Gage Coon: An Earth scientist exploring the power of microbes

Growing up in Waverly, Tennessee, Gage Coon spent much of his childhood outside. His family had everything from chickens to horses and even an emu named Big Bird. Coon and his cousins would explore the woods surrounding their home, and his father, a mechanic, taught him how to build and repair things around the house. His mother, a secretary at the local high school’s vocational school who loves gardening and birdwatching, encouraged him to experience as much of the world around him as he could.

That hands-on upbringing, which taught Coon to appreciate the natural world and the processes that sustain it, continues to influence how he approaches science today.

Now entering his third year as a PhD student in MIT’s Department of Earth, Atmospheric and Planetary Sciences, Coon studies some of the smallest organisms on Earth: microbes. His research focuses on how microorganisms cycle carbon and sulfur through the environment and how to leverage those processes to help address climate change. Though he studies organisms too small to see with the naked eye, the experimental nature of his work — whether in the lab or on a research vessel in the open ocean — is especially satisfying.

“I think I enjoy that physicality of seeing what I’m working with, seeing its change, and being able to touch it,” Coon says.

Coon did not initially set out to study microbiology. His interest in science began with chemistry. A high school chemistry teacher and a summer program introduced him to the subject. But later, at the University of Tennessee at Knoxville, he joined a lab focused on microbial biogeochemistry and was delighted to find a field that brought together the different areas that interested him: chemistry, the environment, and the larger climate processes shaping our Earth.

The transition from rural Tennessee to Cambridge, Massachusetts, and MIT has been a significant one. As a first-generation student, he did not learn about PhD programs until several years into college.

Once he discovered academic research, however, Coon was drawn to the possibility of spending his career learning.

“I discovered this world of academia, and so I was really excited when I learned about it,” he says. “I was like, ‘Oh my god, constant learning. That is exactly what I want to do forever.’”

Coon began studying the microbes that drive carbon and sulfur cycling in marine sediments as an undergraduate, eventually joining research cruises to investigate these processes firsthand.

His first research cruise, in 2022 after his second year of college, took him to the Atlantic continental slope to study methane seeps and how microbes prevent this methane from escaping to our atmosphere. For Coon, experiencing the ocean up close changed the way he understood the microscopic organisms he was studying.

“It is very powerful seeing yourself in the middle of the ocean, with a whole other world of complex life beneath you,” he says.

At MIT, working with his advisor Tanja Bosak, a professor of geobiology, Coon has continued studying microbial carbon and sulfur cycling, but with a greater emphasis on the applications. One of his major projects explores how microbes could be used to reduce methane emissions from wastewater treatment.

When wastewater is treated, microbes break down organic material in large tanks called anaerobic digesters. One of the final products of this process is the powerful greenhouse gas methane. However, Coon and his colleagues found a way to change what the microbes produce by adding gypsum, a waste product that is created from fertilizer manufacturing

The system uses the added gypsum to turn the methane into carbonate, which can be used to make cement, agriculture, and pharmaceuticals. The process also produces elemental sulfur, necessary for global fertilizer production, which is currently sources from oil and gas refinement. The approach effectively turns two waste products, sewage and waste gypsum, into useful materials while reducing greenhouse gas emissions.

For Coon, the possibility of creating a system that is both environmentally beneficial and economically useful is central to the project. Now that the laboratory experiments have ended, the researchers are looking toward conducting pilot-scale testing. Coon and his advisors have been communicating with companies interested in adapting the system to larger facilities, and hope the technology can eventually move beyond the laboratory.

“If enough small places start doing their pilot-scale studies, then hopefully you could convince some place like Boston or another big city to do this and really make a contribution to our global goal to decrease emissions on the gigaton scale,” he says.

The wastewater project is only one part of Coon’s PhD research. He also studies geological processes that could produce molecular hydrogen, a potential carbon-free energy source. His work examines how iron-rich rocks break down and generate hydrogen underground. He is continuing his thesis work by focusing on microbial competition for acetate, and what this means for global methane emissions from coastal wetlands. This work could improve future climate predictions and support engineered mitigation efforts to decrease emissions from these wetlands. 

Across these projects, Coon is interested in the connection between the microscopic and the massive. But Coon’s PhD has also given him an opportunity to think about science beyond his own research.

One of the parts of graduate school he has enjoyed most is mentoring younger researchers. He has worked with a handful of students through MIT’s Undergraduate Research Opportunities Program and from Tufts University, teaching them laboratory techniques and experimental geobiology.

Outside the lab, Coon maintains some of the same connection to the natural world that characterized his childhood in Tennessee. He spends time hiking to explore local geology, playing bluegrass guitar, and speed-solving Rubik’s Cubes. 

