Skip to content ↓

Topic

Machine learning

Download RSS feed: News Articles / In the Media / Audio

Displaying 1 - 15 of 994 news clips related to this topic.
Show:

Nature

MIT researchers have demonstrated how an AI model “can interpret the raw internal values of some other models well enough to predict what different parts of the systems do. It can then edit the models to make improvements—including boosting their mathematical abilities,” writes Nature’s Elizabeth Gibney. “The authors hope that, with development, the findings could boost the efficiency of AI models,” notes Gibney. “We may need much less data, much less compute and much less time to teach our models a specific skill,” says graduate student Rikiya Takehi. 

The Boston Globe

Boston Globe meteorologist Ken Mahan spotlights how scientists from MIT and elsewhere developed a new machine learning technique that could “significantly improve weather predictions and storm forecasts anywhere from two weeks to over a month away, an important window of time often referred to as the ‘predictability desert’ or ‘void.’” The new technique “signals a tremendous breakthrough in the accuracy of longer range forecasting. Such fine-tuning of forecasts could better aid the public, communities, farmers, and others by allowing more time to prepare for extreme weather, from dangerous severe cold snaps to torrential flooding.”  

The Hill

For The Hill, Liberty Vittert Capito ‘10, cohost of MIT’s Data Science Podcast, advises individuals against confiding private information in AI chatbots. “A search query into google is just a few words, but a late-night confession, or a sensitive medical inquiry contains hundreds of words of context. It often specifies why the inquiry is being made. It illuminates exactly what users fear, whom they love, what they regret, their moods and their most intimate health issues,” writes Vittert Capito. “This is unlike anything data brokers have been selling for years and getting rich.” 

The Wall Street Journal

The conversational AI models developed by MIT spinout Liquid AI are being introduced in Mercedes-Benz cars, writes Isabelle Bousquette for The Wall Street Journal. “Liquid AI invented a new architecture known as liquid neural networks, adaptive and efficient systems inspired by the neural activity of worms known scientifically as C. elegans,” writes Bousquette. Using the Liquid AI models in vehicles instead of cloud-based AI systems could allow for more “privacy, speed and the ability to run without constant network connectivity.” 

NPR

Prof. Daron Acemoglu shares his insights on artificial intelligence and the future of work and education during “Ask an Economist,” a virtual event hosted by NPR’s “The Indicator from Planet Money.” “I think there are a lot of pro-teacher AI opportunities where you give better tools to teachers, but teachers still remain central to the learning experience of students,” says Acemoglu. AI “can individualize the teaching experience much more effectively and at low cost in a way that wasn't possible before. And that expands teacher expertise, expands teacher capabilities, enables them to perform new tasks, doesn't replace them, doesn't sideline them.” 

Fortune

CSAIL researchers have discovered that a large frontier AI model can, without any specialized training, understand and fine-tune another AI model’s capabilities, reports Fortune’s Jeremy Kah. The research “could be a big breakthrough in interpretability—our ability to understand the internal ‘thought’ processes of large language models,” writes Kah. “That’s critical because interpretability is one of the ways we are hoping to keep AI agents under human control.”   

Los Angeles Times

Los Angelos Times reporter Mark Swed spotlights Prof. Tod Machover’s “already and not yet,” a new symphony that “handles democracy not as a done deal, but a series of scary questions — and doesn’t trust the heartless machines for answers.” “Already and not yet” challenges us to make the world more beautiful by looking deeply into ourselves,” writes Swed. The orchestral score treats “musicians as creative artists and the audience as capable of opening its collective mind.”  

CNN

CNN reporter Hadas Gold meets with Prof. Adam Berinsky and Prof. Chara Podimata to learn more about their new tool that tracks how popular AI chatbots source their information and respond to questions related to the midterm elections. “Even if people do not necessarily say that they trust AI, they like what AI gives them, which is a single answer,” says Berinsky. “What worries me is that the average voter, someone who might know a little about politics, but doesn’t know the details, is going to get this definitive answer [from an AI chatbot] that might be based on a different stew of information.”

The New Yorker

Prof. John Urschel speaks to The New Yorker’s Rivka Galchen about AI being used as a tool for teaching and research in the field of mathematics. “I would say what I call the more scientist approach [to new technology] is being focused on trying to push the frontier of things as far as we can as quickly as we can,” says Urschel. “And the artist perspective is, ‘let me do something beautiful and enjoyable and manipulate these things,’ and the act of doing this is really, you know, a joyful thing.” 

Forbes

Writing for Forbes, alumnus Paul Hsieh discusses radiologists employing rapidly evolving AI tools to help deliver diagnoses. “Radiologists are willing to employ AI tools that have been statistically validated to provide reliable diagnoses, even if we don’t yet fully understand how those tools work,” writes Hsieh. “But radiologists are not content to stop there. We also continually seek to improve genuine human understanding of how AI works in our field as a matter of professional responsibility for ourselves and for our patients.”  

Forbes

CodeSignal, co-founded by alumnus and CEO Tigran Sloyan, has developed Agentic Skills Intelligence, a new product that “uses AI agents to evaluate the skills people use in their work and links the findings to personalized practice,” writes Forbes’ Ray Ravaglia. CodeSignal has built an AI agent “for each domain in its job taxonomy, such as engineering and sales, along with agents for capabilities that cut across jobs, among them communication and AI literacy,” notes Ravaglia. “The agents draw on systems including Gong, Salesforce, and GitHub, and they score what they find against rubrics the company says are grounded in industrial-organizational psychology.” 

Bloomberg

Prof. Sanjay Sarma speaks to Bloomberg’s Saritha Rai about the difficulty of developing dexterous robots for use in the home versus developing factory machines for use in standardized plants. “Heavy work, hazardous tasks have long been the focus of automation,” says Sarma. “Robots seem to be following an evolutionary path–the finer the motor or cognitive skills required, the harder to automate.”  

TechCrunch

Prof. Ryan Williams speaks to TechCrunch’s Tim Fernholz about AI platforms refuting the Exponential Time Hypothesis, part of a broader class of problems known as P versus NP. “I do think that it kind of calls into question a lot of the intuitions that we thought we had about … how tractable [hard NP problems] are. I also think that this is probably not the final answer either,” says Williams. AI is “forcing people to be a little more ambitious about the kind of work that they do, and to think more deeply about, okay, if all the things I used to work on are now low-hanging fruit for AI, how can I raise the bar for the kind of scientific questions I want to ask?” 

GBH

Live on GBH’s “The Curiosity Desk,” hosted by Edgar B. Herwick III, Prof. Adam Berinsky joins Boston Globe reporter Aaron Pressman and Kate Gage from the Cooperative Lab Project for a conversation about the impact of AI on the 2026 midterm elections. Berinsky notes that in the runup to the midterms, voters “need to be careful about what they are paying attention to. I think we want to separate out what is new to AI and what is just communication in general,” says Berinsky. “In elections where margins are tens of thousands of votes in states that can determine the presidency, even really small effects matter.”  

Forbes

Forbes reporter Marshall Shepherd highlights the Probabilistic Bias Correction tool, a new AI system developed by MIT researchers that could extend weather forecasting predictability beyond two weeks.