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.

If you are crossing an unfamiliar room in the dark, you may grope around a bit to get a sense of your space.

But for many animals, feeling out a space comes more naturally. A mouse, for instance, can efficiently navigate in the dark just by grazing its whiskers against walls and other obstacles.

Fan Wang, a professor of brain and cognitive sciences and an investigator at the McGovern Institute for Brain Research at MIT, has discovered how neurons in a mouse’s brainstem use signals from the animal’s touch-sensitive whiskers to estimate an object’s distance from the face.

Her team’s findings, published June 25 in the journal Neuron, unlock key circuitry the brain uses to represent the space immediately surrounding the body.

Mapping space

The circuit the team discovered is part of the brain’s system for creating an egocentric map of space — that is, understanding where things are relative to one’s own body. Neuroscientists know that the brain calls on specialized circuits to understand space in this way, which are different from its system for mapping space using external landmarks.

In their study, Wang and her team explored how the brain maps the space closest to the body, known as the peripersonal space. This is the space in which we move, and it is vital that we understand where things are in relationship to our bodies so we can reach, step, avoid hazards, and otherwise interact effectively with our environment.

Wang says mice were an appealing model for investigating how the brain understands objects’ distance within the peripersonal space, because a rodent’s whiskers seem so much like a built-in set of rulers. These whiskers, which vary in length, are swept back and forth as the animals explore their environment. As whiskers bend and vibrate, the mechanical sensations are relayed to the brain by sensory neurons at their base. Those neurons fire more when a whisker bends close to the face than they do in response to contact near the whisker’s tip, communicating information about the proximity of the touch.

Wang’s team wanted to know if the brain uses these signals to build an internal ruler-like representation of distance more precise than “near” or “far.” To find out, graduate student Wenxi Xiao and Research Scientist Kyle Severson monitored neural activity in a small sensory-processing region in the brainstem where tactile signals from the whiskers first arrive in the brain. They studied what happened there as mice walked on a treadmill while brushing their whiskers against a wall that passed by at different distances.

Many neurons in the region were sensitive to the whisker bending triggered by the wall. Some behaved similarly to the sensory neurons they were getting their information from, firing more when the wall was closer to the face and thus serving as a proximity-based distance code. But other cells were tuned in to discrete distances, firing only when the distance of the wall the whiskers had touched was within a specific range.

The whiskers rule

For some neurons, activity peaked when the wall was 23 millimeters away from the face, near the tips of the longest whiskers. Others responded most when the wall was at intermediate distances. “Each of these neurons represents a specific distance, and together they span the full range reached by the longest whisker, like tick marks on the ruler,” Wang explains. “We call that the map code.”

The team wanted to know how the brain converts proximity signals from different whiskers into accurate map code of object’s distances from the head. “You cannot just listen to individual whisker neurons, because a contact at the tip of a short whisker would be in the middle of a long whisker. You need a brain circuit to build a unified distance map,” Wang says.

Through computational modeling and by exploring what happened when they manipulated neural signaling in specific ways, Wang’s team showed how distances can be calculated by comparing inputs from different sensory neurons. Their findings suggest that each brainstem neuron that makes up the map code receives both direct excitatory inputs from proximity-sensitive whisker neurons and inhibitory inputs from neurons driven by proximity-dependent whisker touch signals.

“Essentially, the inhibitory pathway allows the brainstem to compare two inputs by subtraction,” Wang explains. “If one input signals ‘this is how far it is’ and the other signals ‘this is how far I estimate it to be,’ subtracting one from the other yields an intermediate value. We think it’s a simple and elegant way to transform tactile input into a representation of discrete distance.”

Wang notes that despite their importance, the brain’s body-centered representations of space have so far received little attention from neuroscientists, who know much more about how we understand locations in space relative to landmarks (an allocentric map). She is eager to investigate how the egocentric map code her team discovered is integrated with other brain systems to guide movement, social interactions, and other behavior, and hopes the findings will further exploration from other groups.

The study was funded by grants from the National Institutes of Health.

In today's world, artificial intelligence chatbots such as ChatGPT and Claude can perform many functions, such as composing work emails and planning travel itineraries. These chatbots are systems built around large vision-language models (VLMs): AI trained on a massive dataset that includes books, websites, code, and images. 

The AI algorithms are then refined on massive amounts of human-generated feedback to follow instructions and avoid harmful or unwanted output, and use that "knowledge" to produce text or images based on input from a user. Although chatbots have clear limitations, they can be very helpful for a wide range of tasks, including in some areas that traditionally require specialized skills, like computer programming.

As part of a project for the U.S. Department of the Air Force–MIT AI Accelerator's Phantom Program, U.S. Air Force cadet Joshua Lynch — with the help of his mentor, Laura Niss, a technical staff member in the Embedded and AI Systems Group at MIT Lincoln Laboratory — wanted to determine if, as a complete novice to coding, he could develop a fully functional program. He used a process called "vibe-coding," in which a user relies entirely on prompts to guide a generative AI chatbot to write and refine code. 

His motivation was to empower anyone familiar with the military problem space, regardless of their technical background, to advance their ideas for useful software applications, essentially bypassing the time and cost constraints of the traditional military software development pipeline. Lynch aimed to build his own application while Niss monitored his experience with the technology.

"The Phantom student wanted to see if he could create a useful application through self-identified vibe-coding, without any previous experience," Niss says. "Within this project, I wanted to understand how his perception of AI changed over time with use. We both wanted to understand better where and how AI could be used by nontechnical users in the military."

Lynch set out to see if, starting with no coding skills and using chatbots, he could create an application specific to his type of tactical team to help reduce collateral damage while enhancing survivability in the broader mission. This application would offer capabilities including AI-assisted target recognition; modular intelligence, surveillance, and reconnaissance; autonomous striking; and communication management on the battlefield. 

During the project, Lynch completed several professional development courses in AI and familiarized himself with both military and nonmilitary uses of the technology. For the basis for his code generation, he used the paid models of three AI chatbots: Anthropic's Claude, OpenAI's ChatGPT, and Google's Gemini. Most of this work was done only through the chatbots' main chat function on a web browser, not as an integrated system within a development environment, as is standard now. The final application was produced using Google AI Studio App, which can create applications that interface with the Gemini application programming interface and has AI integrated in the development environment. 

Over three months, Lynch worked with these models to build his application, called the Remote Operating Modular Augmentation Device (ROMAD-AI). During this time, he learned several methods to improve the code output. For example, he often encountered difficulties with the AI chatbots lacking hierarchical focus and modifying unrelated code sections. He discovered it was important to break problems into small parts, frame questions clearly, and steer conversations back on topic when they stray too far from the objective. 

Learning to recognize the chatbots' limitations and effectively work around them took up most of the project timeline. As Lynch gained more experience with the chatbots, limitations in the AI capabilities and time for development caused him to re-scope the project, moving it from an application that could assist on the battlefield to one that could perform basic document processing, such as analyzing tactical maps of battlefields and generating mission-planning documents through an interface with a VLM-powered chatbot. While the resulting prototype did not perform all capabilities Lynch originally set out to include (and in its current iteration was not secure for the desired use case), it proved the capability and usefulness of such an application for service members.

"I was quite impressed with this final product, and it showed me how powerful these systems can be at prototyping designs from nonexperts," Niss says. "I'm now of the opinion that these can be powerful tools for nontechnical experts to convey problems and possible solutions to technical experts, and aid in communicating desired outcomes."

Niss observed the change in Lynch's perspective of AI language models during his experience. After starting with an impressive goal, Lynch gained understanding of the capabilities of current technology and significantly scoped down his expectations by the end of the project period. Measures of his perceptions of the different AI systems over time and across system updates were particularly interesting to Lynch and Niss, with Claude showing more stability than ChatGPT across traits such as likeability, anthropomorphism, and perceived intelligence. Lynch found AI to be a helpful tutor, but noted its inaccuracies on topics he knew well.

The project showed that AI chatbots can empower nontechnical service members to produce viable software applications for their unique problems, although it works better as a prototyping assistant than as a full production tool when handling sensitive information and for critical applications. Improper vetting of code may lead to security risks, as demonstrated by an instance where Lynch didn't realize that the final application was sending the input documents to a Gemini AI model to analyze, rather than parsing the documents locally on his computer. Although AI can generate significant amounts of functional code, code review remains a bottleneck in this space.

"For me, this project reinforced the expanse between experts in different fields," Niss says. "No matter how good AI gets, I think we'll always need to collaborate to get to the best solutions for the most important problems."

Research was sponsored by the Department of the Air Force Artificial Intelligence Accelerator and was accomplished under Cooperative Agreement Number FA8750-19-2-1000.

When a star dies, a black hole is born. This has been the textbook origin story for most black holes. At the end of a massive star’s life, its outer layers blast away in a brilliant supernova, and its core collapses into a gravitationally tight and dense region, forming a black hole.

Recent discoveries from gravitational-wave detectors have revealed hundreds of merging black holes across the universe. Many of them have been thought to come directly from exploding stars. But black holes can also come from other, smaller black holes. The products of previous black hole mergers can, in principle, merge again, creating a more massive black hole. This alternative, black-holes-birthing-black-holes pathway is known as “hierarchical merging.”

Now MIT scientists are finding that a good number of merging black holes may have indeed merged before. They carried out a new analysis of recent data from the LIGO, Virgo, and KAGRA observatories, containing 155 pairs of binary black holes, and found about 14 percent of merging black holes in the universe may in fact be second-generation black holes that formed from the previous merging of two smaller black holes. 

The results, which the team reports this week in Physical Review Letters, suggest that repeated hierarchical merging is a significant pathway by which black holes form. 

“We’re finding that, for some of these merging black holes, it’s not their first rodeo,” says the study’s first author, Cailin Plunkett, a graduate student in MIT’s Department of Physics. “Overall in the universe, black holes are merging all the time. The question of how often are they repeatedly merging was pretty uncertain. Now we’re seeing a relatively consistent picture where there’s a decent percentage of black holes that are coming from this repeated pathway.”

The study’s co-authors are Salvatore Vitale, associate professor of physics at MIT; Thomas Callister of Williams College; and Michael Zevin of Adler Planetarium and Northwestern University.

Lopsided pairs

When a massive star collapses and dies, the resulting black hole should have very little spin. In addition to losing a huge amount of mass when it explodes, the star should also lose much of its inherent spin, or angular momentum. The black hole left over should then have little to no spin. 

In contrast, when two black holes merge, the collision should create a new, wildly spinning second-generation black hole. 

“They would be spinning very fast, at about 70 percent their maximum possible spin,” Vitale says. 

Scientists suspect that hierarchical mergers occur in dense stellar environments, where stars are so tightly packed together that multiple neighboring stars could die and collapse to form black holes that are then close enough to merge with each other to form second-generation black holes. 

“You might have a ton of stars whizzing around each other, and if some are massive and explode, they become black holes. The black holes continue to whizz around, and can capture each other and merge,” Plunkett says. “This process can repeat potentially ad infinitum, by virtue of the fact that you have a ton of stars and black holes in this really dense environment.”

One sign of a hierarchical merger is that one black hole in a pair of merging black holes has a much higher spin, and higher mass, than the other. Such a lopsided duo would signal that at least one of the black holes came from the collision of two previous black holes. 

In 2024, scientists detected two such lopsided mergers in signals recorded by the LIGO, Virgo, and KAGRA observatories. The observatories detect incoming gravitational waves — incredibly small wobbles in the fabric of space and time — that are the reverberations from distant cosmic phenomena, such as colliding black holes. 

The observatories detected two gravitational-wave signals, labeled GW241011 and GW241110, each of which likely contain a black hole spinning much faster than its partner. The hierarchical mergers were discovered by analyzing each signal in detail to tease out the specific masses and spins of the black holes involved in each merger.

