming trends were most pronounced during the summer and fall, particularly in September and November. Alaska has the highest historical maximum daily temperature warming (an average of 2.5 C of monthly-averaged maximum daily temperature increase per degree of global warming), while Hawaii has the lowest average historical warming, at 0.8 C. 

The authors expect that barring unforeseen local anomalies, these state-by-state temperature trends will continue unabated as ongoing global emissions from human activities continue to raise the average global temperature.

“Based on these results, we encourage decision-makers to assess states that have undergone the steepest increases in maximum and minimum daily temperatures, and identify where continued trends would indicate a rapid progression toward thresholds beyond which natural and managed resources, ecosystems, and infrastructure can function safely,” says MIT CS3 Deputy Director C. Adam Schlosser, a co-author of the study. “In such cases, triaging and adaptive measures would take a high priority. By directing resources to where they are needed most, state and regional planners could significantly reduce climate-related physical and financial risks.”

Unlike previous efforts to report temperature trends across the United States, this study aggregates average maximum, minimum, and mean daily temperature data at the state level for each month of the year, providing actionable information for planners in government, industry, and civil society.

People are peppering artificial intelligence chatbots everyday with questions, but when it comes to politics and the looming 2026 U.S. midterm elections, what role are AI platforms playing? Could an AI chatbot sway a voter to switch parties or re-examine their political beliefs? 

MIT researchers have built a new dashboard, called the LLM Election Observatory, to examine, in real-time, how popular AI chatbots are responding to questions about the 2026 midterm elections. The team behind the dashboard — led by Professor Adam Berinsky, Assistant Professor Chara Podimata, and Professor Charles Stewart III — has extensive experience studying voting and election administration, AI and misinformation, and how AI systems are impacting human decision making. In an effort to better understand how AI is shaping the election, they designed the dashboard to simulate how individuals across the country communicate with chatbots. 

The team tracks how different AI systems respond to the same fixed set of questions on a range of topics, including election certification, hot-button issues, result predictions, news coverage, and more. Every few days, the LLM Election Observatory team runs the questions through the chatbots again, monitoring how their responses vary as the election gets closer depending on user location, political party affiliation, identity and other factors. The goal of this work is not to merely understand how AI systems spread information, but also better comprehend how they are shaping political discourse and voter understanding of political issues across the country. 

MIT News chatted with Podimata, the Class of 1942 Career Development Professor, an assistant professor of operations research and statistics in the MIT Sloan School of Management, and a member of the Laboratory for Information and Decision Systems; and Berinsky, the Mitsui Professor and a professor in the Department of Political Science, about their work with the LLM Election Observatory, how AI could impact the midterms, and what voters should keep in mind when they interact with chatbots.

Q: What was the inspiration for this project and what are you hoping to learn through the LLM Election Observatory?

Podimata: People are increasingly using AI chatbots to find and make sense of political and election-related information. That makes it important to understand what information these systems actually provide, not just at one moment in time but throughout an election cycle. This is particularly challenging because the models themselves are continually getting updates, the political information inputs available to them are themselves changing, and their answers can depend on how a user asks a question or what information the user provides about themselves.

One basic problem is that we currently do not have a good historical record of this information environment. If an AI system gives a particular answer about an election today and gives a different answer several weeks later, it can be very difficult to reconstruct after the fact what voters would have seen at each point in time. One important goal of our project is simply to systematically document that environment as it develops and make the resulting data available to other researchers and to the public.

Ultimately, factual accuracy is only the first frontier. We absolutely need to understand whether chatbots provide accurate political information, but we also need to pay just as much attention to the subtler ways in which they may shape people’s understanding of political issues. Differences in framing, emphasis, uncertainty, and cited sources can matter enormously, even in the absence of outright misinformation.  

Q: Professor Podimata, you studied how LLMs shaped political discourse and information during the 2024 election. What were the main takeaways of that research and how did those findings inform this new project?

Podimata: The LLM Election Observatory builds on my previous work during the 2024 presidential election, which found evidence of meaningful variation in AI models’ responses over time and across prompt framings, as well as differences in how models characterized candidates and implicitly anticipated the election outcome. 

The 2026 midterms give us an opportunity to expand that work substantially. Midterm elections are particularly interesting because there are many state and local contests unfolding simultaneously, alongside broader national political narratives. Our new audit survey therefore asks both contest-specific questions about the U.S. House and Senate candidates, but also broader questions about voting information and other aspects of the election information environment.

Q: Why do you think it’s so important to be studying how chatbots are responding to political queries across different groups of users?

Podimata: We all build our own “microcosms” through our interactions with chatbots. And yet, we lack systematic ways of understanding how these individualized information environments are shaped by who we are or how they evolve over time as we and the information environment changes. 

For example, people may not realize that what their favorite chatbot tells them about a political issue may be quite different from what their friend’s favorite chatbot tells them about the exact same issue. And importantly, these differences matter even when the information being provided is factually correct. Chatbots may frame the same facts differently, emphasize different aspects of an issue, express different levels of uncertainty, or draw on entirely different sources, even when responding to identical queries between different users. As a result, two users may walk away with very different perceptions of the same underlying facts, without ever realizing that their experiences differed. 

In some ways, this brings back many of the concerns we (as researchers and society) have grappled with in the context of social media. On social media, users and researchers could, at least to some extent, observe the information circulating across different Facebook groups, Instagram accounts, or TikTok communities. With chatbots, these information environments are largely invisible to anyone outside the individual conversation. We cannot easily observe what different users are being told, let alone systematically compare those responses or track how they change over time. This is precisely why projects like the LLM Election Observatory are so important: By systematically collecting and publicly sharing chatbot responses, we can begin to make these otherwise invisible differences observable and open to scrutiny.

Q: What sorts of questions and AI responses are you tracking through the dashboard?

