Powering up for a pivotal moment

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

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

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

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

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

Deploying today’s technologies

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

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

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

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

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

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

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

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

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

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

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

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

Early-stage research and basic science

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

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

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

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

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

Data centers and public trust

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

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

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

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

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

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

de MIT News https://ift.tt/5YId1Gh

Regina Barzilay and Jim Collins win Pritzker Prize for AI and Science Research Excellence for work in AI-enabled antibiotic discovery

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

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

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

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

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

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

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

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

de MIT News https://ift.tt/DMHVp1i

A flexible new route to designer materials built from nanoparticles

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

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

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

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

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

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

More flexible structures

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

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

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

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

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

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

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

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

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

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

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

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

A new path toward customizable materials

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

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

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

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

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

de MIT News https://ift.tt/wSyxPtu

MIT engineers design a safer, more efficient vaccination system for farmed fish

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

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

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

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

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

Better vaccine delivery

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

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

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

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

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

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

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

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

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

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

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

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

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

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

Protecting fish

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

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

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

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

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

de MIT News https://ift.tt/WOiIwKQ

J-PAL North America launches ARTEMIS to train early-career researchers

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

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

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

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

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

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

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

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

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

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

de MIT News https://ift.tt/2uM9VPL

Q&A: Hey AI, who should I vote for this year?

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

Q: What are your findings thus far?

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

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

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

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

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

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

de MIT News https://ift.tt/yYZSuRQ

MIT engineers create nanoparticles for delivering personalized cancer vaccines

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

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

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

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

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

Viral vaccine assembly

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

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

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

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

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

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

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

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

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

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

Targeting tumors

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

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

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

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

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

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

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

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

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

de MIT News https://ift.tt/zRDq3rb