That work inspired Plunkett and Vitale to do a search of similar hierarchical mergers using all the gravitational-wave signals that the observatories have captured to date. 

A pattern of wobbles

For their new study, the team analyzed the LIGO-Virgo-KAGRA Gravitational Wave Transient Catalog 4.0 (GWTC-4.0), which comprises gravitational-wave detections from the observatories’ fourth observing run. Rather than analyze each gravitational-wave signal one by one, which is what scientists did for GW241011 and GW241110, Plunkett and Vitale searched for a characteristic pattern of hierarchical mergers across the data overall, to see if any matching signals popped out.

The pattern they searched for represents a range of orbital “wobbles.” Just before they merge, two black holes spiral toward each other in a disk-like, orbital plane. When the spins of the pair are perpendicular to the plane, this remains relatively steady. But when one or both spins are not perpendicular to the plane, the disk will wobble. The degree to which the whole plane wobbles, or “precesses,” can tell scientists about the balance of masses and spins between the two spiraling black holes. 

Plunkett and Vitale developed a model for the range of wobbling that should be a sign of a hierarchical merger, specifically between a first-generation and a second-generation black hole. 

The team applied the model to the entire GWTC-4.0 catalog, which comprises gravitational-wave signals from 153 black hole mergers, in addition to the signals from GW241011 and GW241110. Their analysis revealed that a number of mergers fit the pattern for orbital wobbling that was likely caused by the colliding of first- and second-generation black holes. 

Specifically, they found that roughly 14 percent of merging black holes in the universe may have merged before, and that these second-generation black holes had very particular masses: Black holes of around 10 solar masses (10 times the mass of the sun) and 30 solar masses were run-of-the-mill star-born black holes, while second-generation black holes had masses of around 20 solar masses or 40 solar masses and above. 

“One of the reasons why the 40-and-above regime is interesting is, stellar evolution theory predicts you shouldn’t be able to form black holes in that mass range at all from just a supernova,” Plunkett says. “We think supernovae from really massive stars end up being so violent that they leave no black holes at all above roughly 45 solar masses. Yet we have seen black holes that are that massive. And the question is: Where did they come from?”

The team’s new analysis provides support for the idea that black holes can form from the repeated merging of other black holes, and that this alternate origin story could explain some of the curious black holes that we can detect today. 

This work was supported, in part, by the National Science Foundation, and the Brinson Foundation.

Many experts refer to hydrogen as “the fuel of the future.” It is expected to help decarbonize the global economy in two main ways: burning it or feeding it into a fuel cell produces storable energy with no carbon emissions, just water. And it can be used in place of fossil fuels or as a chemical feedstock in hard-to-decarbonize industrial processes such as steel and cement production.

But for hydrogen to realize its potential, two challenges must be overcome. Researchers worldwide are now working to address the first: finding a method of producing pure hydrogen that’s both cheap and low in carbon emissions.

Just as critical is finding a good means of transporting and storing hydrogen. A team led by researchers at the MIT Energy Initiative (MITEI) has been tackling that less-discussed but important challenge. The location where the pure hydrogen is produced is likely to be far away from where it will be used, so moving it will be critical — and difficult.

The problem stems from two characteristics of hydrogen: It’s the lightest gas there is, and it has low energy density per volume. Therefore, delivering a given amount of energy requires a large volume of hydrogen and a container that’s sealed so tightly that the hydrogen molecules can’t escape. Suffice it to say, moving a liquid fuel such as gasoline is easier. And without a good means of storing and transporting hydrogen, it can’t fulfill its promise as the world’s clean fuel of the future.

In 2024, with funding provided by ExxonMobil Technology and Engineering Co. through MITEI, a team of MITEI researchers and their Exxon colleagues began examining various approaches to transporting hydrogen. The researchers have now concluded that there’s no single answer; the cost and carbon emissions from a given transportation method will vary from one location to another. Therefore, instead of presenting a table showing the “best” outcome, the team created a tool that enables users to understand the various options and choose the best option for their particular use case. 

The researchers present their study and the tool they developed in a new paper published in the journal Fuel.    

The study was led by former MITEI postdocs Gasim Ibrahim, now an R&D engineer/scientist at Honeywell, and Guiyan Zang, former MITEI group lead who is now an associate professor at Washington State University. Additional MIT co-authors include former postdocs Bosong Lin, Jacqueline Garrido, Woojae Shin, and Haoxiang Lai.

The hydrogen challenge and hydrogen “carriers” that can help

The team’s starting assumption was that for hydrogen to become a viable fuel for the world, it would need to be transported over long distances — specifically, overseas, across continents, or across large water bodies. Given the properties of hydrogen gas, it would be best to convert it to some liquid form before shipping.

There are known ways to do that, but what would be best for shipping? How much would various methods cost, and how much would they add to the carbon intensity of the delivered hydrogen?

“There hasn’t been a lot of attention paid to addressing those questions,” Ibrahim says. While some studies have been done, their conclusions are inconsistent and many uncertainties remain, both because the cost and carbon emissions will differ from place to place and because there’s not a lot of data to inform how the large-scale transportation of hydrogen will work.

“So we decided the best thing to do was to develop an adaptive tool that would enable users to perform their own assessments — a tool that could be updated very easily,” Ibrahim explains. “And we would make it open source, so anyone can see and update the numbers that we used in formulating and testing it. As the industry develops, and as scale becomes more a factor, the assumptions made in [our initial] assessments of the economics and the carbon intensity [of different shipping methods] will need to be updated.”

To focus on the transportation and storage issues, their model — called the Hydrogen Carrier Analysis Tool, or HyCAT — doesn’t consider how the starting hydrogen is produced, or how the hydrogen is used after it’s delivered. HyCAT focuses on determining the costs and carbon emissions incurred as the hydrogen is transported and delivered. In addition, while a full life-cycle assessment would include all environmental impacts, HyCAT focuses on emissions of greenhouse gases (GHGs).

The tool is easy to use, says Ibrahim. Built into it is a user interface with drop-down menus for inputting assumptions, and results from an analysis are presented in simple bar charts that include links to tables presenting the details.

Ibrahim clarifies that, while HyCAT has a well-defined boundary — “incoming hydrogen to outgoing hydrogen” — in an analysis of a specific situation, the user will input various factors about the local situation, including the carbon intensity and cost associated with production of the incoming hydrogen. “So that will inform the final values that come out of a HyCAT analysis,” says Ibrahim, and in part explains why the results vary from place to place.

Based on the user’s assumptions, HyCAT calculates the cost and GHG emissions at five steps in the “supply chain”:

• converting the hydrogen into liquid form at the “export” terminal;
• storing the hydrogen-rich liquid;
• shipping it when an empty tanker becomes available;
• storing it at the “import” terminal; and
• releasing the hydrogen as a gas suitable for burning or being fed into a pipeline for distribution.  

Options for liquifying hydrogen gas

The main decision in analyzing the cost and emissions of a proposed hydrogen transport plan is how to convert the gaseous hydrogen to a liquid, and then how to recover the hydrogen gas at the end.

One approach is to simply change the gaseous hydrogen into an easily transportable liquid. But turning hydrogen gas into a liquid requires making it very, very cold. Indeed, notes Ibrahim, “you would need to consume about a third of the energy content of the hydrogen to make the gaseous hydrogen cold enough to liquify.” A further problem arises as the liquified hydrogen is being stored and moved. Unless the vessel containing the liquid hydrogen is properly insulated, the liquid hydrogen can re-gasify and escape. The upside of hydrogen liquefaction is that no chemical reactions are required.

Other options involve using a hydrogen “carrier.” Some liquid chemical compounds will absorb hydrogen atoms under certain conditions, and under other conditions will release them. Therefore, one approach to solving the hydrogen transportation problem is to make a carrier compound absorb the hydrogen where it’s made and then release it when it reaches its destination. This approach therefore involves two chemical reactions — one to bind the hydrogen to the carrier and the other to release it.  

In their demonstration runs, the researchers looked at the hydrogen carriers involving three potential compounds, each of which has known advantages and disadvantages.

One of those carriers is produced by adding hydrogen to toluene. That chemical reaction hasn’t been studied a lot, but there’s one known drawback: the source of toluene is typically the oil and gas industry, so the toluene itself has a relatively high carbon intensity when it picks up the hydrogen. Moreover, over time some of the toluene is lost, so more toluene must be added.
    
The researchers also looked at “synthetic methane,” which is made by reacting hydrogen with carbon dioxide. That reaction has been known for some time. Ibrahim notes that making synthetic methane actually consumes carbon dioxide, often captured from the atmosphere. On the negative side, however, one of the products of the reaction is water, so some of the hydrogen is lost each time the reaction occurs.

The final option they analyzed is ammonia, which forms when hydrogen reacts with nitrogen from the air. That reaction is very well-studied and is used commercially. “We’ve been producing ammonia for a long time,” says Ibrahim. And the infrastructure for transporting and storing it is well established. While Ibrahim refers to ammonia as the “most promising option,” the reaction needed to release the hydrogen has not received much attention.

Varying conclusions and future plans

Based on their sample runs, the researchers observed that the best path to follow will vary from place to place and from situation to situation. “As we developed the tool, we saw that the ‘best’ carrier was very specific to the supply chain at hand,” says Ibrahim. “It’s a function of how far you’re trying to ship your hydrogen, energy and shipping costs at your exporting and importing countries, the capital cost of building the needed facilities at both ends, and more.”

Ibrahim and his team are now planning a follow-up study in which they use HyCAT to analyze specific supply chains under certain conditions. They’ll then select assumptions that are highly uncertain and look at the range of possible values for those assumptions. “Then we’ll be able to say, ‘under these conditions, this carrier is better than that one,’ or ‘this carrier is better at cost, but worse at carbon intensity,’” says Ibrahim.

For now, the main conclusion of the study, says Ibrahim, is that “there’s no conclusion.” He warns decision-makers not to assume that anything they see in the literature can easily be generalized or extrapolated to their specific conditions. Instead, decision-makers should use HyCAT to explore the options available to them. Guided by their results and the objectives and values of their company, they will be able to optimize their supply chains and make clean-burning hydrogen a reality.

Professor Jesse Thaler has been named director of the MIT Laboratory for Nuclear Science (LNS), effective Aug. 1. He succeeds Professor Bolek Wyslouch, who directed LNS for the past decade. Thaler is a theoretical particle physicist who combines techniques from quantum field theory and machine learning to address outstanding questions in fundamental physics. 

“In his research, Jesse has done pioneering work on particle jets at the Large Hadron Collider and is a leader in combining AI and machine learning with fundamental particle physics,” says Nergis Mavalvala, dean of the MIT School of Science and the Curtis and Kathleen Marble Professor of Astrophysics. “The collaborative nature of his research programs will serve the Laboratory for Nuclear Science as science enters a new era of AI-driven discovery.”

Thaler is the William and Emma Rogers Professor of Physics in the MIT Center for Theoretical Physics — a Leinweber Institute (CTP-LI). Since 2020, he has served as inaugural director of the National Science Foundation (NSF) AI Institute for Artificial Intelligence and Fundamental Interactions, or IAIFI, which was recently renewed for another five years. Mike Williams, professor of physics, will succeed Thaler as IAIFI director. LNS is also poised to pursue new research projects through the Department of Energy’s Genesis Mission, which has a focus on AI-enabled scientific discovery.

“In my own field of particle physics, researchers are developing cutting-edge AI algorithms to handle the data deluge from collider experiments and to perform heroic theoretical calculations. This work has direct implications for discovering new physics, but the algorithms themselves turn out to be valuable well beyond our field,” says Thaler. “I’m excited to bring LNS into the next wave of discoveries supported by AI-driven capabilities.”