Looking ahead, Coon sees himself continuing in academia, working in government, or helping to bring environmental technologies into practice.

What matters most, he says, is continuing to produce knowledge that can help people understand and potentially improve the world around them.

“I do think, no matter what,” he says, “I’ll be somewhere thinking about how microscopic life connects to the global ecosystem and carbon emissions.”

de MIT News https://ift.tt/9r8TPyw

Looking beyond natural sequences

A protein’s function is determined by its structure, and structure — the way a protein folds — is determined by its sequence of amino acids, the building blocks of proteins. 

Many methods for designing novel proteins, including examples that could bind to a disease-causing molecule in our cells, involve a two-step process: The structure comes first, and then a machine-learning framework generates a repertoire of sequences that could potentially adopt that structure. 

In nature, many different amino acid sequences can fold into the same structure. At the same time, one amino acid sequence can potentially adopt different structures depending on the protein’s flexibility or a functional trigger. Therefore, when researchers use artificial intelligence to design new proteins, the challenge is to guide AI to “see” that there are many potentially useful answers — that many sequences can adopt the same fold

“For years, the field has measured success by asking whether a model can reproduce the protein sequence that evolution happened to select — our work shows that this isn’t the best metric for protein design,” says Amy E. Keating, Department of Biology head, Jay A. Stein (1968) Professor of Biology, professor of biological engineering, and senior author of a paper recently published in PNAS. 

PottsMPNN, a new machine-learning framework developed in the Department of Biology, incorporates the physical principles that govern protein structure and stability, improving sequence generation and the ability to predict how mutations will affect a protein’s stability. In other words, the model has a better understanding of the sequence-energy landscape, meaning the relationship between the identity of each amino acid and the stability of the protein.

Adding this framework to a protein design pipeline will allow researchers to design structurally feasible proteins with sequences that don’t resemble those of any native protein. 

“If we’re thinking about a completely novel, designed structure, there would be no native sequence to compare it to,” says graduate student and lead author Foster Birnbaum. “What we actually care about is how likely the generated sequences are to fold into the desired structures, how well the model understands the sequence-energy landscape, and how well it can predict the effect of mutations on the stability of the protein.” 

Beyond the noise 

In the same way that AI has recently powered some dramatic social changes, so too has machine learning impacted the pace and breadth of fundamental biological research. Only recently has it become possible to reliably use a computational model to generate a protein structure or sequence. Perhaps the most widely used model today, however, was released in 2022. 

“For a field that’s moving as fast as machine learning in biology, that model has not been surpassed — we’ve been trying to understand why that is, and what it is about that model that makes it so useful,” Birnbaum says. 

Birnbaum was first interested in strategic applications of something researchers call “noise,” or adding variations to a protein structure during training. Noise decreases the tendency of the model to overly mimic native sequences, increasing the diversity of structures for which it’s able to generate sequences.

PottsMPNN also uses a pairwise distribution to capture interactions between amino acids. The ability to account for the physical interactions between all 20 possible sequence options at a pair of positions in the protein is a key reason that PottsMPNN more accurately models the sequence-energy landscape than other methods. 

Finally, Birnbaum says, they introduced sets of evolutionarily related sequences into training the PottsMPNN framework to teach the model how different sequences can adopt the same folded structure.

Birnbaum acknowledges that in trying to shift away from adhering to native sequences, incorporating evolutionary information is, in some ways, still a reliance on them. But PottsMPNN succeeded in demonstrating that as the model depends less and less on native sequences, structural compatibility and energy prediction, including for novel proteins, improve. 

Protein design in the age of AI

“Once we can design any protein we want, that enables us to do a potentially scary amount of biological engineering,” Birnbaum says. “It’s a difficult task, but I’m really optimistic about this century’s progress in biology.”

Birnbaum hopes that the model could be further improved and fine-tuned for a specific task, which has in the past led to better predictions, for example, on the outcome or consequence of a particular mutation. 

Ultimately, according to Keating, “Our methods move the field toward designing useful new-to-nature proteins for diverse applications while providing a stronger foundation for future advances.” 

de MIT News https://ift.tt/iqm60BM

New type of attack can slip past the defenses in your computer’s processor

Modern processors are fast, in part, because they guess. Rather than waiting to find out which way a program will branch, a chip predicts the likely path and races ahead. When the guess is right, time is saved. When it's wrong, the work is discarded, but traces of it linger. Since the Spectre vulnerability was disclosed in 2018, attackers have known how to read those traces to pull secrets out of memory they should never see.

Chipmakers and operating system developers have spent years building defenses. A new study from MIT's Computer Science and Artificial Intelligence Laboratory (CSAIL) shows that a key assumption behind many of them doesn't hold.