Podimata: There are several questions we ultimately want this dataset to help us study. For example: Do different models provide systematically different information for the same questions? Do their responses change over the course of the campaign? Does the answer a particular model gives change depending on how a user frames the question or describes themselves (e.g., does it change if the model gets informed that the user is a Democrat or a Republican)? And, finally, what are the sources that the models pull information from when answering election-related questions? 

Our public dashboard is an important part of the project. We want people both from and outside of academia to be able to see the data we are collecting as the election unfolds rather than only seeing the results months later in an academic paper. The dashboard therefore presents the underlying data and some preliminary descriptive analyses, and we plan to make the dataset available for other researchers to study as well. We expect the dashboard to evolve as the data and our methods evolve.

Through this dashboard, we wanted to build on the scientific methodology we developed in 2024 but also to make the research much more accessible and useful to the public while the election is actually unfolding. Not only that, but the addition of MIT political science experts in the team cannot be overstated; Adam and Charles have really helped expand the questions that we are asking in our audit.

Q: What do you wish voters across the country knew about AI chatbots when it comes to searching for election information?

Berinsky: People should be sure to remember that just because a chatbot sounds confident doesn’t mean it is accurate. AI systems are designed to answer the questions that people ask them in an authoritative voice even when they are wrong, incomplete, or leaning on a weak source. These answers are a reasonable place to start, but voters shouldn’t stop there.

Q: What are your findings thus far?

Podimata: We have not yet done the full, longitudinal analysis, so we are only going to talk about broad strokes here, but a few broad patterns seem to be visible in the 2026 data. Across the different user identities we present, we see differences in how the answers are presented, in which sources a model cites, and in tone — sometimes even when the underlying facts are the same. What we cannot yet say is how stable those differences are, how they move over the course of the election, or how a real person would read and react to them. Our future analyses will take this up. It also underscores why the continuous record of responses we are collecting matters more than any single snapshot that you might see in the days around the election.

Q: Professor Berinsky, your research has also shown that in some cases AI chatbots can be harnessed to help refute misinformation. Can you talk a bit about this?

Berinsky: In my undergraduate course on information and social media I have been talking for a while about how AI can cut both ways. On one hand, these tools can generate misleading claims in text, pictures, and video at a scale we have never seen. Basically, AI can help bad actors “flood the zone” with misinformation. On the other hand, some of the most promising recent work demonstrates that evidence-based dialogues with a chatbot can durably reduce belief in conspiracy theories and other beliefs that are not supported by the balance of publicly available evidence. 

AI has the potential to do such debunking at scale — something we have long struggled to do. The lesson from decades of misinformation research is that any technology (be it print, radio, or television in earlier eras) is not inherently good or bad for promoting truth. The devil is in the details in how it is built and deployed. Independent auditing like this project can help us assess the ultimate impact of such technologies.

Q: Can you talk a bit about your overall goal with this research? 

Podimata: AI is a transformative technology, and people understandably have very different expectations (and feelings) about what it will mean for society at large. All of us have people in our lives that are overly enthusiastic about AI, others that are deeply concerned, and many more that are somewhere in between trying to understand what this technology can and cannot do. Our goal here is not to begin from the assumption that AI is either good or bad for elections. Instead, we hope to shed some light onto one part of this transformation: what information major AI systems are actually giving people about an election as it unfolds.

A nanoscale particle with a unique structure, developed at MIT, holds potential as a customizable delivery vehicle for personalized cancer vaccines.

The backbone of the vaccine is a modified, harmless bacteriophage called M13, which forms a nanofiber that can be coated with cancer antigens. By precisely controlling the density of those cancer proteins, the researchers found that they could create a vaccine that prevents the development of colon tumors in mice.

This type of delivery particle could be adapted to carry many different antigens, allowing for the creation of personalized vaccines to treat individual patients’ tumors, the researchers say.

“The great thing about this is that it’s modular and we could switch out different antigens on demand, so you could design it to be patient-specific and target a known antigen for a particular kind of cancer,” says Angela Belcher, the James Mason Crafts Professor of Biological Engineering and Materials Science and Engineering and a member of the Koch Institute.

Belcher is the senior author of the study, which appears today in the journal Advanced Healthcare Materials. Shengnan Huang, research scientist at the Koch Institute, is the paper’s lead author. Jack Rudrum, Huaiyao Peng, and Hao Nghi Luu, graduate students at the Koch Institute, are also authors of the paper.

Viral vaccine assembly

For more than 20 years, Belcher’s lab has been deploying the M13 bacteriophage in a variety of applications. This virus, which is harmless to humans, consists of a single strand of DNA coated in a protein shell, or capsid. 

Belcher has previously developed techniques for generating materials by engineering these viruses to self-assemble into nanoscale scaffolds. By coating these scaffolds with inorganic materials, they can create functional devices such as batteries, semiconductors, solar cells, and catalysts, with precise control over the electronic, optical, and magnetic properties of the materials.

More recently, Belcher’s lab has been developing methods to control features of the phage that could allow it to be repurposed for medical applications, including vaccine delivery. 

“We’re taking this toolkit that we’ve worked on and developed over decades and formatting it to be specific for vaccine therapeutics,” Belcher says. “Some of the lessons learned from making batteries and all these other things are now being fine-tuned to make personalized vaccines.”

The capsid that surrounds the M13 phage forms a long filament made from viral proteins. To create vaccines, the researchers modify one of these proteins, known as pVIII, by attaching an antigen to the end of the protein. This can be done by adding the gene for the antigen to the virus’s DNA.

Some of the lab’s previous work had suggested that changing the density of cancer antigens attached to the capsid could affect the strength of the resulting immune response. In the new study, Huang wanted to find a way to control that density. 

“The antigen is the specific target we want to initiate the immune response against, which means we have to ensure the right dose of antigen to make sure that we can have a strong immune response,” she says.