At IAIFI, Thaler has championed education and research activities at the intersection of physics and AI. With the MIT Institute for Data, Systems, and Society, IAIFI leadership created a doctoral program in physics, statistics, and data science. IAIFI also created dedicated postdoctoral fellowships to give early-career researchers the freedom to pursue interdisciplinary work. 

“Giving young scientists space to build connections across domains, universities, and career stages has been transformative within IAIFI,” says Thaler, who hopes to bring this type of framework to LNS. Established in 1946 to support nuclear and particle physics, LNS now encompasses research spanning cosmology, gravity, field theory, and quantum information science.

As head of LNS, Thaler will also oversee his home center of CTP-LI, which last year received a donation from the Leinweber Foundation to establish a network of theoretical physics research institutes. According to the Science Philanthropy Alliance, a nonprofit organization that promotes philanthropy for science, this constitutes the largest philanthropic commitment ever for this field.

Thaler received his PhD in physics from Harvard University in 2006, and his BS in math/physics from Brown University in 2002. From 2006 to 2009, he was a fellow at the Miller Institute for Basic Research in Science at the University of California at Berkeley. He joined the MIT faculty in 2010.

As advanced medical technology gets closer to hitting consumer markets, the need for guardrails on protected usage should increase. What might begin as a neural implant to aid in communication could become a device used to police one’s innermost thoughts.

Intrigued by the far-reaching benefits and risks of neural implants, Rachel Sava, a PhD candidate in the Harvard-MIT Program in Health Sciences and Technology, explores how a life-changing medical device can become a tool for surveillance by corporations and government entities in her winning submission, “Superintelligence, Superintimate,” for the fourth annual Envisioning the Future of Computing Prize. 

Sava’s concept was inspired by an internship at IBM, where she worked on a project with the PACE Center in London. “A mentor on the project was Kevin Brown, who had himself designed one of the earliest brain decoders — an EEG-based system he built for a colleague who had suffered a stroke that left him with locked-in syndrome,” she says. “It was this patient population for whom the body has become an unreliable vehicle for the mind that motivated my writing about neuroprostheses some six years later.”

Sava explains that research and applications right now are at a “watershed moment in neurotechnology.” Using examples like companies taking advantage of neural implants to monitor mental productivity, or authorities policing a population for “thought crimes,” Sava said that as this tech hits consumer markets, there is a genuine fear that what starts as a revolutionary medical device could transition into more dystopian usages.

Presented by the Social and Ethical Responsibilities of Computing (SERC), a cross-campus initiative of the MIT Schwarzman College of Computing, in collaboration with the School of Humanities, Arts, and Social Sciences and with support from MAC3 Philanthropies, the competition invited MIT students to identify, in 3,000 words or fewer, which sector stands to gain the highest net positive impact from artificial intelligence. Students were encouraged to explore realistic technological deployments while considering potential risks and ethical concerns. All submissions were eligible for cash awards with the grand prize set at $10,000.

During a live awards ceremony hosted by Caspar Hare, former associate dean of SERC and professor of philosophy, who founded the prize in 2023, three finalists each gave a 20-minute presentation on their concepts and took questions from a panel of judges and audience members.

“SERC and the donors who make this prize possible year after year are asking us, the next generation of scientists: ‘what world do you want to see?’ I think it’s worth taking the time to ask yourself the same,” Sava said. “And if, as it did for me, the sentiment grows bright enough to motivate further action — then it’s worth giving yourself permission to explore it as deeply as you do your other academic work.”

Each year, the Envisioning the Future of Computing Prize asks students to look beyond technological advancement and consider the societal benefits and costs of their work from the outset. From its inception, the competition has consistently attracted undergraduate and graduate students from across a wide range of disciplines.

“This year’s submissions were amazing and included essays on brain-computer interfaces, AI and religion, AI for scientific discovery, finding efficiencies in the power grid, and many more,” says Brian Hedden, co-associate dean of SERC and a professor of philosophy, who holds an MIT Schwarzman College of Computing shared position with the Department of Electrical Engineering and Computer Science. “They showed the breadth and depth of thinking going on at MIT on the social and ethics impacts of technologies.”

Nikos Trichakis, co-associate dean of SERC and the J.C. Penney Professor of Management, adds “what is most striking about these essays is the breadth of imagination they display: the students move fluidly across medicine, neurotechnology, law, ethics, and public institutions, while keeping human agency at the center. Their work is creative, rigorous, and deeply thoughtful, showing a remarkable ability to envision not only what AI can do, but what it should do.”

In addition to awarding Sava the $10,000 grand prize, the judges recognized two runners-up with $5,000 each: Cordiana Cozier, a PhD candidate in the Department of Chemistry, for her paper on the use of AI as a cognitive buffer for public defenders; and Strahinja Janjusevic, a graduate student in the Technology and Policy Program in the Institute for Data, Systems, and Society, for his submission on agency and ownership in the field of neural-controlled prosthetics. The judges also named four honorable mentions, each of whom received a $500 cash prize.

Next generation batteries that use new electrolyte materials could achieve far higher energy density than today’s lithium-ion batteries, without many of the safety concerns. But advanced batteries, such as those that use solid or almost-solid electrolytes, have been plagued by the formation of tiny spikes of lithium metal called dendrites that cause the batteries to lose efficiency and fail.

Exactly how those dendrites form is still up for debate. While the interface between the battery’s electrolyte and electrodes has been the focus of most research, another culprit is the boundary where two grains of electrolyte in a solid material meet. Researchers know these boundaries can seed dendrites within electrolytes, although the effects have been difficult to study.

Now researchers at MIT and the Technical University of Munich have uncovered why such boundaries can lead to dendrites: Hidden electrical imbalances across the boundaries affect how the electrolyte conducts electrical charges, which influences how the ions and electrons move through the material during battery operation. In a paper published today in Nature Nanotechnology, the researchers characterized the electrical and chemical behavior of the boundaries and showed that adjusting how the electrolyte is processed enhances the movement of ions while reducing electron leakage. This adjustment can increase critical current density by more than 300 percent, which could enable solid-state batteries that charge faster and last longer.

“Grain boundaries are like the weather: Everyone talks about it, but nobody does anything about it,” says senior author Harry Tuller, a professor in MIT’s Department of Materials Science and Engineering. “In this paper, we’ve decided to do something about grain boundaries, and by doing something we’ve shown improved performance and demonstrated the importance of grain boundaries more broadly.”

Joining Tuller on the paper are first author Hyunwon Chu PhD ’25; former MIT professor Jennifer Rupp, the Electrochemical Material Professor at the Technical University of Munich (TUM), who led the study; TUM researchers Waldemar Kaiser, Lukas Wolz, Fran Kurnia, Kun Joong Kim, David Egger, and Johanna Eichhorn; Thomas Defferriere PhD ’22; Willis O’Leary PhD ’24; and University of Antwerp researchers Proloy Nandi, Johan Verbeeck, Sara Bals, and Thomas Altantzis.

Investigating grain boundaries

Rupp’s research group, which moved from MIT to TUM during this research, has spent years studying the behavior of next-generation electrolyte materials. Electrolytes in solid-state batteries are made of many tiny crystals of material packed together.

“What we call a grain, like a grain of salt, is actually a single crystal, but it might only be on the order of 1 micron in size,” explains Tuller. “Under high temperature processes, the best materials essentially consolidate to be void or pore-free and can be nearly 100 percent dense, but each of those crystallites is separated from its neighbor by a grain boundary.”

Solid-state battery researchers have increasingly focused on grain boundaries as the source of the lithium metal dendrites that cause them to short circuit. It’s been suspected that grain boundaries have different chemical and electrical properties from the grains, which interact with the ions and electrons shuttling between electrodes during battery charging and discharging. However, the exact mechanisms by which the boundaries slowed the ions down, leaked electrons, and led to dendrites was unknown.

“Grain boundaries are like defects,” Tuller says. “The boundaries have a higher level of defects than in the grains themselves, and generally that means as carriers of charge approach the boundary, whether electrons or ions, there’s some kind of blockage to overcome.”

To better understand that interference, the researchers developed a model to explain how local electrical imbalances at grain boundaries change the movement of lithium ions and electronic charge carriers. They tested the model in a common solid electrolyte material called lithium lanthanum zirconate, or LLZO, using techniques including electron microscopy, machine learning modeling, and electrochemical impedance spectroscopy, which measures how easily a charge moves through a material.

They found the cores of the boundaries carry a local electrical charge, building up local electric fields that lead to enhanced ionic resistance while causing a build-up of electrons in the boundary region, where they can reduce lithium ions, leading to lithium metal dendrite formation.

“For the last 30 years, the world has been dominated by lithium-ion batteries, but there is a growing recognition that other battery types are needed for batteries used in a variety of uses,” Rupp explains. “This work gives us the fundamental understanding of the space charge interface at the grain boundary. If understood properly, we can come up with engineering concepts to increase cycle life, transference of ions over electrons at these interfaces, and ultimately a better battery.”

Better battery materials

The researchers used their observations to adjust the material processing conditions of the LLZO electrolyte material and minimize the negative charges at the boundaries, finding they could ease the movement of lithium ions and reduce the leakage of electrons.

The modifications allowed them to create an electrolyte that had a critical current density more than 300 percent higher than a baseline sample. Higher current density allows for faster charging and discharging. It should also delay short circuiting to extend the life of batteries.

“Fires are currently a huge issue in the battery industry,” Rupp says. “By showing how to engineer these space charges in a controlled way, which is new in the field, we can have a strong impact on safety. It’s a new way to turn up the notch and get these batteries to charge faster and last longer before they break.”

The findings, along with the researchers’ engineering work, present a roadmap for battery researchers to accelerate the development of high-performance, longer lasting solid-state batteries.

“We showed we can control the initiation of these dendrites to maximize solid state batteries’ high performance,” Chu says. “In this paper, we started with a theory for how these dendrites form, then we did the material characterization to support that theory, then we did the engineering to apply the findings and actually improve battery performance.”

The work was supported, in part, by the National Science Foundation and the U.S. Department of Homeland Security.

Lerna Ekmekcioglu, the McMillan-Stewart Professor of History, has been named head of the History Section, effective July 1. 

“Lerna is an exceptional scholar and a proven leader. I am confident that she will guide the unit with thoughtfulness, wisdom, and a deep commitment to its continued success. I very much look forward to working with her in the years ahead,” says Agustín Rayo, the Kenan Sahin Dean of the School of Humanities, Arts, and Social Sciences.

Ekmekcioglu, who joined the MIT faculty in 2011, is a historian of the modern Middle East, the Ottoman Empire, and Turkey, Armenian history, gender, feminism, genocide, and minority politics. She served as director of the Program in Women’s and Gender Studies from 2022 to 2025, where she remains an affiliated faculty member.

Ekmekciouglu succeeds Malick Ghachem, who was named head of the History Section on July 1, 2023. 

“As I begin this new role, my first priority is to sustain and expand the remarkable momentum already underway in the unit. It is truly an exciting moment to be head of History,” says Ekmekciouglu. “We have ambitious new initiatives, extraordinary faculty work, and — this is not a small thing — a group of colleagues who actually like and trust one another.”

She cites the History of Now, launched in 2025, as one of several exciting initiatives underway, adding that her role will be ensuring the section’s projects are sustainable, visible, and intellectually fruitful.

“The work ahead is both practical and intellectual: supporting faculty research and teaching, sustaining new initiatives, expanding public engagement, and demonstrating why historical inquiry is indispensable to MIT’s mission,” she says.

Ekmekcioglu’s first monograph, “Recovering Armenia: The Limits of Belonging in Post-Genocide Turkey” (Stanford University Press, 2016), explored the Armenian community in Turkey after the Armenian Genocide and the limits of minority belonging in the early Turkish Republic.

It won the Der Mugrdechian Society for Armenian Studies Outstanding Book Award.