The defenses work by wiping or isolating the processor's prediction machinery, removing anything an attacker might have planted. The catch, as PhD student Daniël Trujillo and MIT Assistant Professor Mengjia Yan point out, is that the wipe and the moment the predictions get used can't happen at the same instant. There is always a gap — sometimes only a handful of instructions wide. Anything that runs in that gap can dirty the machinery all over again. The researchers call this class of attack "TONTOU."

Mind the gap

Their contribution is a reliable way to get code into that gap. Computers constantly pause whatever they're doing to handle interrupts: small, routine tasks triggered by timers, network traffic, and hardware. Ordinary programs can set those timers themselves. By tuning a timer with enough precision, Trujillo and Yan can make the processor take its detour at exactly the wrong moment, and the interrupt execution does the contaminating. They call the technique "interrupt injection."

The team tested four processor generations from Intel and AMD, and got mispredictions on both. On Intel chips, the attack defeated two different protections, one built in software for older parts, one built into the silicon of newer ones. Curiously, the newer protection held firm on one Intel generation and failed on another, suggesting chipmakers implement the same nominal defense in meaningfully different ways.

AMD's defense, called saferet, cleans the prediction machinery immediately before each use, leaving a vulnerable window just two instructions wide, which typically execute within tens of nanoseconds. The researchers hit it anyway, by slowing down the processor at that exact spot to make the target easier to strike.

From a bad guess to a password file

To show what this means in practice, the team built a working exploit on an AMD system running a current Linux kernel. They first stripped away a defense that scrambles where the operating system sits in memory, succeeding in all 10 tries in about nine minutes each. That helped them read protected memory at roughly five bytes per second — slow, but fast enough to locate and copy "/etc/shadow," the file storing the system's root password hash, in half their attempts.

The paper suggests cleaning the prediction machinery a second time, when the interrupt finishes. That looks workable on AMD. On Intel it may backfire: Because the attack relies on the interrupt leaving behind a consistent state rather than any particular one, the standard fix could make the attack more reliable, not less. Newer Intel chips include a dedicated instruction that appears to help.

The other option, blocking interrupts during the vulnerable window, would likely cost too much performance to be practical.

Trujillo and Yan notified AMD and Intel in early February and reached Linux kernel maintainers in March, coordinating with AMD to warn cloud providers and other downstream customers. AMD then released a patch that mitigates the attack, which can be obtained by updating your operating system. Their code is publicly available.

The research was supported, in part, by the U.S. Air Force Office of Scientific Research under an award made through the U.S. Department of War, and ACE, one of the seven centers in JUMP 2.0, a program sponsored by the U.S. Defense Advanced Research Projects Agency (DARPA). It was presented at both Black Hat USA and USENIX Security this month.

de MIT News https://ift.tt/GPv0rsg

MIT engineers create a system for building shape-changing smart devices

A new set of modular components allows users to create reconfigurable smart devices with electrical connections that keep working no matter which shape the structure forms.

This electrical modularity can enable engineers to design interactive devices that can sense which shape they have taken, without the need for external wires. For instance, the modular components, which the researchers call “bifur-circuits,” could be used to rapidly design and prototype adaptable smart devices, like assistive furniture that helps individuals change body positions while recovering from injuries or reconfigurable robotic grippers that remain electrically connected when they change shapes for different applications. 

Developed by MIT researchers, these 3D-printed building blocks, which are a type of structure known as a mechanical metamaterial, can be combined to form many more possible configurations than traditional metamaterial structures. 

In a study presenting the new system, the researchers demonstrated several interactive objects, including a chair that converts to a table with storage and can also flatten for stowing. The structure senses its configuration and sends corresponding messages to an electronic display. 

These new metamaterials could also be used to design antennas for communications and sensing that form new shapes to adjust their frequencies in changing environmental conditions, without bulky mechanical parts. 

“Metamaterials can make complex mechanical assemblies easy to manufacture just by using repeating units. Our work expands on this design space. If we think of mechanical metamaterials as building blocks, then our work is one way to take advantage of their geometry to embed intrinsic intelligence into hardware, which could open many possibilities,” says Marwa AlAlawi, a mechanical engineering graduate student and lead author of a paper on the devices.

AlAlawi is joined on the paper by co-senior authors Ticha Sethapakdi, an electrical engineering and computer science (EECS) graduate student at MIT; and Stefanie Mueller, an associate professor in MIT’s departments of EECS and Mechanical Engineering and leader of the Human-Computer Interaction Group at the Computer Science and Artificial Intelligence Lab (CSAIL). Their co-authors include others at MIT, the University of Tokyo, and the University of Michigan. The research will be presented at the ACM Symposium on User Interface Software and Technology.