Huang discovered that she could increase the density of the modified pVIII proteins by adding one or two negatively charged amino acids to the protein sequence. The density could be further fine-tuned by adding a genetic mutation that slows down production of the antigen protein.

In this study, the researchers also figured out how to boost the phage’s own ability to act as a vaccine adjuvant, by changing the genetic sequence of the virus’s DNA. Most vaccines need to be delivered along with an adjuvant such as aluminum hydroxide, which helps to stimulate a strong immune response. However, the M13 phage can act as an adjuvant itself because single-stranded DNA activates a pathway called TLR9, which is part of the body’s innate immune system.

“We can control the exact length. We can control the exact antigen expression, density, pattern, the exact adjuvant amount. A lot of times with a vaccine, you’ll have to put a different adjuvant with it, but now it’s all genetically encoded,” Belcher says. 

Targeting tumors

Using these techniques, the researchers showed that they could design phages that express an antigen called SIINFEKL at densities ranging from 6 percent to 70 percent. SIINFEKL, a small protein that can be engineered into tumors, is often used as a model to test vaccine effectiveness.

The researchers tested two of these vaccines in mice, one with 6 percent density and one with 35 percent, and found that the one with 35 percent was the most protective. 

After vaccination with the higher-density particles, mice were much less likely to develop tumors when later exposed to melanoma cancer cells expressing the SIINFEKL antigen. These mice also generated more T cells targeting SIINFEKL. This is, in part, because dendritic cells, which help to activate T cells, were more likely to take up vaccines with higher antigen density, the researchers found.

In another set of experiments, the researchers created and tested a vaccine carrying a neoantigen, which is only present in the tumor cells. This protein, known as ADPGK, is specifically expressed by colon tumors in a mouse model of cancer. When given the higher density vaccine, these mice were much less likely to develop colon tumors.

In this study, the researchers tested the vaccine as a way to prevent cancer from developing, and in their previous work, they adapted it for use to treat mouse colon tumors that have already formed. In the future, they also hope to use it to combat other types of cancer.

“This interesting work adds another cog to improve the translational potential of the M13 phage platform for cancer immunotherapeutics,” says Alberto Danielli, an associate professor of molecular biology at the University of Bologna, who was not involved in the study. “I look forward to seeing whether this platform is also effective against established tumors and whether additional tumor antigens can be displayed with comparable molecular and immunological efficacy.”

This platform could be especially well-suited for personalized cancer vaccines, because many different antigens could be swapped into the vaccine.

“Because this is a modularly engineered platform, we can change the antigen we are interested in. For colon cancer, we use this specific antigen. For other types of cancer, we can just plug in other antigens and they can be assembled in this platform,” Huang says.

The research was funded, in part, by the Koch Institute Frontier Research Program, the Marble Center for Cancer Nanomedicine, and the U.S. Defense Advanced Research Projects Agency.

You know it’s autumn when the scent of pumpkin spice latte begins to waft through the air at the Dunkin’ in the Stratton Student Center, and students in MIT’s Next House residence hall diligently begin to piece together the many parts needed to showcase Next Haunt — a combination of a Halloween haunted house and escape room, built in the basement of the hall.

Every fall since 2014, students — also known as hauntmasters — spend several weekends building the room and preparing puzzles, physical tasks, or word challenges to obtain a code to open locks that let participants escape the house. This year’s theme, titled “Where the Light Ends,” follows “Terror Under Destruction,” “Next Stop: MBTA,” and last year’s theme: “Get in the HOLE.”

Next Haunt is free and open to all MIT students, also known as “test subjects.” Participants typically spend around 20 minutes in small, dark, enclosed rooms, working with four to six teammates to find the clues they need to escape.

Along the way, they have to navigate zombies — played by student actors — who may shout at them or splash them with blood (fake) while they solve the puzzles. If a zombie touches a participant, they become “paralyzed” and can no longer move, but they can still communicate and relay information to their teammates.

Teams must work closely, strategize, delegate tasks, and communicate effectively as the clock counts down. The group with the quickest escape time gets bragging rights, and every group signs their names on a wooden board to show that they participated.

MIT junior Kiyoko Hayano, Next Haunt co-producer, says, “I became involved in Next Haunt during my first semester at MIT because I had experience with larger-scale painting. I split the build director’s role with two other freshmen, including Michelle Liang, our current assistant producer. I thought this would be a great opportunity to take a break from psets and work with my hands, and fell in love with the club.”

Safety first

Before the room is built and the puzzles and costumes are created, careful planning takes place to ensure the escape room is safe, fire-resistant, and passes an on-site inspection. MIT’s Campus Construction and Environment, Health, and Safety offices are involved from the start, and all students who partake in the build are required to attend safety orientation and tool training.

“We currently have 18 people on our executive team, and we have 12 outside volunteers come help over the first few weekends of Next Haunt build,” says Hayano. “We then run Next Haunt for six-and-a-half hours every night for two weekends at the end of October.”

Once the build is complete, several alumni volunteers conduct a “test run” to make sure the puzzles and clues are hard, but not impossible, and anyone who wants to leave the room before the allotted time can easily find the door out built with Velcro closures. 

Escaping by solving the puzzles is not easy, and not all participants are successful. Last year, only 16 percent of the 73 teams successfully got out of the room. The winning team escaped in just under 15 minutes.

Alumnus and previous Next Haunt producer Allen Lin ’26 notes, “I became involved with Next Haunt in my freshman year after being a test subject myself, and I loved the creative and building process so much I wanted to join. My favorite part was seeing all the teams go through the house, especially on the last day when we hosted past Haunt alumni. It was just as rewarding to see how much our test subjects enjoy the house as it was to share the many laughs and moments of camaraderie among my team. I’m excited to come back as an alum and see what the Next Haunt team built this year!”

Environmentally mindful

At the end of Next Haunt, the escape room is dismantled, and anything that can be repurposed safely is stored for the following year’s build. The executive crew then goes to a Boston-area escape room to celebrate their hard work.