Her forthcoming book, “Feminism in Armenian: Lives and Texts Through Empire, Genocide, and Diaspora,” co-authored with Melissa Bilal of the University of California at Los Angeles, continues her long-standing work on Armenian feminist thought, activism, and archives across empire, violence, and dispersion.

Ekmekcioglu is a 2016 recipient of the the James A. and Ruth Levitan Award for excellence in teaching. She also organizes the biannual McMillan-Stewart Lecture Series on women, gender, religion, politics, and law across the Middle East and North Africa.

Ekmekcioglu earned a BA from Boğaziçi University in Istanbul 2002 and a PhD from New York University in 2010.

On a Wednesday afternoon in April, a cohort of scholars from the School of Humanities, Arts, and Social Sciences (SHASS) gathered in MIT’s Lewis Music Library. 

This group of seven professors are the inaugural SHASS Faculty Fellows, a semester-long program launched this past spring. The faculty represent a variety of disciplines across the school. They met biweekly through the spring to connect over lunch and present updates on their respective research projects. 

At this particular meeting, associate professor of music Emily Richmond Pollock presented some of her work — a chapter about an opera festival in Sarasota, Florida — which, she says, started from “my own curiosity about how American institutions relate to opera’s traditions and practices.” 

After Pollock’s presentation, the group discussed and provided a sounding board for her work. It’s precisely the type of scholarly environment the SHASS Faculty Fellows program was designed to foster.

“The fellows program is a recognition of the fact that not only do we benefit from being in conversation with other scholars, but even more so when in conversation with scholars who do things differently than we do, who approach problems with different opening questions and methodologies,” says Anne McCants, the Ann F. Friedlaender Professor of History and Faculty Fellows Program Committee chair.

Along with committee member and literature professor Arthur Bahr, McCants serves as a kind of moderator during the discussions, asking pointed questions and interrogating participants’ assumptions.

“A small group of people coming from diverse scholarly backgrounds meeting regularly to share a meal and sustained conversation can have a truly outsized impact on their scholarship,” McCants adds.

Time to focus and connect

Faculty must apply to take part in the program, and are selected by the program committee. The program is administered by the MIT Human Insight Collaborative (MITHIC). 

Participants take advantage of opportunities to share and discuss ideas with students, too. Volha Charnysh, a Faculty Fellow and the Ford Career Development Associate Professor of Political Science in the Department of Political Science, presented research on the effects of large-scale humanitarian aid to the Burchard Scholars. The Burchard Scholars program connects faculty and promising MIT sophomores and juniors who have demonstrated excellence in some aspect of the humanities, arts, or social sciences.  

Projects can run the gamut. Participants might develop scholarly articles, develop book manuscripts, or dig deeper into existing research. 

“The Faculty Fellows Program has two primary aims: to enrich faculty members’ scholarly programs, and to foster collegial community within the school,” says Heather Paxson, associate dean for faculty in SHASS, the William R. Kenan, Jr. Professor of Anthropology, and MITHIC faculty co-lead. “Participants in the program gain a better sense of the breadth and depth of our school’s scholarly contributions, and some may forge lasting connections with colleagues they might not otherwise have gotten to know.” 

For Pollock, the fellows program this past spring was an opportunity to focus on her current research.

“I’m working on a book about a set of five opera festivals in the United States,” Pollock says of the project, “Opera on Uncommon Ground: Five American Festivals.” 

“These are annual, seasonal opera companies where rare repertoire is often performed alongside canonical works, in places that are outside of major cities, and performed in unusual spaces.” 

“I hope that anyone who loves opera will be able to read and enjoy my book,” she says, including “opera ‘superfans’” Pollock says she has in mind while writing.

Pollock says the program gave her the space she needed to continue her project. “This semester [in the program] has been wonderful so I could get back to drafting and really concentrate on a book I am excited to write.”

“I am so inspired each week when we meet”

Faculty Fellow Richard Nielsen, associate professor of political science, faculty director of the MIT-MENA Program, and a Security Studies Program affiliate, is hard at work on his project, “Fighting War with Divine Intervention,” a book about how combatants’ beliefs affect wars. Using material from a diverse set of cases — the Islamic State, the Confederate States of America, and the current U.S. engagement with Iran — he wants to understand when claims about divine intervention motivate fighters and citizens to fight harder and longer for victory, even when the state of the battlefield strongly suggests they have lost already. 

“We understand a lot about how religion might shape the conditions for war and peace, but religion matters during wars, too, and we understand surprisingly little about how religious claims affect leaders and fighters in combat,” he says. 

Nielsen lauds the collegial atmosphere available in the fellows program, citing the importance of engagement with scholars outside his research area as a significant draw. “The best part has actually been the engagement with a diverse set of fellows,” he notes, “pursuing a dizzying variety of humanist and social science projects. I am so inspired each week we meet, and every single project has me exclaiming ‘I wish I was writing this!’”

“It adds a regular ongoing conversation with scholars not like yourself who will push you, likely accidentally, in unexpected directions,” McCants says of the fellows’ meetings. Conferring with other participants about their projects, meanwhile, helps Nielsen “return to my research with fresh eyes and enthusiasm,” he says.

Pollock appreciates the camaraderie available as a program participant. “I value my colleagues so highly — the other fellows and mentors are people I really admire and respect — and it’s been fun to trade work and get to read work in progress far outside my field,” she says. 

Twelve professors have been named SHASS Faculty Fellows for the 2026-27 academic year, with six taking part in the fall and another six in the spring. 

The inaugural group of fellows included: 

• Héctor Beltrán, the Class of 1957 Career Development Associate Professor of Anthropology; 
• Volha Charnysh, the Ford Career Development Associate Professor of Political Science; 
• Kevin Dorst, associate professor of philosophy;
• Richard Nielsen, associate professor of political science;
• Emily Richmond Pollock, associate professor of music; 
• Jessica Ruffin, assistant professor of literature; and 
• Robin Scheffler, associate professor of science, technology, and society.

Applications for the next cohort of fellows will open this fall.

In the study of bacteria, a longstanding dogma held that two molecular machines — RNA polymerase, which leads the way in transcribing DNA into RNA, and ribosomes, which bring up the rear translating RNA into proteins — worked so closely in tandem that they were effectively attached. 

This close coupling of transcription and translation in bacteria was thought to be fundamental to gene expression in part because the trailing ribosome could shield nascent gene products from an effective and omnipresent quality-control protein called Rho. 

In bacteria that exhibit something called runaway transcription, however, the polymerase instead speeds ahead, unhitched from its protective ribosome. Inexplicably, however, in bacteria that exhibit this runaway transcription, such as Bacillus subtilis, Rho targeted primarily noncoding, useless RNA products. 

New research from the Department of Biology reveals that the secret to Rho’s quality-control specificity lies in the sequence composition of nucleotide bases that make up coding strands of DNA. 

“We started with a hypothesis that Rho was regulated by sequence, but the fact that the sequence alone was enough to protect any gene in the entire B. subtilis genome from Rho was really surprising,” says Julia Dierksheide PhD ’26, a graduate student in the Li Lab and first author of a paper recently published in Nature Microbiology. “That’s a really diverse range of sequences — what sequence feature is shared by every single gene in the genome?” 

Barricading with bias

Rho serves as a termination factor, meaning that it is a crucial mechanism for preventing bacteria from wasting precious resources by making RNA transcripts that serve no purpose. 

All the information a bacterial cell needs is encoded in its DNA, which is made up of two strands of nucleic acids. These strands twist together to form a double helix, with genetic information codified in pairs of bases: purines guanine and adenine are matched with pyrimidines cytosine and thymine, respectively. Any sequence that gives rise to RNA transcripts is stored in complement to a parallel, noncoding strand, meaning that a large portion of genetic material is transcriptionally useless. 

Coding DNA strands in certain bacteria were known to be significantly higher in purines guanine and adenine compared to the rest of the bacterial genome. The researchers found that this purine bias alone shields productive mRNA transcripts from Rho-mediated termination.

“I love having a big, complicated dataset and trying to reduce that to biological meaning,” Dierksheide says. “It seems like Rho itself has been broadly shaping the evolution of the B. subtilis genome to create these sequence composition biases.” 

Bacterial species that, over generations, have lost Rho no longer exhibit this strong purine bias. 

Rho also serves as a regulatory factor in bacteria becoming motile, forming biofilms, or sporulating, all of which are critical for biology and survival. The purine bias could also provide a layer of protection against the insertion of foreign DNA, for example, when a viral bacteriophage infects bacteria.

“Bacteria exist as single cells, so everything that they do, they have to do through gene expression,” Dierksheide says. “Understanding the fundamental details about how gene expression works, how a cell encodes all the information it needs to survive in the nucleotide sequence of the genome, is really exciting.”

Future directions

Although the exact mechanism underlying Rho’s specificity remains unclear, these results crack an underlying code in the composition of bacterial genomes. 

Dierksheide said she hoped to perform a similar screen to characterize Rho’s specificity in Escherichia coli, which diverged from B. subtilis on the evolutionary tree an estimated 2 billion years ago and still exhibits coupled transcription-translation, where the transcribing RNA polymerase is closely followed by a translating ribosome.

The high sequence specificity of B. subtilis Rho is crucial for the protection of its runaway RNA polymerase, in which that molecular machine speeds ahead of the ribosome. A systematic comparison to E. coli Rho could help reveal how this heightened stringency arose. 

This information will be critical for engineering diverse bacterial species for applications including the production of therapeutic agents. Other bacterial species, such as B. subtilis, may be better models for this process because they have abundant secretion pathways, according to Dierksheide, making it much easier to produce and isolate proteins in large quantities. 

“Our findings reveal an important criterion for successful sequence design that must be considered in expression engineering,” says associate department head, associate professor of biology, and Howard Hughes Medical Institute investigator Gene-Wei Li, the lead author of the study. “There are so many cryptic messages in the genome, like the purine bias, and we are just beginning to be able to decipher what they mean.”

Without federal support for curiosity-driven research, the innovation and talent pipeline that has helped ensure our nation’s prosperity and safety could run dry, warned President Sally Kornbluth during a Washington Post Live event. 

During "The Next Generation," a panel discussion moderated by Washington Post reporter Zachary Goldfarb at The Washington Post’s “Building America Summit,” Kornbluth and Arizona State University (ASU) President Michael Crow joined forces for a spirited discussion on the importance of curiosity-driven research, examining how universities are preparing the next generation of scientists to lead in America’s rapidly changing technological landscape. 

“Many of the things we have in our everyday lives, whether they be medical advances, technological advances, a lot of these things came from 30, 40, 50 years of scientists just trying to figure out how things work,” emphasized Kornbluth.

Kornbluth pointed to MIT’s curriculum that focuses on teaching foundational skills that can be applied to a myriad of technological advances, skills that will be indispensable to leading in an AI-enabled world.

“I do not think that any of our traditional subjects are now outmoded [by AI]. It’s how you approach them,” said Kornbluth. “In our new curriculum, not only are we leaning into basic STEM fields. We really feel we have to resurrect some of the old, moral and civic and ethical educational goals much more strongly because we want all these kids that are learning to be leading-edge technologists, to come at it from a moral, civic and ethical perspective.”

Artificial intelligence

Key to Kornbluth’s mission is maintaining a human-centric approach to AI. Inspired by MIT’s motto, “mens et manus” (mind and hand), she shared: “We really want students to be able to use physical AI. We want our students to still be able to build things, but use AI as an augmentation tool.”

Kornbluth expressed the importance of teaching interested faculty and students how to best use AI as a tool and her commitment to uplifting student collaboration. 

“We’re putting a big emphasis on things like teamwork. So, [students] need to be able to use these tools and come together towards goals, because you could imagine a situation that AI becomes your buddy instead of your study group. We don’t really want that to happen,” said Kornbluth. 