Shape-changing interactive structures

Mechanical metamaterials are programmable, three-dimensional structures of repeating units that can form complex shapes due to their geometries. When squeezed, pushed, or pulled, metamaterials can bend or twist in precise ways. 

For instance, “auxetic” metamaterials get wider when stretched, instead of narrowing.

In prior work, the MIT researchers used auxetic metamaterials to build reconfigurable antennas that formed three shapes depending on how the structure was stretched. This allowed the antenna to dynamically adjust its frequency range without complex, moving parts.

Next, the team wanted to expand the number of antenna configurations but were limited because the auxetic metamaterials could only form three fixed states.

In this work they created “bifur-circuits,” which are auxetic metamaterials that can form many more shapes based on how the modular units are connected and rotated. 

The units are also designed to be electrically modular. Due to the way conductive material is integrated into the bifur-circuits, electrical connections throughout the structure are maintained no matter how the object is rotated, pressed, or twisted to form new shapes. 

To create interactive objects with many possible configurations, bifur-circuits leverage a property known as mechanical bifurcation. 

Mechanical bifurcation is a sudden change in how a mechanism behaves when a force exerted on it passes a tipping point. For instance, when you gently bend the ends of a plastic ruler, once that force reaches a critical threshold, the ruler buckles.

In bifur-circuits, this bifurcation occurs when connected blocks are rotated in certain ways around a pivot point. The property allows connected blocks to form more stable configurations than one block could on its own.

Adding more bifur-circuits to a structure exponentially increases the number of potential configurations.

“Bifurcation allow us to significantly expand on this reconfigurability space. Just adding one extra unit gives us so many more combinations out of the same structure,” says AlAlawi.

Connecting and rotating components activates a unique circuit between adjacent units. This interactivity allows the units to communicate with one another, enabling the structure to sense its configuration.   

One of the biggest challenges the researchers faced was incorporating a conductive material that was flexible enough to bend, but still offered enough efficiency in the flow of electricity.

“The conductive material was a constraint we had to work around in the design process, and it dictated how the sensing between blocks would happen,” AlAlawi says.

Once they perfected the design, the researchers tested the durability of reconfigurable structures by compressing them more than 10,000 times. The structures showed no degradation in electrical connectivity.

The researchers also developed a user-friendly construction and simulation tool to simplify the bifur-circuit design process. The software generates instructions for a multimaterial 3D printer, which can fabricate the reconfigurable objects in one pass.

They demonstrated the versatility of bifur-circuits by fabricating a chair that can sense its geometry when its shape is changed to a tea table, as well as a shape-shifting controller that will launch one of several video games based on its configuration.

Bifur-circuits could someday be used in applications like interactive rehabilitation tools, shape-changing grippers for modular soft robots, or reconfigurable shelters that could respond to changing environmental conditions after a natural disaster.

In the future, the researchers want to explore more applications for bifur-circuits. They also want to add more interactivity into the structures and investigate additional metamaterial shapes.

“Bifur-circuits are one step toward developing mechanical building blocks with integrated intelligence. It would be interesting to build on this work and come up with building blocks that allow us to create a structure with any form or shape we want, and which are structurally stable and can be actuated,” AlAlawi says. 

This work was funded, in part, by Japan’s Science and Technology Agency and the Bahrain Crown Prince International Scholarship Program.

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

MIT student leaders: Q&A with McCormick Hall co-president Sydney Baller

A Colorado native who originally planned to attend college close to home, Sydney Baller decided to come to MIT for the strong academic community. She knew she had found a new home in McCormick Hall after attending its Campus Preview Weekend (CPW) events in 2023. 

McCormick Hall opened in 1963 as MIT’s first women’s residence — a move that provided women the first real opportunity to attend the Institute in significant numbers. To support continuity of the McCormick community through its first renovation in its 63-year existence, MIT has established a dedicated McCormick lounge in the Stratton Student Center (Building W20), funded efforts to maintain dorm traditions, and more.

Now a rising senior in mechanical engineering, Baller is a year-round athlete (basketball and outdoor track), crafting enthusiast, and co-president of McCormick’s student government. With construction underway to renovate the residence, Baller is taking time to help support future MIT women so they can have the same powerful residential experience she’s had.

McCormick is scheduled to open again in August 2028 — well after Baller graduates. In this interview, she describes her thoughts on the transition and how she is working to maintain a sense of community among the dorm’s residents and incoming first-year students while updates are ongoing.

Q: Why did you choose to live in McCormick? 