“It’s a joy working collaboratively with students and connecting them to campus partners who lend their expertise as the build develops," says Milo Alto, manager of special projects, housing and residential services. “Students get very creative every year with the theme and costumes, but they also learn valuable leadership skills along the way — time management, delegation, working with peers, and implementation of a detailed plan. It’s also a great way to build community.”

Interested MIT students can sign up to help plan and work on the 2027 build.

MIT researchers have shown that they can accurately track muscle movement using tiny magnetic beads implanted in the muscle. These measurements could make it easier for people with certain types of amputations to control their prosthetic limbs.

For decades, researchers have been working to develop neuroprosthetic limbs, which use neural information from muscles remaining after an amputation to guide the prosthetic limb. Measurements of a muscle’s length, velocity and activation can be fed into the control algorithms that guide the limbs, allowing individuals with leg amputation to walk with a more natural gait, or those with arm amputation to manipulate objects more effectively.

Until now, these devices have relied on muscle activity measurements from electrodes placed on the surface of the skin. However, those measurements have not proven reliable enough to ensure safety while walking with a neuroprosthetic leg. 

The MIT team hopes that their new magnetic tracking system, which offers more accurate measurements, will enable commercial development of such limbs.

“For four decades, the standard methodology for neurally controlling prostheses has been to use skin electrodes to measure muscle activation. However, to control a prosthetic limb, joint position, torque, or impedance has to be controlled, and one cannot directly determine such mechanical signals using electrical activation alone,” says Hugh Herr, a professor of media arts and sciences, co-director of the K. Lisa Yang Center for Bionics at MIT, an associate member of MIT’s McGovern Institute for Brain Research, and the senior author of the new study.

In the new study, which appears today in Nature Sensors, the researchers demonstrated the accuracy of the magnetic measurements in three participants with below-knee amputations.

“This interface is minimally invasive, so it could be accessible to more people. You place two beads in the muscle and you have access to all of this rich information about the muscle dynamics,” says Christopher Shallal, an MIT graduate student in the Harvard-MIT Program in Health Sciences and Technology and one of the lead authors of the paper.

The paper’s other lead author is Cameron Taylor PhD ’20, an assistant professor of biomedical engineering at the University of North Carolina at Chapel Hill.

Tracking muscles

For many years, Herr’s lab has been developing human-machine interfaces that can be used to control an artificial limb in a way that makes it more like using a natural limb. Obtaining accurate measurements of how the user is moving the muscles at the amputation site is critical to the success of this type of prosthesis.

Electrodes on the surface of the skin can reveal the electrical activation of the muscle below, and those measurements can be converted to estimates of how much a muscle is contracting or flexing. Based on that input, a robotic controller generates commands that guide the prosthetic limb in the user’s intended direction. 

However, that measurement approach, known as electromyography, does not always provide accurate enough information. The electrical signal is not a perfect proxy for muscle dynamics; for example, when a muscle is held at a fixed length for several seconds, the signal fades. Additionally, the electrodes may pick up signals from muscles other than the intended target, and sweat can interfere with detection of the signal.

Last year, Herr and his colleagues reported a “bionic knee” prosthesis that uses electrodes implanted in the muscles to provide accurate measurements and feed them into a robotic controller. This approach proved successful in a small clinical study, but alongside that work, Herr’s team has also been working on a less invasive approach that relies on magnets instead of electrodes.

With this approach, known as magnetomicrometry, pairs of magnets are implanted into muscles. When a muscle contracts or stretches, external sensors placed on the skin can measure the change in distance between the magnets, allowing muscle length and velocity to be calculated.

“Magnets can be tracked wirelessly through the body, so you can get very specific muscle signals and measure the mechanical output of the muscle without needing to be extremely invasive,” Shallal says.

Better control

In a 2021 study led by Taylor, Herr’s lab first demonstrated the magnet-based technique in animal models. The new Nature Sensors study extends the approach to humans with long-term magnetic implants. For the study, the researchers recruited three participants who have below-knee amputations.

All of these participants had a type of amputation that creates what’s called an agonist-antagonist myoneural interface (AMI). With this surgical procedure, muscle pairs are reconnected during surgery so that they still dynamically communicate with each other within the residual limb. This sensory feedback, called proprioception, helps the wearer of the prosthesis to decide how to position and move their prosthetic limb. 

Participants had magnets implanted in up to four muscles of the ankle joint. The surgeries were performed at Brigham and Women’s Hospital in collaboration with plastic surgeon Matthew Carty, a co-author of the study. The beads are 3 millimeters in diameter, about the size of a sesame seed. The magnets are made from an alloy of neodymium, iron, and boron, and are coated with gold and a polymer called parylene C to make them biocompatible. The magnetic signals can be read from a depth of up to 6 centimeters from the skin’s surface, using a customized external sensor that is worn over the skin, positioned over the muscles that are being tracked. 

Each of the participants in this study also had electrodes implanted into the same muscles that received magnetic beads. This allowed the researchers to compare the accuracy of three different types of measurements: magnetomicrometry, electromyography with implanted electrodes, and electromyography with electrodes on the surface of the skin.

During the experiments, the participants were not wearing a prosthetic limb but were asked to perform tasks such as moving a cursor on a screen to a particular target using their muscle movements. Using each of the three measurement techniques, the signals generated by the muscle movements were translated into control signals that, in this case, moved the cursor but in the future could be fed into the controller of a robotic ankle-foot prosthesis.  

“The cursor on the screen represented the degrees of freedom that they would control on a robotic ankle-foot prosthesis. This virtual neuroprosthetic task allowed us to compare the performance across different neural interfaces, and our results suggest that the magnet-based approach can offer more accurate and stable control,” Shallal says.

The researchers are now performing a larger study in which they are evaluating how well people can control prosthetic limbs using measurements from the magnetic beads. 

