Research

A working paper by Prof. Drew Fudenberg and a team of researchers suggests that major AI companies should share their advances with one another to mitigate competitive pressures and decrease the potential of a catastrophic event, writes Amanda Taub for The New York Times. “If the players, including companies like OpenAI and Anthropic, can see that no one is racing into dangerous territory, they will have less motivation to do so themselves, the paper reasons,” explains Taub.  

In a video interview, Prof. Turkle speaks to CNN’s Madeline Holcombe about how AI chatbots may contribute to feelings of loneliness and isolation in adults. “Technology offered the possibility of feeling like friendship without the demands of vulnerability,” says Turkle. “It turns out you can’t have intimacy or friendship without vulnerability.” 

In a conversation with Slate’s Nitish Pahwa, Prof. Sherry Turkle discusses how to better bolster human connection and limit AI chatbot dependence, which she explores in her new book, “Artificial Intimacy: Who We Become When We Talk to Machines.” “We’re not going to solve this problem by manipulating the chatbots,” says Turkle. “We’re going to solve this problem by getting back together as us. We have to summon what we think is important about being people, both for ourselves and for our kids.”  

Prof. Sherry Turkle speaks to The San Francisco Chronicle’s Jessica Zack about the risks related to humans developing intimate relationships with AI chatbots, detailed in her book “Artificial Intimacy: Who We Become When We Talk to Machines.” “The experience of being with an object that uses the first person — the ‘I,’ and, even worse, the ‘we’— is problematic. We are evolved to think that if something talks to us that way, it is a person, so you’re fighting against your very deep evolutionary buttons that are being pushed,” says Turkle.  

MIT researchers have developed a paper-thin biohybrid robot “capable of swimming that is powered by living muscle cells instead of a traditional motor,” writes Forbes’ Conor Murray. The muscle cells are engineered to respond to flashes of light, allowing the robot to swim through mazes. The researchers believe that in the future biohybrid robots could eventually be used for everything from precision microsurgery to exploring fragile environments.  

Hamid Doosthosseini PhD ‘25, an MIT postdoc, speaks to Tech Brief’s Andrew Corselli about his team’s work engineering bacteria to act as transistors on “living circuit boards.” “Simply, each cell works somewhat like a traffic light at an intersection, it determines whether traffic (here a chemical signal) should continue or stop based on traffic in the intersecting road,” says Doosthosseini. “Printing a network of cells can therefore combine, compute, or direct signals.” 

MIT researchers have invented a digestible, bioresorbable battery that can deliver drugs directly to the GI tract, reports Smithsonian Magazine’s Sara Hashemi. Prof. Giovanni Traverso and his colleagues “made two versions of the battery of different sizes and voltages,” writes Hashemi. “Exposing them to an acidic liquid similar to gastric fluid revealed they would function normally in the stomach for about three days.” 

Prof. David Autor and his team of researchers found that junior patent attorneys who used AI for assistance showed no net improvement in their skills, reports Andy Mukherjee for Bloomberg. “When researchers subsequently tested everyone on an unassisted, offline task to gauge internalized judgment, the learning gains went entirely to senior lawyers,” writes Mukherjee. “Foundational expertise may be a prerequisite for extracting durable skill from AI-assisted practice,” the researchers noted.  

In The Wall Street Journal’s “Economics Newsletter,” reporter Greg Ip highlights a study by Prof. David Autor that found that AI did not significantly improve the performance of the 133 patent lawyers who participated. “Among junior lawyers, the performance of some got better, others got worse, and on average they were the same. Among senior lawyers, the lowest performers improved, the best didn’t,” writes Ip. “The conclusion: Lawyers don’t acquire durable new skills with AI.” 

The Economist highlights MIT researchers’ new “paper-based battery whose components are safely absorbed by the body,” to help deliver drugs to the GI tract more effectively. “The new battery is a stack of magnesium, gel and chemically active paper—all coated in wax to protect it from the harsh environment of the gut. It is also compact, taking up only about a quarter of the space inside a standard gelatine capsule, leaving plenty of room for the device it is to power, as well as other useful payloads, such as a drug.”  

MIT researchers have developed an ingestible battery that could be used to monitor drug delivery and avoid the need for surgery; a “milestone towards potential human use for health monitoring and treatments,” writes Michael Peel for the Financial Times. Improving the effectiveness of therapies “moves the technology beyond a materials demonstration toward an integrated biomedical system,” says Prof. Giovanni Traverso, senior author of the study.  

MIT researchers found that AI assisted work leads to the ‘diversity collapse’ of a brand’s voice, and users that rely on AI to do their thinking lose skills over time, highlights Forbes’ Michelle O'Connor. “Judgment, taste and the instinct to catch when something’s off isn’t a soft skill you can afford to lose,” writes O’Connor. “The MIT data is the first hard evidence of that.” 

Quanta Magazine’s Marlowe Starling highlights a 2014 study by MIT researchers that found that coral cilia “interact with the boundary layer by rapidly whipping about to generate swirls of fresh, oxygenated seawater.” The researchers “not only modeled the tiny vortices created by the cilia for the first time, but also revealed the cilia’s importance for survival and metabolism,” notes Starling. 

A recent study by MIT researchers found that AI chatbots easily generated misleading election-related answers, images and videos, writes Emily Condon for USA Today. “This, along with other findings from our study, suggests that in an ongoing election with little authoritative information for models to rely on, AI may not be as helpful in pushing back against unsubstantiated rumors and misinformation,” the researchers explain.  

A new report published in part by MIT FutureTech finds that AI does not significantly accelerate scientific research, writes Peter Hall for the Scientific American. “As the experiments become larger or more complex, so, too, do the models that simulate them, and the data to train those models is hard to come by,” Hall explains. “Robots handling real-world materials raise safety questions that are unknown to software.”