Using AI effectively requires writing strong prompts. Kornbluth discussed how foundational knowledge in fields like math, physics, biology and chemistry, along with teaching students how to write and communicate clearly and effectively, enables students to use AI responsibly when it comes to applying these new technologies to scientific research.

Students need to be able “to take that knowledge and think about how they can use AI to the greatest good and also learn to write the right prompts,” said Kornbluth. 

Kornbluth noted the MIT Sloan School of Management’s unique role in AI exploration. “It’s because the students are all coming with business experience and the demand out there in the field for them to have really strong AI knowledge is very high,” she said. 

The impact of frozen funds

Federal funding fuels curiosity-driven research—the groundwork of medical, technological and countless scientific breakthroughs.

“It is very difficult to make a groundbreaking discovery that’s going to revolutionize human life because you want to do that. You really have to be figuring out how things work and traditionally that sort of research in this country has been funded by the government because it does not have an immediate return,” said Kornbluth.

Discussing issues with federal funding, Kornbluth said that although money has been appropriated for universities, it has not been released to them by and large.

“We’re really trying to figure out what the funding stream is going to be going forward,” said Kornbluth. 

When asked about the consequences of these frozen funds, Kornbluth pointed to the long timeline required to develop life-saving treatments. 

As one example, Kornbluth pointed to diabetes treatments. 

“[Treatments] started with injections of insulin saving people and now it’s automated pumps and CGMs [Continuous Glucose Monitors],” said Kornbluth. “The next phase is going to be an actual functional cure, which is stem cell implantation—masking the cells so they’re not rejected by the immune system. But it takes a lot of basic work to be able to get there.”

“That [diabetes] is just one area. You can extrapolate that to cancer therapy,” said Kornbluth. 

Investment in basic research can advance treatments such as immunotherapy. 

“Immunotherapy is just in its infancy—it doesn’t work in every possible kind of cancer at this point. But all of the modifications that are being done now in basic science laboratories through to pharmaceutical companies and biotech are making it more and more broadly applicable so that pancreatic cancer is not absolutely a death sentence now,” Kornbluth emphasized.

National impact

Beyond research and AI, the president concluded by highlighting the strength of MIT’s student body, programs, and spinouts. 

Kornbluth underscored the value of an MIT education for students and the greater economy. 

Twenty percent of MIT’s class of 2029 were first-generation students. Education“is the best pathway to economic mobility,” said Kornbluth. 

She continued: “MIT has spun out north of 30,000 companies. The economic impact of MIT on this country is equivalent to the 14th largest GDP in the world. We are having a huge impact on the economy and we’re producing the next generation of talent.”

Though MIT is highly selective, Kornbluth noted it is financially accessible through its free tuition program for students with parental incomes under $200,000. She further highlighted MIT for America, an initiative expanding access to calculus, a required course for institutions such as MIT, in under-resourced high schools nationwide.

Kornbluth and Crow concluded the panel by highlighting how their respective universities learn from one another.

“What we [ASU] learn from MIT is, where’s the edge of technology,” said Crow. “We learn how master technologists, and master scientists work in small groups.” For ASU, which has a student population of over 150,000, “ it’s instructive to learn and then operate at a different scale and in a different way. There’s a lot of back and forth,” he said.

Kornbluth expressed her hope for MIT to continue its longstanding tradition of research and education in service of the nation’s next 250 years.

“As a smaller private institution, we’re putting a much stronger footprint in how we can impact people well beyond the MIT walls,” said Kornbluth, “as well as having a scientific impact on society through our discoveries.” 

After more than 10 years at the helm of the Laboratory for Nuclear Science (LNS), Boleslaw “Bolek” Wyslouch will step down to continue research in nuclear physics as director of the Bates Research and Engineering Center, a subgroup of LNS.

“LNS scientists, including Bolek himself, are world leaders in particle and nuclear physics,” says Nergis Mavalvala, dean of the MIT School of Science and the Curtis and Kathleen Marble Professor of Astrophysics. “Bolek has ensured that LNS has flourished during his time as director, supporting our teams’ critical large-scale, international, collaborative research.”

The largest university-based program of its kind in the country, LNS was established in 1946 to provide support for basic research in the fields of nuclear and high-energy physics. Wyslouch has served as LNS director since 2015.

Since Bolek’s appointment as LNS director in 2015, he has helped significantly increase the Laboratory’s research volume. This growth reflects expansion across many areas of nuclear and particle physics, with LNS supporting several new faculty members. His vision was instrumental in bringing low-energy nuclear physics into the laboratory as a major new research area, the only subfield of nuclear physics in which the laboratory had not previously engaged.

“The leadership to inspire this capacity growth brought in young and vibrant faculty research groups, which helped lead to the expansion in LNS research volume,” says Rick Peterson, executive director of the lab. “Further, this new technical expertise facilitated new partnerships across the national laboratories, enabling LNS to develop and build a presence at all U.S.-based nuclear physics labs.” Most recently, LNS is engaged in an effort to compete for bids to the Department of Energy’s Genesis mission, a potential source of funding in the AI era. 

During his tenure, LNS saw the successful bid for the National Science Foundation-funded AI Institute for Artificial Intelligence and Fundamental Interactions, led by LNS scientists and supporting more than 25 physics and AI senior researchers at MIT and Harvard, Northeastern, and Tufts universities. Last year, the Center for Theoretical Physics (CTP), part of LNS, also received a $20 million donation from the Leinweber Foundation to create a Leinweber Institute within CTP.

“Perhaps most importantly, Bolek led LNS toward a culture where each individual is valued for their own contributions, regardless of their status within a lab group,” says Peterson, adding that he developed new pathways for postdoc support and sponsored other community-building activities. 

At Bates, Bolek has led and overseen a wide range of complex engineering and scientific projects. These include the development of advanced particle detectors for major international research facilities such as CERN, Brookhaven National Laboratory, and Jefferson Lab. Under his leadership, the laboratory established collaborations with industry partners on innovative technologies, including next-generation batteries, advanced accelerator systems, and medical applications of nuclear science. Through these efforts, the laboratory is helping advance both fundamental research and the development of technologies with broad scientific and societal impact.

In his own research, Wyslouch is one of the founders and leaders of the relativistic heavy ion program in the Compact Muon Solenoid (CMS) experiment at the Large Hadron Collider (LHC) at CERN in Geneva.

Wyslouch studies the interactions between subatomic particles by looking at the very energetic collisions of heavy ions. The earliest runs of the LHC showed that hot plasma strongly suppressed production of high-energy jets, redistributing the jet energy among slow particles. Wyslouch’s CMS group further discovered surprisingly strong collective effects in ion-ion collisions, as well as in proton-proton and proton-ion collisions.

Before joining CMS, Wyslouch conducted high-energy and nuclear experiments at CERN and at the Brookhaven National Laboratory Relativistic Heavy Ion Collider facility, and took a leadership role at Brookhaven in creating PHOBOS, a project designed to create and study a quark-gluon plasma.

After completing his undergraduate work in physics at the University of Warsaw, Poland, in 1981, Wyslouch began his association with MIT as a doctoral student, earning a PhD in physics in 1987. After postdoctoral appointments at LNS and CERN, he joined the MIT faculty in the Department of Physics in 1991. He has also served as the head of the Nuclear and Particle Physics Division of the Department of Physics since 2013. 

Wyslouch was recognized for his contribution to education at MIT with a 2004 William W. Buechner Teaching Prize. He was elected as a fellow of the American Physical Society in 2013, and as a member of the American Academy of Arts and Sciences in 2024.

For people at high risk of developing breast cancer, yearly mammograms may not be enough to detect tumors early. To make earlier diagnosis easier, an MIT team has developed portable detectors based on ultrasound, which could be used much more frequently.

In a new paper, the team reports that they have improved the resolution of the images produced by their system, making it easier to spot potential tumors, as well as cysts and microcalcifications. The researchers also created a user interface that makes it simple to use the ultrasound probe, even for people with no expertise in ultrasonography.

This system, they believe, could not only enable earlier detection, but also allow for long-term monitoring following breast cancer treatment — either in a doctor’s office or at home.

“At each time interval, the computer interface guides you to position the device in exactly the same location, which is important for the longitudinal monitoring of a given tissue. It’s very intuitive and quite easy to use,” says Canan Dagdeviren, an associate professor of media arts and sciences at MIT and the senior author of the study.

Former MIT postdoc Md Osman Goni Nayeem and MIT graduate students Shrihari Viswanath and Hyeokjun Yoon are the lead authors of the paper, which appears today in Nature Communications.

Higher-quality imaging

While many people receive annual mammograms to check for breast cancer, it is possible for cancer to develop in between these annual screenings. These cancers, known as interval cancers, tend to be more aggressive, and they account for 20 to 30 percent of all breast cancer cases.

After losing an aunt to an interval breast cancer in 2015, Dagdeviren was motivated to develop a screening technique that would be more effective on women with dense breast tissue and could be performed more often than mammography. She decided on ultrasound, which uses sound waves to create images of tissue. Ultrasounds are often used to follow up on abnormal mammograms, but current ultrasound technology requires large equipment and a trained operator.

Earlier this year, Dagdeviren’s lab published a study in which they demonstrated a small ultrasound probe attached to an acquisition and processing module that is a little larger than a smartphone. This compact system can create a 3D image of the entire breast by scanning just two or three locations.

In the new Nature Communications study, the researchers reported several advances that allow for higher resolution imaging and greater ease of use.

One key advance is the addition of a “backing layer” to the ultrasound transducer. This layer helps to contain and focus the ultrasound waves, improving the resolution and quality of the resulting images. It also increases the range of soundwave frequencies that can be absorbed, and reduces both acoustical noise and electrical noise, further enhancing the images.

“With the backing layer, the device produces more accurate and sharper images, with a wider operating range of frequencies,” Nayeem says.

To further improve the quality of the images, the researchers designed an algorithm that adaptively performs a process called beamforming. This algorithm allows the system to compensate for differences in the speed at which sound waves travel through different types of tissue, such as skin and fat. 

“What we are trying to do is predict the speed of sound properties of the tissue you’re imaging, and then use that to reconstruct the image more accurately. We see up to a 10 percent improvement in the resolution just by applying this technique,” Viswanath says. 

The researchers asked 10 volunteers, who were not experts in ultrasound technology, to use the system to try to identify small micro targets embedded in a “tissue phantom” — a gel-like material engineered to mimic human tissue. Participants had a much higher success rate locating the spheres when they used the new system than when they used a traditional ultrasound probe.

A user-friendly system

For the new version of this system, the researchers also created a user interface, displayed on a computer screen, that guides the user to place the probe in the correct location. This could be especially important for tracking progression of treatments such as neoadjuvent therapy, or long-term monitoring of known abnormalities such as fibroadenomas or microcalcifications.

In a trial with seven people, the researchers found that the users were able to accurately place the probe in the correct location each time they did a scan. 

“Conventionally, you need an operator to move the probe around the breast, but we made a computer-vision interface for users to do it by themselves. This is very user-friendly and it shows live images on the screen,” Yoon says.

For future versions of this technology, the researchers hope to create an interface that could be used with a cellphone or tablet, making the system easier to carry. In addition to enabling earlier diagnosis, this type of system could make ultrasound more accessible to patients in areas where there aren’t enough trained ultrasound technicians, the researchers say.

Dagdeviren and some of her students now hope to form a company to work toward making the technology commercially available. While breast cancer diagnosis is their first target application, they hope to expand it to many others.

“The technology is so versatile that it can be used for any soft tissue imaging, from ovarian cancer to measuring endometriosis progression, or fetal monitoring,” Dagdeviren says.

The research was funded by the National Science Foundation, the 3M Non-Tenured Faculty Award, the Lyda Hill Foundation, the MIT Media Lab Consortium, and a Tata Center Technology and Design Fellowship.