The research was funded, in part, by the Yang Tan Collective at MIT as well as a National Science Foundation Graduate Research Fellowship.

A user who is frustrated with changes a social media company made to their favorite messaging platform, like the introduction of an annoying AI chatbot, can’t do much to address the situation, short of leaving the platform.

Now, MIT researchers are offering an alternative path.

They’ve created Social.Wiki, a system that allows users to easily edit any social application made within Social.Wiki, and choose which of the edits made by others they want to see. 

For example, a user could change a messaging site on Social.Wiki to remove an AI chatbot and instead add a restaurant voting feature so their group chat can finally decide where they are going out for dinner.

“The infrastructure that we are all dependent on online is very rigid. It is based on these one-size-fits-all applications that we don’t have a lot of power to change. This project is about trying to imagine what a different web would look like where anyone can participate,” says Theia Henderson, an electrical engineering and computer science (EECS) graduate student, and lead author of a new paper on Social.Wiki.

The researchers envision a world in which, if a user is unhappy with their social app, they can edit it to change it however they like — without getting cut off from their social connections.

The platform enables users with no coding expertise to create a range of websites, including social sites like photo sharing networks, discussion forums, and dating apps. It handles all back-end work, eliminating the need for creators to set up servers or arrange databases.

Social.Wiki employs a decentralized infrastructure that allows users to retain ownership of their data as they create, edit, and navigate social websites within the platform. It also includes a “data guard” that protects user data from malicious or buggy applications. 

Through case studies and user tests, the researchers demonstrated the breadth of applications people can create with Social.Wiki, including a rideshare app where riders propose the prices they are willing to pay and a collaborative canvas for creating art together.

While Social.Wiki is still in its early days, the researchers plan to continue refining the platform as they deploy it to a wider audience. 

“We want to empower the creation of far more social applications that meet the needs of unusual small groups. Each group is unusual, but everyone is in unusual groups. Everybody has their own needs for social applications that we want to be able to address,” says David Karger, professor of EECS, a member of the Computer Science and Artificial Intelligence Laboratory (CSAIL), and senior author of the paper.

Henderson and Karger are joined on the paper by EECS graduate students Carmel Schare and Ana Dodik; Clemens Klokmose, a professor of computer science at Aarhus University in Denmark; Ziv Epstein, a postdoc in the MIT Media Lab; and David D. Clark, an MIT research scientist. The paper will be presented at the ACM Symposium on User Interface Software and Technology in November.

Letting users choose

Social.Wiki is a Wikipedia-like system in which users can navigate to different web applications or create their own web application within the Social.Wiki ecosystem. Each website has an “edit” button that allows any user to change the content of the site, from the text color on a message board to the pricing algorithm in a ride-sharing app.

Clicking the “edit” button in Social.Wiki opens a page that lets users view and edit the source code manually or prompt an AI assistant to make changes to the site. 

This platform builds on the researchers’ prior work on Graffiti, a web framework with a decentralized infrastructure that makes building personalized applications easier. 

With Social.Wiki, the researchers sought a collaboratively editable system that would allow anyone to revise web applications to be ideally suited for their needs and the needs of groups they belong to. 

To enable this structure, Social.Wiki relies on a strategy the researchers call “plural governance.” Instead of a central moderator deciding which content users will see, each Social.Wiki user can choose the rules that determine which edits they will see on Social.Wiki webpages.

Users can also identify “trusted editors,” so they will see edits from individuals or organizations they consider reliable.

For instance, a user could customize their rules, or “governance policy,” so they only see edits to a site that have been endorsed by at least two of their friends. When the Social.Wiki browser loads a site for a user, it applies the governance policy that user selected to decide which version of the site to present.

For example, based on each user’s choices and edits, the system might show one user a ride-sharing app with a uniform pricing algorithm while another user might see the same app with a dynamic pricing algorithm.

“No website is tied to a particular machine. This means we have no centralized ownership over the websites or the underlying infrastructure,” Henderson explains.

Security and creativity

In this decentralized infrastructure, user data are stored separately from the websites that access it, which helps keep data secure.

Social.Wiki also employs a data guard, which prompts a user to review each request from a website that is attempting to access their data.

“In Social.Wiki, these websites are held in common and anybody can edit them, but the data guard makes sure that each user stays in control of what happens to their data,” Karger adds.

The researchers evaluated Social.Wiki through several case studies. For one, they developed a messaging platform that allows participants to modify the chat’s styling and interface without affecting the experience of others. In another, they built a collaborative ride-sharing app that enables riders to adjust the fare they want to pay using a transparent “living wage” calculator.

They also sponsored a hackathon in which users built Social.Wiki applications.

Participants developed a range of creative sites, including a housing co-op communication tool that organizes chats on a map of the co-op, as well as a garden-themed microblogging site where posts are “planted” as flowers.

“Even the little bit that has been done so far with Social.Wiki demonstrates that there is so much more that could be done with social apps if users could specialize them to meet a particular group’s needs,” Karger says.

Moving forward, the researchers are working to deploy Social.Wiki to a broader audience and gather feedback from more diverse user groups. They also want to make technical upgrades to enable end-to-end encryption on private message data and provide better developer tools.

This work was funded, in part, by Schmidt Sciences and the National Science Foundation (NSF). 

For decades, scientists have been able to look at individual moments in the development of an embryo: a snapshot of a developing brain, a picture of how the heart takes shape, or a glimpse of cells beginning to form the nervous system.

But what if instead of taking snapshots, they could watch the entire process unfold?

A new study out of the lab of MIT Professor Jonathan Weissman, a member of the Whitehead Institute for Biomedical Research and a Howard Hughes Medical Institute investigator, offers one of the most detailed views yet of how a mammal is built, cell by cell.

By following the family history of over a million cells as a mouse embryo develops, the researchers have created a kind of developmental blueprint, or movie, showing how cells divide and ultimately become the tissues and organs in the body.

