Prof. Yoon Kim speaks with Vox reporter Adam Clark Estes on how to address hallucinations and misinformation within large language models. “I don't think we'll ever be at a stage where we can guarantee that hallucinations won't exist,” says Kim. “But I think there's been a lot of advancements in reducing these hallucinations, and I think we'll get to a point where they'll become good enough to use.”
Researchers from MIT have developed RoboGrocery, a soft robotic system that “can determine how to pack a grocery item based on its weight, size and shape without causing damage to the item,” reports Jennifer Kite-Powell for Forbes. “This is more than just automation—it's a paradigm shift that enhances precision, reduces waste and adapts seamlessly to the diverse needs of modern retail logistics,” says Prof. Daniela Rus, director of CSAIL.
Prof. Dylan Hadfield-Menell is interviewed by Wired’s Will Knight about CriticGPT, a new tool developed by OpenAI that will assist human trainers in developing AI. “It might lead to big jumps in individual capabilities, and it might be a stepping stone towards sort of more effective feedback in the long run,” Hadfield-Menell says.
Writing for the Boston Business Journal, MITEI Director William Green emphasizes that in order to address the climate crisis, "we need to convene universities, industry, and government to address the challenges of every sector including construction, manufacturing, agriculture, and the electric grid." Green notes: "We at MIT are searching for real climate solutions that the public will adopt, and that merit the huge investments necessary for wide deployment. By working collaboratively to solve these complex issues we will successfully address the greatest threat facing humanity today."
MIT scientists have shown a simple paper-strip test can detect the flu and identify the specific strain, which could prove useful in improving outbreak response and infection care, writes Dennis Thompson for HealthDay. Jon Arizti-Sanz PhD ’24 says “being able to tease apart what strain or subtype of influenza is infecting a patient has repercussions both for treating them and public health interventions.”
Prakash Govindan PhD '12 and Anurag Bajpayee SM '08 PhD '12 co-founded AlkaLi, a startup working to extract lithium from brine and process it for use in batteries, reports Amy Feldman for Forbes. AlkaLi uses resins and membranes to more easily extract the lithium from brine, then relies on its own technology to concentrate the mineral, which ultimately is precipitated into a solid for use in batteries,” writes Feldman.
Researchers from MIT and elsewhere have developed a new technique that removes lead from water using repurposed beer yeast, reports Boston 25 News. The researchers “developed a hydrogel capsule to hold the yeast after it is cleaned, freeze-dried, and ground into a powder,” explains Boston 25. “Researchers said the yeast capsules could be modified to remove other dangerous contaminants from water, including PFAS and microplastics.”
Researchers from MIT have developed “sustainable, offshore, hydrodynamic,” artificial reef structures capable of dissipating “more than 95% of an incoming wave’s total energy,” reports Nick Warburton for Materials World. The design “comprises vertical cylinders with four rudder-like slats attached to them, so that water can flow through the structure to generate 'swirling masses of water' or large eddies,” explains Warburton.
MIT researchers have improved upon the diffusion models used in AI image generation, reports Alok Jha for The Economist. Working with electrically charged particles, the team created “Poisson flow generative models,” which “generate images of equal or better quality than state-of-the-art diffusion models, while being less error-prone and requiring between ten and 20 times fewer computational steps,” Jha explains.
Ronald Davis III '18, SM '22, PhD '24 speaks with Jason Groves of Las Cruces Sun-News about how his research at both MIT and an Army Department of Defense lab inspired his interest in applying AI technologies to improve wireless communications, work he is now using as the foundation of his startup VectorWave. "When I think of a student earning multiple degrees from a school such as MIT along with a Ph.D., I'm blown away that it could come from a small place like Las Cruces [New Mexico]," says Brandon Sprague, Davis’ former high school mathematics teacher.
MIT scientists have discovered that metals become harder when heated under certain conditions, reports Isabelle Dume for Physics World. In an interview, Prof. Christopher Schuch says by bombarding metals with tiny bits of sapphire small enough to avoid shock waves, “the result is a clean and quantitative set of data that clearly reveals the counterintuitive ‘hotter-is-stronger’ effect.”
Prof. Regina Barzilay spoke at The Futurist Summit: The Age of AI – an event hosted by The Washington Post – about the influence of AI in medicine. “When we're thinking today how many years it takes to bring new technologies [to market], sometimes it's decades if we’re thinking about drugs, and very, very slow,” Barzilay explains. “With AI technologies, you've seen how fast the technology that you're using today is changing.”
Prof. Sara Seager, Prof. Robert Langer and Prof. Nancy Kanwisher have been awarded the 2024 Kavli Prize for their work in the three award categories: astrophysics, nanoscience, and neuroscience, respectively, reports Michael T. Nietzel for Forbes. According to the Norwegian Academy of Science and Letters, this award honors scientists with outstanding research “that has broadened our understanding of the big, the small and the complex,” writes Nietzel.
MIT researchers used tools of computational complexity and mathematical concepts to prove that no analysis of the Super Mario Bros video game level “can say for sure whether or not it can ever be completed,” reports Matthew Sparkes for New Scientist. “The idea is that you’ll be able to solve this Mario level only if this particular computation will terminate, and we know that there’s no way to determine that, and so there’s no way to determine whether you can solve the level,” says Prof. Erik Demaine.
Washington Post reporter Carolyn Johnson spotlights how Prof. Laura Schulz and her colleagues have been exploring why ChatGPT-4 performs well on conversation and cognitive tests, but flunks reasoning tests that are easy for young children. Schulz makes the case that to understand intelligence and create it, childhood learning processes should not be discounted. “That’s the kind of intelligence that really might give us a big picture,” Schulz explains. “The kind of intelligence that starts not as a blank slate, but with a lot of rich, structured knowledge — and goes on to not only understand everything we have ever understood, across the species, but everything we will ever understand.”