A new big-data analysis of the U.S. pinpoints how urban design aids the health of city residents — especially when cities provide walking opportunities, greenery, and mixed-use streets with a blend of commercial and residential activity. 

The study examines tens of thousands of urban census-bureau tracts in the U.S., seeing how city features correlate with population health measures, while accounting for socioeconomic considerations as well. 

“We found that on a very large scale, urban planning and design, such as the availability of different amenities and their spatial arrangement, plays a critical role in population health outomes,” says Winston Yap, a visiting scholar at the MIT Senseable City Lab, a postdoc at Cornell University, and co-author of a new paper outlining the study’s findings. 

While there is not one design template for all locations, short and well-connected blocks with a variety of amenities, as well as the strategic placement of parks, all help well-being — physiologically and psychologically. 

“We usually think about physical health first, but we also found a high correlation between good design and mental health,” says Fabio Duarte, an MIT researcher and co-author of the paper. “If you are walking more, it is not only a matter of physical fitness, but gives people a chance to avoid isolation, have serendipitous meetings with people, and at least see there are others around.”

The paper, “Urban motifs associated with population health,” appears today in Nature Health. The authors are Yap; Duarte, who is associate director and a principal research scientist at MIT Senseable City Lab; postdocs Yu Zheng, Kee Moon Zhang, and Peng Luo, who is also an incoming assistant professor at the University of Iowa; Paolo Vineis, a professor at Imperial College, London; Carlo Ratti, director of the MIT Senseable City Lab; and Filip Biljecki, an associate professor at the National University of Singapore.

Only connect

The researchers say they conducted the analysis not just due to an interest in cities, but out of recognition that health care systems are often swamped, and preventative health measures are ever-more important. 

“We wanted to do this study because health care systems around the world are overloaded,” Yan says. “There’s a lot of burden on health care systems, and there is a need not just for treatment but for prevention as well, for obesity, high cholesterol, depression and other mental health issues, and more.” 

To conduct the study, the researchers analyzed 28,323 census tracts, using data from the U.S. Census Bureau along with health data from the U.S. Center for Disease Control and Prevention (CDC). They then used geospatial data, including more than 8 million street view images, to see how urban form related to the health status of residents in those areas. The study accounts for socioeconomic factors and other variables in building an assessment of the relationship between design and health. The study confimed that by themselves, socioeconomic factors are associated with urban health disparities; it then examined the relative impact of differences in urban design in those different settings. 

“By bringing together open demographic, health, and environmental data, the study highlights the importance of open data accessibility for planning healthy cities,” says Ratti.

The scholars also applied a graph deep-learning model to the data, an emerging machine-learning technique they used to help understand which key factors in urban design are most connected to health outcomes. 

The research reveals that in some cases, rectangularity in city blocks, and “building spread,” meaning structures that cover the full size of their lots, can enhance wellness. Examples of this include Manhattan or Boston’s Back Bay neighborhood, where mixed-use buildings on relatively short blocks create many amenities and a variety of walking routes. That said, circular and curving street forms can also work, as long as they feature a lot of interconnectedness as well. 

Urban greenery is almost always a significant factor in urban wellness, with parks scoring high as a facet of city design that helps resident health. Beyond that, expanding the tree canopy can also help urban health outcomes. 

The presence of cultural institutions and restaurants are also linked to general health, while access to health care amenities are understandably connected to physical health improvements. In general, access to points of interest, broadly defined, whether cultural or commercial, is a significant factor in abetting better health, in cities across the country. 

“One of the major contributions of the study is that we look at not only one or two cities, but the entire United States,” Yap says. “In a large-scale study, we were trying to find patterns that were consistent across different urban contexts, as well as populations with different characteristics. Just using this data, we can predict very confidently the population health outcomes for a neighborhood.”

Knowing where to intervene

The research also provides a kind of road map for urban planners and city officials when it comes to policy decisions and local improvements. Among other things, the study suggests where cities might see the greatest return on investment in urban improvements, in health terms. Improvements in lower-income neighborhoods, on aggregate, may generate about four times the added health benefits than the same level of investment in better-off areas that already realize the benefits of good urban amenities. 

“It’s important to know where to intervene,” Yan says. 

“I think for me it shows how intertwined different policies are,” Duarte adds. “Some funding for urban development could have a direct influence on health, and could be more inexpensive than [direct spending on health].”

The researchers regard the study as just one empirical step in this domain. As they note, additional studies could observe changes over time, to further enhance our picture of the connection between urban design and health. Still, as the authors write in the paper, “we believe that our broad picture provides an overarching scaffolding for the understanding of the social and material determinants of health and can guide [further] analytical studies.” 

The research received support from the Campus for Research Excellence and Technological Enterprise (CREATE) program of the National Research Foundation Singapore; the Singapore-MIT Alliance for Research and Technology (SMART); and the MIT Senseable City Lab consortium. It is part of the Largescale 3D Geospatial Data for Urban Analytics project, supported by the National University of Singapore.

For decades, neuroscientists have known that specific regions in the brain’s left hemisphere are responsible for processing language. However, a new study by MIT researchers shows that language processing also occurs in many other parts of the brain.

Using functional magnetic resonance imaging (fMRI) data from more than 700 people, the researchers identified 17 additional regions of the brain that appear to play a role in language. These regions are scattered across the brain, including parts of the cerebellum, hippocampus, and cerebral cortex, and they make up about 5 percent of the total volume of the adult brain — about the size of a large strawberry.

“Even though there are all these distant components, it’s pretty restricted in terms of volume. You don’t need that much of the brain to do language,” says Evelina Fedorenko, an MIT associate professor of brain and cognitive sciences, a member of MIT’s McGovern Institute for Brain Research,and the senior author of the study.

Exactly how these regions contribute to language processing is still to be discovered, although the researchers have made some progress toward determining the functions of the cerebellar regions that they identified.

MIT postdoc Agata Wolna is the lead author of the paper, which appears in the Journal of Neuroscience. Other authors include Aaron Wright, a K. Lisa Yang Post-Baccalaureate Research Scholar at MIT; Colton Casto, a graduate student at Harvard University; Samuel Hutchinson, a graduate student at MIT; and Benjamin Lipkin PhD ’26.

Tracking language

The brain’s language processing centers include Broca’s area, first discovered in the 1800s, plus additional regions in the left frontal and temporal lobes of the brain. Scientists have found that some of the corresponding areas of the right hemisphere also contribute to processing language, especially the social-emotional components of language.

There have also been hints that other parts of the brain might be involved in language processing. Early in her career, Fedorenko’s language studies often showed active brain regions outside of the canonical language centers, but she says she was discouraged from including them in her papers.

“When we initially started looking at language, in the first couple of papers, I tried to be comprehensive and include anything that seemed consistent across participants, and there was a huge amount of resistance,” she says. “People would say things like, ‘Well, we know those are not language areas, so please focus on the language areas.’”

In the new study, she and Wolna wanted to revisit those brain scans and see if they could systematically identify language regions outside of the standard language-processing areas.

To do that, they analyzed data from 772 people who had been scanned in Fedorenko’s lab since 2013. Each of these participants underwent a task known as a language localizer, which is used to determine the location of language processing areas for each subject. 

During the test, participants read or listen to sentences as well as sequences of nonwords. For each person, the researchers measure the difference in strength of response when reading real sentences or nonsense sequences. The brain areas that work harder during the sentence condition are considered to be doing something relevant to language, especially if they respond while both reading and listening to sentences.

“It’s a very simple paradigm that lets you identify this core language system in individual brains,” Wolna says.

When searching for language areas, the researchers usually use a relatively strict statistical threshold. In this study, they relaxed the threshold and also used some targeted searches in subcortical areas, in hopes of finding all areas that may contribute to language processing.“We always see this frontal temporal network, but there’s quite a lot of evidence that there are other regions that are also critical for language processing,” Wolna says. “By using a laxer threshold and zooming in on areas with weak MRI signal, we tried to maximize the chances of finding small and weakly responsive regions outside of this left frontal temporal system.”

A widespread network

For about 490 of the participants, the researchers also had data on how their brain responded during a spatial working memory task — remembering the locations of flashing squares on a grid. This task engages a brain network called the multiple demand system, which does not overlap with the core language areas.

This task allowed the researchers to ask whether any of the newly identified language-sensitive regions specifically respond to language and not more general cognitive processes.

Of the 17 new language sites that were revealed by this study, five are located in the cerebellum, which is mainly involved in coordinating the body’s movement. In a study published earlier this year, researchers led by Casto found that three of those cerebellar regions also became engaged during some nonlinguistic cognitive tasks, which was also seen in the new study.

“Those areas that respond to both language and some other tasks could be really interesting and important because they may be doing something like integrating information from different cortical systems,” Fedorenko says.

They also found language-selective regions in the medial frontal cortex, the bottom surface of the left temporal lobe, the hippocampus, and the amygdala. The researchers now plan to further study how these brain regions might contribute to language processing.

“We can now test some ideas from past work, and also more rigorously characterize these regions across different kinds of language manipulations, and different kinds of nonlinguistic tasks, to try to understand what it is that they’re doing,” Fedorenko says.

The research was funded by the Simons Center for the Social Brain at MIT, the McGovern Institute, MIT’s Department of Brain and Cognitive Sciences, and the MIT Siegel Family Quest for Intelligence.

As a new academic year dawns, the MIT-Kalaniyot program is welcoming its second cohort of scholars to campus, expanding an innovative effort to build new connections between MIT and researchers from Israel. 

In fall 2026, MIT-Kalaniyot has 11 new scholars arriving at MIT to pursue research, collaborating with Institute faculty across a wide variety of disciplines. They consist of seven new Kalaniyot Postdoctoral Fellows and four new Kalaniyot Sabbatical Scholars, who are faculty on leave from institutions in Israel. 

It is another step forward for a program which, less than two years ago was still an idea on a drawing board. The project aims to enhance research and create stronger community ties — not only among those connected to the program, but across the MIT campus.

“The goals of the program are to build academic ties between MIT and Israel, alongside a strong, supportive community,” says Or Hen, an MIT nuclear physicist and a co-founder of MIT-Kalaniyot. “MIT has a mission that revolves around research, education, and entrepreneurship, and MIT-Kalaniyot strengthens MIT, to help meet that mission for the world.”

The scholars will be working on a wide range of topics, including mathematics, materials science, behavioral economics, architecture, modern history, chemistry, quantum computing, and computational methods for examining cellular activity.

“We designed Kalaniyot to strengthen MIT’s research and its community at the same time,” says Ernest Fraenkel, a professor of biological engineering and a co-founder of  MIT-Kalaniyot. “We now have scholars in the program working in each of MIT’s five schools. The academic breadth shows our model is working.” MIT-Kalaniyot will also feature its first teaching fellow at the Institute, hosted by MIT’s History program. 

MIT-Kalaniyot was founded by Hen and Fraenkel as a constructive response to discord over conflict in the Middle East. Hen is the Class of 1956 Associate Professor of Physics and associate director of the Laboratory for Nuclear Science; Fraenkel is the Grover M. Hermann Professor in Health Sciences and Technology.

Fraenkel and Hen credit multiple members of MIT’s community and upper administration for backing the MIT-Kalaniyot idea from the start, making it feasible for the program to launch. 

“When we first shared the idea, we were very encouraged by the response from MIT’s senior leadership,” Fraenkel says. “They understood the value of a faculty-led effort, and their constructive response gave us confidence that our approach could be successful.”

“This would be impossible to do the way we’re doing it without the administration’s support,” Hen says. “The program is faculty-led and institution-backed. That’s what you want.”

Hen adds: “I think MIT today is home to one of the most, if not the most, accepting and welcoming communities for Israelis, and I can stand by that statement very strongly. The way our community grew these past years is remarkable.”