This open-access work, published Oct. 8 in the journal Cell, in collaboration with Zachary Smith’s group at Yale University and the Max Planck Institute for Molecular Genetics, provides a comprehensive view of a process that has long been studied in pieces: how a single fertilized egg gives rise to hundreds of different kinds of cells, arranged in the right places and at the right times to build an animal.

“This is really a view of how the information in the genome plays out over space and time to go from linear DNA to a three-dimensional animal,” Weissman says.

The map could have broad implications for understanding birth defects and cancer. Many birth defects arise when cells change course as tissues and organs are being assembled. A detailed map of typical development could help researchers pinpoint where and when those processes go awry. In cancer, the same approach could help scientists trace when cells acquire traits such as the ability to spread or resist treatment. 

The map could also inform stem cell research by revealing the sequence of cell states that leads to mature, functional cell types.

In the 1980s, scientists ventured into similar territory by painstakingly following the development of Caenorhabditis elegans, a tiny worm whose transparent body contains only about 1,000 cells. Using microscopy, they could trace how individual cells divided and where their descendants ended up.

But achieving something similar in a mammal was a challenge on a vastly different scale. Mammalian embryos contain far more cells, develop inside the uterus — out of sight — and do not follow a completely predetermined script.

The new study tackles this challenge using a technology developed in the Weissman Lab called PEtracer. The system acts somewhat like a flight recorder inside a cell. PEtracer makes small, heritable changes to DNA that are passed down as cells divide. By reading those changes later, scientists can work backward to reconstruct which cells came from which ancestors, essentially building a family tree for the cells in the developing embryo.

In this work, the researchers combined these family trees with information about what genes were turned on in each cell and where these cells were located in the embryo. They analyzed over a million cells from mouse embryos collected at multiple points during development, allowing them to follow not just where cells ended up, but when they began heading toward particular fates.

A new study from Jay Shendure’s group at the University of Washington, published concurrently in the journal Science, uses a related approach called DNA Typewriter to track cell divisions in developing mouse embryos and build a record of their cellular history. 

Both studies provide a much more comprehensive look at mammalian development than existed previously. The Weissman team, however, studied embryos at closely spaced stages of development, looking at multiple embryos at each stage. This gave them a more detailed picture of how development unfolds and allowed them to ask how consistently cells follow the same paths from one embryo to another. 

With this detailed timeline, the researchers could also determine when cells commit to particular fates, how many early cells go on to form a tissue, and how they ultimately contribute to the developing organ.

While earlier studies might focus on one cell population, this approach puts those individual stories into the context of the entire developing animal — a perspective that is particularly useful for understanding disease. 

Cancer, for example, is not a single event, but a process in which cells acquire new characteristics over time, including the ability to spread or resist treatment. A lineage map could allow researchers to trace when those characteristics emerge and how cells change as they acquire them.

“Development and cancer both depend on decisions cells make about what they will become,” says Luke Koblan, a postdoc in the Weissman Lab and co-first author of the paper. “By tracing how cells develop in mouse embryos, we hope to better understand how cancer cells change in ways that help them spread or resist treatment.”

The work could also provide a new guide for stem cell research. Scientists have spent decades trying to coax stem cells into becoming mature, functional versions of particular cell types for research or potential therapies. But reproducing the precise conditions in an embryo is difficult.

“If we could have a step-by-step recipe for how to create every cell type in the body,” says graduate student and co-first author William Colgan, “that would be incredibly powerful for stem cell biology.”

Weissman compares the work to the Human Genome Project: knowing the sequence of the human genome gave scientists the ingredients — but not the instructions for how those ingredients are used — to build an organism.

“The genome is like the ingredient list for a four-course meal,” he says. “It tells you what the ingredients are, but it doesn’t tell you which ones you use when, how much you use, or where you put them. Without the recipe, it’s impossible to know how those ingredients can become the meal.”

This new map begins to fill in the details of that missing recipe. The result is a resource that researchers can use to explore mammalian development in a fundamentally new way. Rather than studying development as a collection of separate processes, scientists can now begin to see how those processes fit together into one unfolding story. 

During its annual meeting on Oct. 19, the National Academy of Medicine (NAM) announced election of 100 new members, including MIT faculty member Michael Yaffe and six MIT alumni: Steven Allen Abrams ’79; Susan L. Ettner PhD ’91; Diane Medved Harper ’80, SM ’82; Albert Icksang Ko ’81; Jennifer Lewis ScD ’91; and Alexander Marson PhD ’08.

Election to the academy is considered one of the highest honors in the fields of health and medicine and recognizes individuals who have demonstrated outstanding professional achievement and commitment to service.

Michael Yaffe, David H. Koch (1962) Professor of Science, is the founding director of the MIT Center for Precision Cancer Medicine, as well as a professor in the MIT departments of Biology and Biological Engineering and a member of the MIT Koch Institute for Integrative Cancer Research. He is also an attending surgeon and intensivist at the Beth Israel Deaconess Medical Center, where he specializes in injury and surgical oncology, and was recently appointed as senior advisor to the director of the National Cancer Institute (NCI) for the NCI Intramural Research Program. 

The NAM recognized Yaffe for “the discovery of phospho Ser/Thr-binding domains in proteins and how they regulate signaling pathways controlling the cell cycle or in response to stress.” His contributions in this area were pivotal to “the emergence of systems biology and bioinformatic approaches to map motifs across the kinome and characterize biochemical principles that determine kinase-substrate selectivity.” 

Yaffe’s work in cell signaling more broadly seeks to understand how signaling pathways that control DNA damage responses, survival, cell cycle progression, and immune activation are integrated at the molecular and systems levels to control tumor outcomes after treatment, and to translate discoveries into more effective cancer drug combinations and treatment strategies.

