STAT reporter Katie Palmer spotlights Principal Research Scientist Leo Anthony Celi’s research underscoring the importance of improving the diversity of datasets used to design and test clinical AI systems. “The biggest concern now is that the algorithms that we’re building are only going to benefit the population that’s contributing to the dataset,” says Celi. “And none of that will have any value to those who carry the biggest burden of disease in this country, or in the world.”
MIT researchers and two high school seniors have developed DualFair, a new technique for removing bias from a mortgage lending dataset, reports Hiawatha Bray for The Boston Globe. “When a mortgage-lending AI was trained using DualFair and tested on real-world mortgage data from seven US states,” writes Bray, “the system was less likely to reject applications of otherwise qualified borrowers because of their race, sex, or ethnicity.”
The MIT AI Hardware Program seeks to bring together researchers from academia and industry to “examine each step of designing and manufacturing the hardware behind AI-powered technologies,” reports Emily Bamforth for EdScoop. “This program is about accelerating the development of new hardware to implement AI algorithms so we can do justice to the capabilities that computer scientists are developing,” explains Prof. Jesús del Alamo.
MIT researchers have developed reconfigurable, self-assembling robotic cubes embedded with electromagnets that allow the robots to easily change shape, reports John Koetsier for Forbes. “If each of those cubes can pivot with respect to their neighbors you can actually reconfigure your first 3D structure into any other arbitrary 3D structure,” explains graduate student Martin Nisser.
The MIT AI Hardware Program is aimed at bringing together academia and industry to develop energy-optimized machine-learning and quantum-computing systems, reports Katyanna Quach for The Register. “As progress in algorithms and data sets continues at a brisk pace, hardware must keep up or the promise of AI will not be realized,” explains Professor Jesús del Alamo. “That is why it is critically important that research takes place on AI hardware."
TechCrunch reporters Christine Hall, Anita Ramaswamy, Connie Loizos and Mary Ann Azevedo spotlight Sribuu, an AI-powered personal financial advisor in Indonesia, co-founded by Nadia Amalia ’20. The company is aimed at helping “users make better money decisions with our wealth management tools and give personalized saving advice based on their financial habits,” they write.
MIT researchers have utilized a new reinforcement learning technique to successfully train their mini cheetah robot into hitting its fastest speed ever, reports Matt Simon for Wired. “Rather than a human prescribing exactly how the robot should walk, the robot learns from a simulator and experience to essentially achieve the ability to run both forward and backward, and turn – very, very quickly,” says PhD student Gabriel Margolis.
Fortune reporter Jeremy Kahn spotlights a study co-authored by Prof. Marzyeh Ghassemi exploring issues associated with “explainable” AI systems that are being applied in fields such as healthcare, finance and government. The researchers explain that those using such systems “might have misunderstood the capabilities of contemporary explainability techniques—they can produce broad descriptions of how the AI system works in a general sense but, for individual decisions, the explanations are unreliable or, in some instances, only offer superficial levels of explanation.”
Profs. Ruonan Han and Qing Hu speak with Popular Science reporter Rahul Rao about their work with terahertz waves. “There’s a laundry list of potential applications,” says Hu of the promise of terahertz waves.
Prof. Stuart Madnick writes for The Wall Street Journal about how flaws in a company’s cybersecurity defenses can lead to cyberattacks. “Every decision regarding cybersecurity must weigh the benefits of not doing something (cost savings or the faster growth) against the increased risk to the organization,” writes Madnick.
MIT researchers have created a new computer algorithm that has allowed the mini cheetah to maximize its speed across varying types of terrain, reports Shi En Kim for Popular Science. “What we are interested in is, given the robotic hardware, how fast can [a robot] go?” says Prof. Pulkit Agrawal. “We didn’t want to constrain the robot in arbitrary ways.”
MIT researchers have used a new reinforcement learning system to teach robots how to acclimate to complex landscapes at high speeds, reports Emmett Smith for Mashable. “After hours of simulation training, MIT’s mini-cheetah robot broke a record with its fastest run yet,” writes Smith.
CSAIL researchers developed a new machine learning system to teach the MIT mini cheetah to run, reports James Vincent for The Verge. “Using reinforcement learning, they were able to achieve a new top-speed for the robot of 3.9m/s, or roughly 8.7mph,” writes Vincent.
Gizmodo reporter Andrew Liszewski writes that CSAIL researchers developed a new AI system to teach the MIT mini cheetah how to adapt its gait, allowing it to learn to run. Using AI and simulations, “in just three hours’ time, the robot experienced 100 days worth of virtual adventures over a diverse variety of terrains,” writes Liszewski, “and learned countless new techniques for modifying its gait so that it can still effectively loco-mote from point A to point B no matter what might be underfoot.”
Graduate student Matt Groh speaks with Scientific American reporter Sarah Vitak about his team’s work studying whether human detection or artificial intelligence is better at identifying deepfakes and misinformation online. “One of the things that we would suggest for the future development of these systems is trying to figure out ways to explain why the AI is making a decision,” says Groh.