Machine learning

Postdoctoral Associate Pat Pataranutaporn speaks with Wall Street Journal reporter Heidi Mitchell about his work developing Future You, an online interactive AI platform that “allows users to create a virtual older self—a chatbot that looks like an aged version of the person and is based on an AI text system known as a large language model, then personalized with information that the user puts in.” Pataranutaporn explains: “I want to encourage people to think in the long term, to be less anxious about an unknown future so they can live more authentically today.” 

MIT engineers have developed a new training method to help ensure the safe operation of multiagent systems, including robots, search-and-rescue drones and self-driving cars, reports Jijo Malayil for Interesting Engineering. The new approach “doesn’t focus on rigid paths but rather enables agents to continuously map their safety margins—the boundaries within which they must stay,” writes Malayil. 

Prof. Pattie Maes speaks with Wired reporter Reece Rogers about the potential benefits and challenges posed by AI agents. “The way these systems are built, right now, they're optimized from a technical point of view, an engineering point of view,” says Maes. “But, they're not at all optimized for human-design issues.” 

Noubar Afeyan PhD '87 and a member of the MIT Corporation speaks with Boston Globe reporter Aaron Pressman about the future of artificial general intelligence (AGI) and “superintelligent” AI. “Humans have long developed tools, microscopes, mass spectrometers, you name it, to help them be able to understand nature better,” says Afeyan. “Now one of the tools, in the case of machine [learning], we’re elevating to the level of a whole new intelligence.”

Ars Technica reporter Jacek Krywko spotlights how MIT researchers have developed a new photonic chip that that can “compute the entire deep neural net, including both linear and non-linear operations, using photons.” Visiting scientist Saumil Bandyopadhyay '17, MEng '18, PhD '23 explains that: “We’re focused on a very specific metric here, which is latency. We aim for applications where what matters the most is how fast you can produce a solution. That’s why we are interested in systems where we’re able to do all the computations optically.” 

Prof. Daron Acemoglu speaks with Financial Times reporter Rana Foroohar about the impact of automation on the labor market. “It’s likely that the short- to midterm gains from AI will be distributed unequally, and will benefit capital more than labor,” says Acemoglu. 

Iqbal Dhaliwal, executive director of the Abdul Latif Jameel Poverty Action Lab (J-PAL), speaks with NPR reporter Ari Daniel about the positive social impact that can be brought forth by AI. "As this technical revolution unfolds in real time," says Dhaliwal, "we have a responsibility to rigorously study how these technologies can help or harm people's well-being, particularly people who experience poverty, and scale only the most effective AI solutions."

Writing for Wired, Prof. Daniela Rus, director of CSAIL, highlights the future of “physical intelligence, a new form of intelligent machine that can understand dynamic environments, cope with unpredictability, and make decisions in real time.” Rus writes: “Unlike the models used by standard AI, physical intelligence is rooted in physics; in understanding the fundamental principles of the real world, such as cause-and-effect.”

Sloan research fellow Michael Schrage speaks with Fortune reporter Sheryl Estrada about the impact of AI on CFO roles. “The ongoing ‘Compound AI’ revolution, which involves approaching AI tasks by combining multiple interacting components, will increasingly transform the CFO role into that of an AI-powered chief capital officer (CCO),” says Schrage. “This is an analytics-driven shift that isn’t optional but imperative for enterprise growth.”

Knowable Magazine reporter Katherine Ellison spotlights Future You, a new program developed by researchers at MIT that “offers young people a chance to chat with an online, AI-generated simulation of themselves at age 60.” 

Prof. Daron Acemoglu highlights the importance of adopting alternative technologies in the face of AI advancements, reports Jared Newman for Fast Company. “We need investment for alternative approaches to AI, and alternative technologies, those that I would say are more centered on making workers more productive, and providing better information to workers,” says Acemoglu.

Forbes reporter Joe McKendrick spotlights a study by researchers from the MIT Center for Collective Intelligence evaluating “the performance of humans alone, AI alone, and combinations of both.” The researchers found that “human–AI systems do not necessarily achieve better results than the best of humans or AI alone,” explains graduate student Michelle Vaccaro and her colleagues. “Challenges such as communication barriers, trust issues, ethical concerns and the need for effective coordination between humans and AI systems can hinder the collaborative process.”

In an interview with CNBC, Prof. Max Tegmark highlights the importance of increased AI regulation, specifically as a method to mitigate potential harm from large language models. “All other technologies in the United States, all other industries, have some kind of safety standards,” says Tegmark. “The only industry that is completely unregulated right now, which has no safety standards, is AI.” 

Researchers from MIT and elsewhere have found that “AI doesn’t even understand itself,” reports Peter Coy for The New York Times. The researchers “asked AI models to explain how they were thinking about problems as they worked through them,” writes Coy. “The models were pretty bad at introspection.” 

Liquid AI, an MIT startup, is developing technology that “holds the same promise of writing, analyzing, and creating content as its rivals while using far less computing power,” reports Aaron Pressman for The Boston Globe.