Yaffe earned his doctoral degree in biophysical chemistry and his medical degree from Case Western Reserve University. Before joining the MIT faculty in 2000, he completed residencies in both general surgery at University Hospitals of Cleveland and surgical oncology at New England Deaconess Hospital, a fellowship in surgical critical care at Harvard Medical School’s Harvard-Longwood Critical Care Program, and a postdoctoral fellowship with Lewis Cantley at Harvard Medical School. Yaffe is a decorated veteran of the war in Afghanistan and retired as a colonel from the U.S. Army Reserve Medical Corps in 2021. He is also the scientific editor-in-chief of Science Signaling, an elected member of the Association of American Physicians, and a 2021 MacVicar Faculty Fellow.  

MIT alumni who were elected to the NAM for 2026 are affiliated with the following departments:

• Steven Allen Abrams ’79 is an alumnus of the Department of Biology.
• Susan L. Ettner PhD ’91 is an alumna of the Department of Economics.
• Diane Medved Harper ’80, SM ’82 is an alumna of the departments of Chemical Engineering and Materials Science and Engineering.
• Albert Icksang Ko ’81 is an alumnus of the departments of Chemistry and Biology.
• Jennifer Lewis ScD ’91 is an alumna of the Department of Materials Science and Engineering.
• Alexander Marson PhD ’08 is an alumnus of the Department of Biology.

Established originally as the Institute of Medicine in 1970 by the National Academy of Sciences, the National Academy of Medicine addresses critical issues in health, science, medicine, and related policy, and inspires positive actions across sectors.

“I am thrilled to welcome these outstanding health and medicine leaders into the National Academy of Medicine,” said NAM President Monica M. Bertagnolli. “Their extraordinary talent and commitment to advancing evidence-based guidance and solutions for our country’s greatest health-care challenges will strengthen our efforts to improve health for all.”

One of the main ways that the APOE4 gene variant, the biggest genetic risk factor for late onset Alzheimer’s disease, undermines brain circuit function is by disrupting the process of insulating neurons with a fatty sheath that optimizes their electrical transmission. A new study led by neuroscientists at the Picower Institute for Learning and Memory at MIT provides several lines of evidence to implicate a specific molecular pathway in that disruption, and uses that insight to demonstrate an intervention for preventing it in lab experiments.

The study, published Oct. 15 in Cell Stem Cell, shows that APOE4 leads to a suppression of “Wnt” signaling in the oligodendrocyte cells that provide neurons with a fatty insulation, called myelin. When Wnt signaling is deficient, they showed, the cells accumulate droplets of lipids rather than deploying fats to properly myelinate the neurons’ axon branches. When the researchers restored Wnt signaling in APOE4-carrying oligodendrocytes, myelination rebounded, both in human cell cultures and in mice modeling Alzheimer’s.

The research team, led by Picower Professor Li-Huei Tsai and study lead author Leyla Akay PhD ’25, a former graduate student in Tsai’s lab, discovered Wnt’s key role via multiple methods, including a screen in which they used CRISPR gene editing to systematically knock out each protein-coding gene in the genome, one by one, in cultures of human APOE4-carrying oligodendrocytes. The screen showed that when genes promoting Wnt activity in oligodendrocytes were lacking, lipid droplets accumulated, and that when genes suppressing Wnt were lacking, lipid droplets were, too.

“This proof of principle of doing CRISPR screens for genetic modifiers of APOE4-associated lipid accumulation in a central nervous system cell type is very exciting,” says Tsai, who also directs MIT’s Aging Brain Initiative. “I think this type of screen not only provides new insights into cellular pathways contributing to lipid metabolism defects, but really also has the potential to provide new targets for drug development.”

For want of Wnt

In 2022, Tsai’s Lab, led by Akay, Djuna Von Maydell PhD ’25, Jose Davila-Velderrain, and Joel Blanchard (also a co-author of the new study), showed that APOE4 caused lipid droplet accumulation in oligodendrocytes and upset myelination. But, Akay says, that droplet accumulation could happen for many reasons. The new study sought to dig deeper into the underlying molecular mechanisms.

“This CRISPR screen allowed us an unbiased way to test every single possibility,” Akay says. “What kind of lipid enzymes are involved in this process? And what are upstream pathways that could be targeted?”

In parallel with the CRISPR screen, the team also analyzed the chromatin accessibility of the DNA in human oligodendrocyte cell cultures and in postmortem human brain samples, in each case comparing APOE4-carrying samples to ones carrying the neutral APOE3 variant. This analysis of ATAC-seq data allowed them to assess which genes were most accessible for expression versus which ones were locked down from transcription. The chromatin analysis showed that genes promoting Wnt signaling were less accessible in APOE4 carriers.

“There was a striking overlap between ATAC-seq and CRISPR screen results,” Tsai says. “Both pointed to a dysregulation of the Wnt signaling pathway.” 

Toward a therapy

Further analysis, including direct observations in cells at the lab bench, revealed how Wnt signaling became depressed in cultured APOE4 oligodendrocytes. Two proteins, beta-catenin and GSK3b, each compete to regulate Wnt — beta-catenin promotes it, and GSK3b suppresses it. The team found that in APOE4 oligodendrocytes compared to APOE3 ones, there was reduced beta-catenin expression and excessive GSK3b expression.

The team then tested interventions that manipulate GSK3b activity. In human APOE3 oligodendrocyte cultures, they showed that increasing GSK3b with a chemical called Dif3 increased lipid droplets. In APOE4 oligodendrocytes, they showed that inhibiting GSK3b with a chemical called CHIR99021 reduced droplets. In more complex “organoid” cultures combining APOE4 oligodendrocytes with neurons, they found that inhibiting GSK3b increased myelination and promoted the growth of neural axons.

The tests suggested that GSK3b inhibition could be a proof-of-concept therapeutic strategy for restoring Wnt signaling and myelination. As a final test, they treated mice carrying APOE4 and genetically modeling Alzheimer’s tau pathology with doses of CHIR99021. The treated mice exhibited a reduction in lipid droplets in the hippocampus of the brain, a key memory-making region. Moreover, they also showed increased myelination there, compared to untreated controls.