Embedded at MIT

MIT-Kalaniyot, named for a well-known flower that grows in Israel and other parts of the region, welcomed its first cohort of scholars to the MIT campus for the 2025-26 academic year. Hen and Fraenkel also give Tal Cohen, an associate professor in MIT’s Department of Civil and Environmental Engineering, substantial credit for developing the concept. 

Scholars at Israel’s nine state-recognized universities are eligible to seek the MIT-Kalaniyot fellowships, which enable research, collaboration, and training at the Institute. The scholars come from a range of academic and personal backgrounds, including both Arab and Jewish citizens of Israel. 

The program is highly competitive, with many more applicants than positions currently available. Applicants are encouraged to identify in advance MIT faculty they would like to work with; accepted applicants then already have a “faculty host” lined up. Many of the new fellows will be working with researchers in established MIT labs, for instance. 

“When they’re here, they are treated exactly like anybody else in an academic unit at MIT and that’s really important,” Fraenkel says. “They’re embedded in these places.”

The program is also intended to generate the kinds of community connections that help scholars flourish, both professionally and personally. MIT-Kalaniyot features weekly lunches, attended by people from the larger community, where scholars can forge connections and friendship. 

The program also features informal academic talks and discussions, with the talks given by MIT researchers both within and outside of MIT-Kalaniyot. Hen, for one, has already seen the benefits of such events; one paper he has recently co-authored directly stemmed from discussions he had at a program event. 

“The range of MIT faculty who stepped forward as hosts has been one of the most gratifying parts of the program,” Fraenkel says. “It shows that this is not confined to one field or one corner of the Institute. It is becoming part of MIT’s broader academic life.”

Adds Hen: “I think it sends a very strong and important message. We’re able to move forward at MIT and build collaborative partnerships with strong ties.”

An additional facet of the program is the potential impact of MIT-based research in practical, tangible ways. One of the 2025 fellows, a leading physician, focused her MIT work on new methods of breast cancer detection, and now, back in Israel, is working to apply those findings in active medical settings. 

Plans for future growth

Having first taken root at MIT, the MIT-Kalaniyot concept is now spreading to other places. In the last two years, Columbia University, Cornell University, Dartmouth College, Harvard University, the University of Pennsylvania, and the University of Southern California have implemented the concept, with other universities in the process of adopting it as well. 

“This national movement all started by replicating the MIT model,” Hen says. “Each university then innovated in their own way. They start from the MIT approach, and then they adapt to what’s happening on their campus. They learn from us, we learn from them, and together we support a broad academic network.”

The progress at MIT and elsewhere has led Hen and Fraenkel to feel optimistic about the ongoing evolution of MIT-Kalaniyot. 

“We started at a tense time on our campus, not really knowing what the future would hold, and it’s exceeded our hopes,” Fraenkel says. “Now we want Kalaniyot to become a recognized center at MIT, funding seed grants for research that wouldn’t happen any other way.”  

While Fraenkel and Hen do not yet have a firm timetable for those developments, they regard them as being realistic. 

“Now we see Kalaniyot as a program that helps MIT well beyond our community,” Hen says. After all, he observes, simply as a vehicle for research, the program has the potential to provide added capacity for MIT, as well as the further connections to top scholars being generated by the effort. 

Indeed, Hen reflects, he is motivated the question: “How do we best support MIT in realizing its mission for the world?” Overall, he says, “I think that’s the ultimate goal of Kalaniyot. We do it in one way, other people can do it in other ways, and as long as you do net good, and support the MIT mission, we value and treasure that, and just want to be part of it.”

“I really believe this is the DNA of MIT,” Fraenkel says. “We’re all about finding practical solutions to society’s biggest problems. Kalaniyot brings extraordinary people here to do exactly that, and the whole Institute is stronger for it.”

Three teams comprising 35 students across eight different MIT departments and Wellesley College have been at work since fall 2025, designing critical early infrastructure elements that a moon base would require. This June, their designs were recognized with five awards at NASA’s 2026 Revolutionary Aerospace Systems Concepts — Academic Linkage (RASC-AL) Forum. 

Among 75 submissions and 14 finalists, the MIT teams earned first and second place in the competition, as well as three best-in-theme awards. The Exploration-Class Lunar Integrated Power SystEm (ECLIPSE) team won first place overall and first in its theme category, lunar surface power. The communications and navigation constellation team, MELIORA, won second place overall and first in its theme category on Mars communications, position navigation and timing, which included a strategy for proving the design at the moon. And CHEESEBURGER, a campaign to mine and process lunar regolith into oxygen, metals, and bricks, won first in its theme category, lunar technology demonstrations. 

“NASA spent the spring telling the world what critical early infrastructure their upcoming permanent moon base will need,” says George Lordos, a research scientist and lecturer in the Department of Aeronautics and Astronautics (AeroAstro) and in System Design and Management (SDM), who co-advised all three teams. “Over 30 MIT students spent this academic year designing much of the moon base — systems for generating, storing, and distributing power; robust systems for positioning, navigating, and communicating; and early experiments with essential technologies to live sustainably off the moon’s own dirt.”

A power grid for surviving lunar night and winter

The hardest constraint on NASA’s moon base is staying powered, because a failure in life-support power would doom the crew within hours. ECLIPSE is a reference design for a lunar grid engineered to stay up for more than 99.995 percent of the time — fewer than 27 minutes of downtime a year in the worst-case scenario, the standard demanded of the most critical data centers on Earth. It pairs two power sources that fail in different ways: banks of 20-meter solar masts in the sunlit highlands near the south pole, and, for the roughly 18-day stretch each year when the sun drops below the horizon, a pair of buried 20 kilowatt microreactors the team named CARROT, (Compact Autonomous Regolith-shielded Reactor Operating for Ten years). The CARROT reactor, a novel design developed independently by the ECLIPSE team, ended up being similar in design to NASA’s SR-1 reactor for the 2028 mission to Mars, both aiming to maximize speed-to-deployment. 

“Burying each reactor 1.3 meters down shrinks the keep-out zone from kilometers to meters, so crews can work nearby, and it saves tons on required shielding mass,” says Taylor Hampson, a PhD student in the Department of Nuclear Science and Engineering and ECLIPSE team co-lead.

The full design delivers an initial 120 kilowatts using a grid of buried aluminum cables and shielded direct-current power equipment. Laser-equipped rovers provide “Frontier Power” capability, beaming up to 10 kilowatts to sites beyond any cable, from a shadowed crater to a new outpost before its own grid exists. Patrick Riley, a graduate student in the Department of AeroAstro and ECLIPSE team co-lead, says the design’s point is to put reliability ahead of mass: “We sized it so the most likely failures never reach the moon base inhabitants, and so it scales from a first crew of six up to industrial demand without interrupting a commercial lunar economy.”

A network for exploring the moon and Mars, and calling home

MELIORA acts as the base’s relay and GPS. Although RASC-AL framed the communications, positioning, navigation, and timing competition sub-theme around Mars, the team also proposed a plan to validate their design in lunar geometry first, in step with the agency’s strategy to prove technology on the moon before extending it to Mars. To find the best design, the team ran a trade study across 5,764 candidate constellation geometries. The result grows from an initial three satellites to 23, returns more than 100 megabits per second to Earth-orbiting data networks over free-space optical links, and pins a user’s position to within 10 meters. For the Mars design, four relay satellites parked at gravitationally stable Lagrange points keep the link alive even during solar conjunction, the weeks when the sun sits between the two worlds and ordinarily cuts communication. On the surface, a user needs only a portable radio terminal and a chip-scale atomic clock — a timekeeper the size of a matchbox. 

“You should never have to think about whether the network is there — it just is, the way you don’t think about a cell tower,” says Ekaterina Tiukhtikova, an undergraduate studying both AeroAstro and electrical engineering and computer science (EECS), and a MELIORA team co-lead. “We put almost all the complexity up in orbit, so everything on the surface stays portable and simple,” adds Clayton Lieberman, a graduate of the SDM program and team co-lead who wrote his thesis on MELIORA.

Making oxygen, metal, and bricks from lunar dirt

After power and communications, the third essential pillar of a lunar base is living off the land. The moon’s own regolith can supply oxygen to breathe and burn, metal to build with, and shielding to hide behind for protection from deadly radiation. CHEESEBURGER is a campaign of five robotic payloads that prove the supply chain one link at a time, followed by integration of the five into the first end-to-end lunar industry. 

The payloads carry a kitchen’s worth of names: SWISS prospects for the richest ore, BRIOCHES digs and sorts the regolith, BACON casts it into bricks, GRILLED MEAT melts it electrically to pull out metal and oxygen, and AVOCADO is the robotic builder that stacks the products into structures, including interlocking Moon BRICCSS that shield a habitat from radiation. The food theme was born during a January team outing at Sandwich, Massachusetts. “Naming the prospector SWISS and the metal extractor GRILLED MEAT turned a wall of acronyms into something the whole team could enjoy,” says Cesar Meza, a graduate student in AeroAstro and CHEESEBURGER co-lead. “It sounds like a joke until you see that each acronym clearly describes a serious piece of hardware doing one job in the pipeline.”

Thirty students, eight departments, and three teams for one moon base

More than 30 students contributed across the teams, from AeroAstro, SDM, Nuclear Science and Engineering (NSE), EECS, Mechanical Engineering (MechE), the Technology and Policy Program, the MIT Sloan School of Management, and Earth, Atmospheric and Planetary Sciences (EAPS), along with a student from Wellesley College. Several student mentors and faculty advisors worked across more than one team, which is why ECLIPSE’s grid is sized to power CHEESEBURGER’s processing, CHEESEBURGER’s regolith handling is used to bury and shield ECLIPSE’s grid, and all three projects are designed to translate moon base lessons for a future mission to Mars. The teams were advised by Olivier de Weck, the Apollo Program Professor of Astronautics and Engineering Systems and interim department head of AeroAstro, who led ECLIPSE; Kerri Cahoy, the Sheila Evans Widnall Professor of Aerospace Engineering, who led MELIORA; Jeffrey Hoffman, professor of the practice in AeroAstro and a former NASA astronaut, who led CHEESEBURGER; Koroush Shirvan, Atlantic Richfield Career Development Professor in Energy Studies in Nuclear Science and Engineering, who co-advised ECLIPSE; and Lordos, who co-advised all three. Much of the day-to-day mentorship work is led by PhD student volunteers and runs through the MIT Space Resources Workshop, which Lordos founded in 2019.

“The winning teams demonstrated how academic innovation can support Artemis mission goals,” says Daniel Mazanek, RASC-AL program sponsor and senior space systems engineer at NASA’s Langley Research Center, in NASA's announcement of the awards. “Their work highlights the important role student research plays in shaping future space exploration.”

NASA expects astronauts living on the lunar surface for months at a time by the early 2030s — the window ECLIPSE, MELIORA, and CHEESEBURGER were designed for. The picture the three teams had worked toward is unified: a crew at the lunar south pole, the lights on through the winter night, the network always up, and the first oxygen and bricks coming out of the ground beneath them. 

“A permanent base is no longer a slide in a strategy deck; NASA begins landing the first elements in 2027,” says de Weck. “Studies like these three let the agency see, before the concrete sets, how its power, communications, and resource choices depend on one another. That is precisely when independent, integrated architecture work has the most influence on the real plan.”

RASC-AL is administered by the National Institute of Aerospace on behalf of NASA. MIT has a long record in NASA’s student design competitions, with recent winning teams including the  HYDRATION Mars water production system, the Pale Red Dot Mars homesteading architecture, the deployable lunar tower MELLTT, the MARTEMIS lunar Mars analog campaign, the MAPLE autonomous lunar robot pathfinding system, the CERBERUZ lunar recycling project, and the THERMOS cryogenic fluid management system. This work was supported in part by NASA, the Massachusetts Space Grant, MIT AeroAstro, and the MIT Space Resources Workshop. One student was supported by a NASA Space Technology Graduate Research Opportunity Fellowship.