A: I was recruited to play basketball in college, and other schools were pressing me for a decision. MIT was like, “Well, you got in. It’s up to you what you want to do.”  

So I came to CPW to find out what the campus is like. I stayed in [co-ed] Baker with one of my teammates. It was weird for me. I could have probably adjusted to being in a living space with men, but I guess I just bristled at the concept. I grew up in a Christian household, so I was used to certain things. I shared a bathroom with my sister, not my brother. 

What really set it in stone was going to the other CPW events at McCormick. They were like — “We’re an all-women’s community. The dorm is quieter. Everyone’s super nice. We like to do crafts.” Then they showed us the craft room. A whole room dedicated to crafts? I was sold.

Q: How would you describe the community in McCormick?

A: On the day my dad helped me move in, we had three suitcases I brought from Denver. He and I sorted out my stuff, and then we went down to the laundry room. I thought I saw a big spider or something, and a girl who was standing there asked, “Are you talking about Despereaux?” I had never even talked to this girl before. Even this was a way to bond! 

There’s a lot to love. Our heads of house are amazing. After the last day of class every semester, they have a tea and churro study break. They make the churros themselves. So we just come down, drink tea, chat with our friends, and eat churros and little tea sandwiches. That’s very McCormick — a little break with some good chatting. 

The heads of house also run something called “karao-cake.” When I first heard about it, I was like — “I’ll go for the cake.” They have a karaoke machine with a bunch of microphones attached, and we all sing songs together. And if they pick a song we don’t like, we all yell “No!” Everyone's on the same page. It gives really good sisterhood vibes. 

Also, I personally loved our all-women’s gym. As someone who has been an athlete for many years, I can say: We had amazing equipment in there. I’d rather work out where I don’t have to fight for a rack. I can just go and lift and do my workout.

Overall, the McCormick community is what you make of it. You can choose to be invested and have a great time. You can also just choose for it to be the place you come back to every night. I was in the same room sophomore and junior year, and so were a lot of the girls around me. By the end of last year it was like that scene from the “Barbie” movie — when they’re all in their houses, and say “Good night, Barbie! Good night, Barbie!”

Q: How did you get involved in the renovation project? 

A: I originally joined house government to be the craft chair and athletic chair. Later, I decided to run for co-president because McCormick was my first home away from home. I had honestly planned to go to college close to home, or where my friends were going. The thought of going to another state and being on my own just seemed too out of the ordinary. When I came here, I knew one person. 

When MIT first told us the dorm was being renovated, I was pretty excited to see what they were going to do. They held all-dorm events, brought donuts, and asked us to come and talk about what we envisioned for renovation. I said I wanted the biggest craft room you can imagine, pianos in every corner, and to get rid of the study cubicles in the penthouse no one uses. We really got to dream, right? 

But then MIT announced a one-year delay in the renovation, and you have the emotions. McCormick was home — and then they say it’s going to get renovated, then they say it will be next year. When my friend and I decided to run to become co-presidents, the rest of the dorm really didn’t want to talk anymore. We started meeting with [the Division of] Student Life on Zoom, but it was hard to get resident engagement. I appreciate that we’re in the conversations. We get to hear the numbers before other people do, but that’s just information.

Q: What’s the role of house government while the residence is being renovated? 

A: We do things that keep the energy alive. McCormick is more than just a building. The housing office just told us more than 300 incoming students expressed interest in the McCormick community, even though the dorm is being renovated.

To bring momentum into the renovation, we held an end-of-semester party where we dropped nice crewnecks, got a food truck, and had popcorn, cotton candy, a DJ, games, face paint — all the stuff. We also enjoy dorm movie outings, which would be a great tradition to continue. When the Taylor Swift “Eras Tour” movie was in theaters, we all got to have the experience of going over on the T together, and then sitting together singing Taylor Swift songs. We also saw “Wicked” and “Wicked for Good.” We have chill events, too, like crochet, painting, and eating pastries. All of this is about being together. Even if we’re not sitting there having a conversation, we’re existing together. That feels like home.

I’m also trying to help people who are dealing with the transition. It can be hard. You can’t have our heads of house move with you, or the craft room. You can’t have the cute merch that one of our students designs. If someone who has been moved to Maseeh doesn’t know anyone else on their floor, they don't get to have that Barbie moment. But maybe McCormick is holding a study break where they can hang out — a throwback to the old days, where we can craft, or drink boba, or whatever. 

Q: Has the effort been worth it? 

A: It’s worth it to me because I get to keep the momentum going, but I won’t know for sure until the dorm is open again and a freshman checks into their room and experiences the community.

They took our feedback doing the donuts and stuff, and they put a lot of our ideas into the design, but now I’ve got to see the finished product. I know MIT has to balance a lot of things, so they’re not necessarily going to do everything just because we asked. 