In addition to highlighting suppression of Wnt signaling, the CRISPR screen and chromatin analysis provided key clues about the molecular mechanisms by which loss of Wnt alters how oligodendrocytes metabolize lipids. Akay and Tsai say they are eager to follow up on those clues. They are also interested in investigating GSK3b’s linkage to tau phosphorylation, a major feature of Alzheimer’s pathology.

In addition to Tsai, Akay, and Blanchard, the paper’s other authors are Anna Bright, Daiyun Dong, Ethan Smith, Robert Hurt, Carles Boix, Kate Louderback, Janine Medrano, Dingcheng Sun, Haonan Lin, Oisín King, Gwyneth Welch, Hiba Nawaid, Emre Agbas, Alan Jiang, Adele Bubnys, and Ji-Xin Cheng.

The National Institutes of Health, the Robert and Renee E. Belfer Family Foundation, the Manthripragada McManama Alzheimer's Fund, the CAMA Seed Fund, the J. Crayton Pruitt Foundation, Joseph P. DiSabato and Nancy Sakamoto, and the Freedom Together Foundation provided funding for the study.

Raw data from the study is available on the Open Science Framework. Akay and Tsai have applied for a patent on the CRISPR screen results.

Disorders of the small intestine are often difficult to diagnose and treat. More than 20 feet long, an inch in diameter, and located deep within the body, this organ is harder to access than other parts of the digestive tract, such as the colon or stomach.

To overcome those challenges, MIT engineers have designed a new robotic enteroscope that can move easily through the narrow passage of the small intestine, allowing it to perform imaging, deliver drugs, and remove small polyps or tissue samples for biopsy.

In designing the new robot, the researchers were inspired by the locomotion of snakes, as well as the scaly skin that helps them gain traction on slippery surfaces. A special textured coating helps the robot remain attached to the mucosal lining of the intestine as it inches its way through the narrow tube. 

Similar to an endoscope, the robot contains a variety of tools to help detect and treat any abnormalities that it might encounter.

“When we evaluate the small intestine, we do it for either diagnosis or treatment. If we find a little polyp that needs to be removed, this robot allows us to take a biopsy or remove it,” says Giovanni Traverso, professor of mechanical engineering at MIT, a gastroenterologist at Brigham and Women’s Hospital, and an associate member of the Broad Institute of MIT and Harvard.

Traverso is the senior author of the study, which appears today in Science Robotics. The paper’s lead authors are former MIT postdoc Binbin Ying, who is now an assistant professor at University of Texas Southwestern Medical Center; MIT graduate student Ahmad Mujtaba Jebran; and former MIT research scientist Sahab Babaee.

Snaking through the intestine

The most common problems that doctors need to evaluate within the small intestine include bleeding, lesions, and blockages.

Currently, the only way to diagnose those intestinal complications is with a procedure known as double balloon enteroscopy. During the procedure, an enteroscope advances through the small intestine as a health care worker alternately inflates and deflates two balloons that help propel it through the tract. 

This process is slow and laborious, so the MIT team wanted to design an alternative enteroscope that would be easier to control and could move faster through the intestine.

“The fundamental challenge that we’re trying to overcome is how to evaluate that 20-to-25-foot segment of the small intestine in a way that can be more automated,” Traverso says.

The new tube-like robot moves through the narrow small intestine using a locomotion strategy similar to that of the snake. When slithering along the ground, a snake can steer itself by first contracting muscles on one side of the body, then the other. 

To mimic this, the researchers designed multifunctional actuators that control the robot’s movement. Each actuator contains four chambers, which can be inflated in different combinations, prompting the tube to extend or bend in any direction. 

“By tuning the amount of air that goes into these chambers, you can tune the direction that the robot is moving in,” Traverso says.

The head and tail segments of the robot contain additional expandable anchoring balloons that take turns inflating and deflating, helping the robot move forward. These head and tail segments are coated with special patterned sleeves inspired by kirigami — a variation of origami that involves cutting paper as well as folding it.

Using this technique, the researchers created coatings that lie flat until they are stretched by the expansion of the balloon, which causes three-dimensional kirigami needles to pop out. Similar to the scales of a snake, these needles give the robot more traction within the slippery lining of the intestine. 

In previous work, Traverso’s lab has used this approach to create coatings for shoe soles that give them more traction, as well as stents that can temporarily lodge in the esophagus and release drugs. 

“By using kirigami sleeves to enhance surface friction, the robot can generate localized tissue engagement traction for locomotion without relying on excessive stretching of the delicate small intestinal tissue,” Jebran says.

Diagnosis and treatment

The system also includes electronics that allow for navigation, with a controller that the person operating the device can use to guide the robot through narrow passages and around tight corners. The soft robot is tethered to the external controllers by a cable and is designed so that it could enter the body through either the mouth or the rectum. 

The robot contains tools including a camera for imaging, a sprayer for drug delivery, and clips that can take tissue biopsies or remove polyps. 

In animal studies, the researchers showed that the robot could travel about six and a half feet through the small intestine. They also successfully demonstrated the drug sprayer, tissue clips, and camera. The team is now working on improving the navigation systems and testing this approach for additional applications in the small intestine.

“We hope we can continue to move it forward to eventually develop systems that can help achieve more automated evaluation of the small intestine,” Traverso says.

The research was funded, in part, by the Karl van Tassel Career Development Professorship, the MIT Department of Mechanical Engineering, a Banting Postdoctoral Fellowship, an NSERC Postdoctoral Fellowship, the UT Southwestern Medical Center Startup Fund, Samsung Semiconductor and Eurnekian Nanotechnology Graduate Student Fellowships, the Cooke Undergraduate Transfer Scholarship, and the MIT Undergraduate Research Opportunities Program.