The full teams:

ECLIPSE — Team leads: Taylor Hampson (graduate student, Nuclear Science and Engineering) and Patrick Riley (graduate student, AeroAstro). Reactor team: Liliana Arias, Sydney Menne, Julian Rocher and Pavel Shilenko (graduate students, NSE). Power management and distribution team: Evrard Constant and Mary Foxen (graduate students, AeroAstro), Janhavi Joglekar and Asma Patel (undergraduate students, AeroAstro). Solar and architecture team: Zachary Dawson (graduate student, System Design and Management), Sreeja Akula and Ian Jimenez (undergraduate students, AeroAstro; EAPS), Yohan Lim (graduate student, AeroAstro/Technology and Policy Program), CJ Taglienti (graduate student, AeroAstro/MBA). Student co-advisors: Yana Charoenboonvivat, Lanie McKinney (AeroAstro), Palak Patel (MechE). Industry mentor: Sully Marigliano-Crevecoeur (Technetics). Faculty: Olivier de Weck (lead) and Jeffrey Hoffman (AeroAstro), George Lordos (AeroAstro and SDM), and Koroush Shirvan (NSE).

MELIORA — Team leads: Clayton Lieberman and Katiyayni Balachandran (System Design and Management), Ekaterina Tiukhtikova (undergraduate, AeroAstro and EECS), Celvi Lisy (AeroAstro). Team members: Thomas Harrington and Zachary T. Barnes (SDM), Asael Acosta (undergraduate, AeroAstro). Student co-advisor: Lanie McKinnery (AeroAstro). Faculty: Kerri Cahoy (lead), Jeffrey Hoffman and Olivier de Weck (AeroAstro), and George Lordos (AeroAstro and SDM).

CHEESEBURGER — Team leads: Cesar Meza (graduate student, AeroAstro) and Elizabeth Romero (undergraduate, AeroAstro). Team members: Rachel Dunphy, Shreya Kothnur, Hailey Polson (undergraduates, AeroAstro), Christopher Kwon, Jose Soto, Lanie McKinney (graduate students, AeroAstro), Marvin Martinez (undergraduate, MechE), Ananda Santos Figueiredo (graduate student, Technology and Policy Program), Evangeline Haiqi Wang (undergraduate, Computer Science and Psychology, Wellesley College). Faculty: Jeffrey Hoffman (lead) and Olivier de Weck (AeroAstro), and George Lordos (AeroAstro and SDM).

An MIT-led research program aimed at creating future microsystems capable of sustainably transmitting data with greater bandwidth and higher efficiency than is possible today has made several significant advances since it was established in 2022. 

These include the invention of devices within systems that can much more easily integrate electronics — manipulating data with electricity — with photonics, which does the same with light. The microsystems, the first of their kind, also promise to be cost-effective because, among other advantages, they can be manufactured using existing equipment in traditional electronics foundries and packaging houses.

“Our disruptive electronic-photonic integrated solutions will enable us to leap from [transmitting data at] hundreds of terabits per second to greater than 1 petabit per second,” said Anu Agarwal, who leads MIT’s FUTUR-IC, at an April webinar titled, “Shaping the Future of Semiconductors: Power, Performance, and Possibility.” The event was sponsored by the MIT Industrial Liaison Program and Startup Exchange.

An advanced system using co-packaged optics can provide improved bandwidth and energy savings compared to what is used today, which is electronics-only or pluggable optics.

Toward sustainability

The microchips behind everything from smartphones to medical imaging can be traced to about 500 megatons of carbon dioxide-equivalent lifetime emissions in 2021, and every year the world produces more than 50 million tons of electronic waste. Further, the huge data centers necessary for complex computations like on-demand video are growing, and will require close to 10 percent of the world’s electricity by 2030.

“This is neither scalable nor sustainable, and cannot continue,” Agarwal has reiterated over the years. FUTUR-IC, funded by the National Science Foundation Convergence Accelerator, was created to address these resource-efficiency issues.

For example, integrating photonics with the electronics that underpin today’s microchips could address energy use because the transmission, or communication of data, using light is much more energy efficient. “Our mantra is to use electronics for computation and photonics for communication to bring this energy crisis under control,” says Agarwal.

Currently, however, it is difficult and expensive to connect electronic chips with their photonic counterparts within a single package. That’s partly because the supply-chain ecosystem for co-packaged optics is still immature.

New devices

Enter two new devices developed through FUTUR-IC aimed at making it easier — and less expensive — to integrate photonic chips with microchips. One, the evanescent coupler, was featured on the cover of Advanced Engineering Materials last year. Another, known as the graded index coupler (GRIN), was reported in the March 2026 print issue of the Journal of Physics: Photonics. 

A third new coupler was developed by an MIT team led by Professor Juejun Hu of the Department of Materials Science and Engineering. It was reported in a 2023 issue of Laser & Photonics Reviews. That work was supported by the Department of Energy. 

The three couplers are the first optical equivalents of “solder bumps,” or the tiny dots of metal that allow chip-to-chip or chip-to-substrate connections for electron flow. Until this MIT work, there were no analogous “optical bump” options for photonics.

And if photonics is to be integrated with electronics, “you’ll need both metal bumps and optical bumps, because there are devices on your photonics chip that will require both an electrical signal and an optical signal,” says Drew Weninger PhD ’25, first author of the papers on both the evanescent and GRIN couplers. Weninger is now at the National Institute of Standards and Technology.

As with electronics, many options of optical bumps will be necessary, as “each type has substantial trade-offs,” wrote Weninger and colleagues in a review article in Nature about coupler advances published earlier this year.

For example, the GRIN coupler can be used over a wider spectrum of light than is possible with the evanescent coupler, Weninger says. The evanescent coupler, however, is easier to fabricate and can be packed in tighter to form a higher number of connections.

Additional advances

FUTUR-IC is organized into three dimensions: Technology (the coupler work is a good example), Value Chain Innovation, and Workforce. 

Under the Value Chain sector, researchers developed a new tool to support companies’ decisions toward sustainability. Earthster provides a visual model for quickly determining the energy, materials usage, and environmental sustainability across a company’s products. For example, says Agarwal, “looking at [Earthster], a supplier can tell right away their hot spots for carbon emissions, and start working to minimize them.”

FUTUR-IC has also developed several programs aimed at developing a future workforce for next-generation microchips. For example, “it is introducing an online course on semiconductor resource efficiency,” Agarwal says. “We also offer gamified digital learning and problem-based learning, plus a summer academy and a hands-on bootcamp.” For K-12 awareness, FUTUR-IC has created TED-Ed videos.

Agarwal concluded her April webinar by acknowledging the range of industries FUTUR-IC aims to help. “If you’re a packaging vendor, a materials vendor, or you are in the supply chain for data centers, FUTUR-IC can provide value.”

Additional authors of the paper on the GRIN coupler are Agarwal; Lionel Kimerling, the Thomas Lord Professor in the Department of Materials Science and Engineering; Christian Duessel BS ’25, now at SiLC Technologies, a silicon photonics company; and Samuel Serna, professor of physics, photonics, and optical engineering at Bridgewater State University.

Additional authors of the Nature review paper are Serna; Luigi Ranno PhD ’25, now at Ayar Labs; Kimerling; and Agarwal.

The deployment of automated software systems called AI agents has recently exploded. A November 2025 report by MIT Sloan School of Management and Boston Consulting Group found that 35 percent of surveyed businesses had already deployed AI agents, while another 44 percent planned to implement agentic AI soon. 

To understand the fundamentals and potential impacts of these increasingly popular tools, MIT News spoke with Phillip Isola, an associate professor in the Department of Electrical Engineering and Computer Science (EECS) and a member of the Computer Science and Artificial Intelligence Laboratory (CSAIL), who studies the intelligence AI agents possess, as well as the underlying models and mechanisms that power agentic AI systems.

Q: What is agentic AI and how is it different from generative AI models like ChatGPT and Claude?

A: Agentic AI is AI that takes actions in the world. These actions could be a physical action, like robotic manipulation, or a digital action, like booking a flight. On the other hand, we think of generative AI as making up stories, poems, art, and images, rather than taking actions for us. 

The word “agent” is just a brand name. It usually means AI that is going to help people interact with an application, a website, or the physical world. Most agents we encounter today are digital agents, like customer service agents you can talk with about product complaints. 

Most companies that offer agents use the same few AI models under the hood and give them the ability to take actions and remember what happened. An agent starts with a fundamental generative AI system, like Claude, at the core. Then companies put different wrappers around that foundation model for their product or application. Those wrappers might be specific tools that agent can use, and those tools depend on the application. Maybe the agent has access to a calculator so it can solve math problems, or maybe it has access to a more complicated hard drive and operating system so it can remember a firm’s financial data and past business negotiations. 

The biggest challenge in developing agentic AI comes from a lack of training data. If I want to create a system that can go online and book a flight for me, that seems pretty simple. But we don’t have a lot of data that spells out exactly how to do that — where to move the mouse, which buttons to click on, what to do if something goes wrong, or how to call somebody and negotiate about the price of the airline ticket. One way to train a system like this is to have the AI agent visit airline websites, try things out, and see what works and what doesn’t work. These environments are hard to model, so often the agent must learn by trial and error.

Q: What are some promising applications of agentic AI?

A: I think the area where we’ve seen the most success has been with coding agents. This is something that evolved from generative AI. People trained language models on code, and then they can predict what a human would do to solve a coding problem. In addition, an agent can learn to do this by going through a feedback loop where it tries out different solutions and checks to see if it got the answer right. As long as it can check the answer, the AI agent can perform this trial-and-error loop until it figures out a good strategy.

But there is always a balance between automating decision making versus simply assisting and informing humans. Analytical AI methods, like the systems that help predict possible outcomes of decisions, are not agentic in nature, but are very informative to human decision-makers. For cases that are either high-stakes or safety-critical, like medicine, security, high-level business policies, etc., the technology might not be ready for AI to completely automate those processes, or we might not even be comfortable with that.

Q: Are there risks we should be thinking about when using AI agents?

A: One big risk area comes from the fact that it is often very easy to get agents to do certain types of work for you. With coding agents, you can “vibe code” and just ask the agent to make a code for you, so you don’t have to do the hard work yourself. There is a big risk that, because it is so easy, people will not put enough effort into verifying that it is doing the right thing. Bugs will be introduced, private data will get leaked — this is already happening.

Agents aren’t perfect, in the sense that they might make mistakes because they are not well-trained and don’t know what to do. But even if they are very competent, if a human doesn’t use them appropriately or gives them an instruction that is too vague, the AI agent could make a mistake because the human made a mistake. If humans are less involved in thinking through all the consequences, I think we might be more prone to making those mistakes. 

An additional aspect is the risk of de-skilling. It is unclear how far this will go, but when we are relying on agents to do our homework, our coding, and our math, we might lose the ability to do that ourselves, and we might lose that ability too soon because the technology is not yet ready to fully automate those processes.

Q: What does the future hold for agentic AI?

A: What we think of now as agentic AI refers to large language models using tools to interact with digital and physical systems. One obvious limitation is that, under the hood, these have the architecture of a language model and are trained on text data. To make even more powerful AI agents, we might need to model videos, physical forces, time series, radar scans, and other modalities. We might need to have models with fundamentally different architectures that can handle continuous data, high-dimensional data, stochastic data, and so on. 

But, on the other hand, maybe an extremely good coding model could act as a puppeteer to interface with sensors, actuators, and web APIs? Perhaps, once you have a super-smart reasoning system that understands math, language, and code, you can give it a camera and a keyboard and it will figure out what to do in the spatial domain. Is the next wave of AI just going to be Claude with sensors, actuators, and tools, or is it going to be something built in a new way from the ground up? That’s the big question a lot of people in AI are grappling with right now.