Q: What are your goals for when the renovation’s finished? 

A: The building won’t reopen before I graduate, so I guess there’s two things.

When I graduate, I would hope to see a house government team that’s excited to continue the traditions. It’s different to be affiliated with a community than to be living in it. I would love to graduate and leave here knowing McCormick is in good hands and the momentum our generation started helped drive us through to reopening. 

And when the dorm reopens, I want to come back and get a tour. I would just love to see the excitement around being back in the dorm. I’ll buy my own plane ticket!

de MIT News https://ift.tt/Js6TKQR

AI helps design new materials that work in the real world

Today, anyone with a large enough artificial intelligence model can generate millions of new material designs in minutes. Unfortunately, that hasn’t led to a huge leap in the number of new materials being used to improve the performance of products like computer chips and rockets.

One reason for the translation gap is that current models don’t reliably factor in the chemical stability of the materials they generate, and unstable materials aren’t very useful in the real world. That forces industries to allocate huge computational budgets to screening out all the unstable materials they generate, in some cases leaving behind a tiny fraction of usable options.

Now, MIT researchers have developed a framework that can be applied at the beginning of the materials generation process to vastly improve the stability rate while achieving targeted material properties. It works by ensuring every design satisfies certain key rules of chemistry relating to the electrons around the materials’ atoms before the expensive generation step begins. The researchers call their approach “crystal generator with valence-constrained design, or CrysVCD.

In a paper published today in Nature Computational Science, the researchers show how CrysVCD allowed several commonly used material models to meet those valence shell rules more often, and used it to achieve high lattice-dynamics stability — a stringent stability test — in nearly 70 percent of computational material generations. They also showed the approach could support the creation of materials with specific desired properties, like high thermal conductivity or high dielectric constant, which is important for computer chips and data centers.

A hint of how the researchers envision people using their system is in the name.

“If material-generating models are like DVDs, we are like the DVD player,” says associate professor of nuclear science and engineering Mingda Li. “You can plug this into any kind of model, not only existing diffusion models but also future models, where people can’t generate enough stable materials, and it can improve stability.”

Joining Li on the paper are Mouyang Cheng SM ’26 and Weiliang Luo, MIT doctoral students in materials science and engineering and chemistry, respectively; Hao Tang PhD ’26, a recent graduate in materials science and engineering; Bowen Yu, a senior undergraduate in physics; Yongqiang Cheng, a staff scientist at the Oak Ridge National Laboratory; Weiwei Xie, an associate professor at Michigan State University; Ju Li, MIT’s Carl Richard Soderberg Professor in Power Engineering; and Heather Kulik, MIT’s Lammot du Pont Professor of Chemical Engineering.

More efficient materials

Computational approaches to materials design have been around for decades, but recent advances in artificial intelligence have increased excitement about their potential. Of particular interest are models that can start with a desired material property and work backward to deliver a material that achieves that goal.

Some of those models use an AI technique known as diffusion, which is commonly used to generate images, while others use large language models like the one powering ChatGPT and Claude, but both approaches struggle to ensure their material generations achieve chemical stability or follow fundamental principles about how chemicals interact and behave.

The solution has been to add another layer of computing on top of the generative process to filter out unstable materials.

“It’s becoming easy to generate the material structure,” Cheng says. “But the validation process, especially the part where you test the stability, has a huge computational cost. It’s something like 90 percent of the computational cost for creating usable materials, and it can take weeks or months.”

Big companies with huge computing budgets can afford to run those processes, but many small companies and research labs can’t, potentially limiting innovation in the field.

“In academia, where we have fewer resources, I think we can still achieve strong performance with smarter designs and other approaches,” Kulik explains. “Generating a model and then down-selecting for stability is inefficient. There’s a high computational cost. But if we put a language model in the beginning of the process to constrain the generation, you can significantly enhance the ratio of stable materials generated.”

The new study involved MIT researchers affiliated with the departments of Materials Science and Engineering, Chemistry, Chemical Engineering, Physics, and Nuclear Science and Engineering. Together the researchers combined AI diffusion models with a language model. In the first stage of their process, the language model produces chemically valid formulas. In the second stage, the diffusion model uses that formula to generate the corresponding atomic structure of the crystal material in coordination with the underlying material generation model.

“Diffusion for typical material generation is a slow process — you can think of it like 1,000 steps to create one material,” Luo says.

“In contrast, when our model is used in the beginning, you can think of it like five steps. It allows you to screen out the unstable materials to generate higher quality materials. And it works with any models generating materials,” Tang adds.

The researchers showed their approach created more stable materials an order of magnitude more efficiently than approaches that rely on screening materials after they’re generated. When fine-tuned on stability metrics, their approach produced crystalline materials that achieved 68 percent mechanical stability and 85 percent metastability, which measures if a material stays in a stable state when undisturbed.

The researchers then used their approach to generate material candidates with high thermal conductivity and easy polarization in an electric field.

“These are materials useful for the semiconductor industry and high thermal conductivity materials relevant to data center cooling,” Ju Li says. “In principle, you could also use this to create other properties, but thermal conductivity has become really important for cooling data centers. There’s been a huge increase in energy use in that industry, and 30 percent of that energy goes to cooling. The industry needs materials with high thermal conductivity to more efficiently remove the heat.”

Democratizing material design

The new approach doesn’t work with every kind of material — it works best with solid structures with highly ordered internal arrangements. Still, the approach could be used to generate stable new crystalline materials with a host of important properties.

“We are not just generating stable materials, we’re also prioritizing performance,” Cheng says. “Any time you have two goals, achieving those goals with anything over 50 percent is hard in this field. In the past, people might have a goal for specific properties and not stability, or vice-versa, and get a single-digit percentage of materials that fit their goal.”

Ultimately the approach will enable more researchers to develop novel materials for a range of next-generation applications.

“This will save huge computation costs and time by removing downstream selection requirements,” Li says. “That will help not only large efforts that generate hundreds of millions of materials, but also smaller research groups with targeted applications.”

The work was supported, in part, by the U.S. Department of Energy, a Mathworks Engineering Fellowship, the National Science Foundation, and the U.S. Defense Threat Reduction Agency.

de MIT News https://ift.tt/V0PsFMJ

Retooling to help democracy revive

In the 20th century, the United States built the world’s dominant manufacturing powerhouse. A thriving middle class grew, well into the 1970s. The U.S. was a beacon of democracy, defeating fascism in World War II and beating back communism and other forms of authoritarianism during the Cold War. 

To MIT economist Daron Acemoglu, there is a deep intertwining among these things. Democracy, his work has shown, helps economies grow. As the industrial economy expanded, in Britain, the U.S., and other countries in the 19th and 20th centuries, so did democratic participation, as people tried to stake out new rights, or make real the rights ascribed to them. 

“The industrial age created the tools for shared prosperity around which democracy organized,” says Acemoglu, a Nobel Prize-winning economist and Institute Professor at MIT. 

Today, though, income inequality has grown markedly in the U.S., starting around 1980. The U.S. has deindustrialized to a significant extent, offshoring production and hurting shop-floor workers and their families. As Acemoglu sees it, this economic realignment has had deep civic consequences: A stranded working class has become more alienated from the institutions and ideas traditionally buttressing democracy.

And for those around the world supporting democracy, he says, “You really need to have the working classes in your coalition for it to make any sense.”

Acemoglu explores these topics in a new book, “What Happened to Liberal Democracy? Remaking a Politics of Shared Prosperity,” published by Penguin Random House. In it, he looks broadly at the benefits of democracy, the tensions it faces in everyday life, and democracy’s trajectory in recent decades.

Broadly, Acemoglu favors rebuilding “working-class liberalism,” essentially seeking the largest coalition that favors self-government and the rule of law. “Working-class liberalism has strong communal roots, eschews social engineering, and prioritizes shared prosperity, jobs, and public services,” Acemoglu writes in the book.

After all, Acemoglu believes, democracy is the one form of rule that promotes rights and liberties, and allows the flexibility and “experimentation” we need to address all the challenges a complicated world throws at us.

“Democracy is the only way we can make progress in society,” Acemoglu says. “Trying to impose top-down solutions to all our problems will ultimately not work.” 

Along the narrow corridor

Acemoglu has long studied the relationship between economic growth, rights, and democracy. With economist Simon Johnson of MIT and political scientist James Robinson, now of the University of Chicago, Acemoglu published a landmark series of studies in the early 2000s demonstrating that economic growth is helped by the development of stable democratic institutions, including property rights. For that work, Acemoglu, Johnson, and Robinson later shared the 2024 Sveriges Riksbank Prize in Economic Sciences in Memory of Alfred Nobel. 

Acemoglu’s 2012 bestseller “Why Nations Fail”— written before democracy’s current challenges seemed as acute — synthesized his research on these topics. His 2019 book “The Narrow Corridor,” co-authored with Robinson, casts democratic governments as essential to liberty because they protect people simultaneously from overreach by an authoritarian state, on the one hand, and from domination by other groups in society, on the other. 

However, as Acemoglu has consistently emphasized, self-governance is an ongoing effort; this machine does not run on its own. Relatedly, in the new book, Acemoglu critiques some famous attempts to formulate governance as a neat “social contract,” including Jean-Jacques Rousseau’s conception of a “general will” in society. 

Those ideas helped make the case for political rights, but actual governance in a pluralistic society will always be a messy process. 

“You have to allow communities, and societies in aggregate, to build rules around shared values for anything to stick as institutions, norms, or aspirations,” Acemoglu says. “When you go down the social contractarian path, you sometimes fool yourself into thinking there are clear solutions to dilemmas that in reality don’t quite have such obvious ways of being resolved.” 

For instance, Acemoglu notes, democracy itself “is built on tolerance and acceptance of plural perspectives, but how do you deal with people who are intolerant?” In the book, he largely regards interventions to stamp out seemingly intolerant thought as being unwise and politically counterproductive.

“You’re not going to have a clear-cut solution to all cases,” he says.

The economic realignment

Even with leaders backing a pragmatic, flexible approach to self-governance, democracy faces another challenge: supporting the material welfare of citizens. And here, “What Happened to Liberal Democracy?” takes an unflinching look at the postwar economy, finding fault lines that have shaken the political order. 

The roughly three decades after World War II were fantastic for many workers in democracies, and certainly in the U.S., where middle-class incomes grew by 2.5 percent annually into the 1970s. 

There were always going to be forces pushing back on this trend, and U.S. companies started offshoring and outsourcing production work to clamp down on wage growth. But one other technological and economic trend occurred just as the middle classes of the industrial economy were reaching new heights. 

“Then computers happened,” writes Acemoglu in the book — referring to a complex set of economic and civic realignments involving technology-driven shifts in work. 

Over time, computers started replacing significant numbers of clerical office workers, shop-floor industrial workers, and other types of employees who were earning middle-class wages without holding a college degree. In recent decades, middle-class incomes have only grown by about 0.5 percent annually. 

To be sure, computers have produced plenty of benefits, and created many new forms of work. But as research shows, those jobs have tended to go mostly to college-educated employees, creating a significant split in society between well-educated, well-paid, white-collar workers, and less-educated, worse-paid workers in blue-collar and service jobs. That national share of income hauled in by the top 1 percent of earners has basically doubled in this time, from 10 percent to nearly 20 percent.

Crucially, in Acemoglu’s analysis, this material gap between more-educated and less-educated social cohorts has translated to U.S. politics, with political groupings reshuffling along educational lines, and cultural politics following suit. That’s the dynamic the U.S. faces now — even as one also accounts for the effects of social media and other polarizing features of contemporary society. 

“We now live in a less-industrial world, and we also live in an age defined by social media, much greater levels of conflict, more polarization, and now AI, and all of that complicates things,” Acemoglu says. 

Always a work in progress

This precise feature of contemporary society — deindustrialization fueling an earnings gap that has led to more political polarization — is what shapes Acemoglu’s prescription in response, the idea that “working-class liberalism” is needed to strengthen democracy. 

There are many potential ingredients in this formula, from politicians determined to reach across class lines to workers regaining the impetus to organize in their workplaces. 

“Trade union participation itself is a very important form of local governance that’s very difficult or unimaginable in an authoritarian society,” Acemoglu says, even while noting that he has not always agreed with the actions of particular unions in the past. 

Still, Acemoglu adds, “I don’t think the economic aspect is the only one, in that you cannot just gain the trust of workers by ensuring there are wage gains. That is an important step but it is not sufficient.” Voters need asurances that politicians are thinking about them, at least share their concerns, and have a grounding in similar values. More candidates today need to seek a shared language about those things.

That’s not easy in a world characterized, in part, by global migrations, increasingly diverse national populations, and economic flux. But it is possible, Acemoglu thinks. 

“There are deep dilemmas faced by liberal democracy that were sometimes going to come to boiling points, and this becomes more heightened when societies such as the U.S. and some European ones are simultaneously becoming more complex, more globalized, and more hetereogeneous,” Acemoglu says. On the other hand, he adds, “Multiracial tensions, I would say, were much worse for the U.S. in the 1950s and 1960s. We made democracy work then, in the face of much more difficult race problems, so why not today?”

None of this is a straightforward task, of course. “Forging working-class liberalism is a tall order in the best of times and much more challenging in today’s polarized environment,” Acemoglu writes in the new book. Still, he adds, even in frustrating moments, the stakes are too important for people to relent.

“Democracy is a success,” Acemoglu says. “It’s easy to fall into a trap of painting the democratic project as being doomed to failure, and I want to avoid that.” He adds: “It continues to be a work in progress.” 

de MIT News https://ift.tt/1utM5nX