Machine learning

Prof. Jonathan How and graduate student Yi-Hsuan (Nemo) Hsiao speak with Tech Briefs reporter Andrew Corselli about their latest work developing an aerial microrobot that is “agile enough complete 10 consecutive somersaults in 11 seconds, even when wind disturbances threatened to push it off course.” Hsiao explains that: “This work demonstrates that soft and microrobots, traditionally limited in speed, can now leverage advanced control algorithms to achieve agility approaching that of natural insects and larger robots, opening up new opportunities for multimodal locomotion.” 

MIT researchers have developed “GelSight,” a system that provides robots with a sense of touch, reports Ben Guarino for Scientific American. “GelSight can identify by touch the tiny letters spelling out LEGO on the stud of a toy brick,” explains Guarino. 

Lila Sciences, a startup co-founded by Prof. Rafael Gómez-Bombarelli, is developing new platforms aimed at enabling AI-driven laboratories to accelerate materials discovery for energy, sustainability, and computing, writes David Rotman for Technology Review. “If they succeed, these companies could shorten the discovery process from decades to a few years or less,” Rotman notes. 

MIT researchers have developed a deep-learning model “capable of predicting the precise movements, divisions, and restructuring of thousands of cells during the embryo’s transition from a simple cluster to a complex organism,” reports Mrigakshi Dixit for Interesting Engineering. “This model currently provides a sneak peek into the fruit fly’s earliest developmental stage,” explains Dixit. “In the future, it could be used to predict how more complex tissues, organs, and organisms develop.” 

Prof. Daron Acemoglu spoke at the City University Graduate Center’s panel discussion about the development of AI in the workforce. Acemoglu says “[AI could take] very different directions, and which direction we choose is going to have great consequences in terms of its labor market impact.” 

A study by MIT researchers examining the carbon emissions of self-driving cars found that “the power required to run one billion driverless vehicles driving for one hour per day could consume as much energy as all existing data centers in the world,” reports Claire Brown for The New York Times. Graduate student Soumya Sudhakar explains that another big unknown is how autonomous vehicles could change the way people travel, adding to the uncertainty over the overall long-term emissions outlook for self-driving cars. 

Prof. James Collins and his colleagues are using AI to develop new compounds to combat the growing problem of antibiotic resistant bacteria, reports Lisa M. Krieger for the San Francisco Chronicle. Thus far, “Collins and his colleagues have synthesized several compounds that combat hard-to-treat infections of gonorrhea and MRSA,” writes Krieger. “These techniques are also being harnessed to fight diseases, like cancer, lupus and arthritis.” 

Prof. James Collins speaks with Financial Times reporter Patrick Temple-West about his work using AI to design new antibiotic compounds to combat drug-resistant bacteria. “At present, the [AI] models are doing quite well at designing compounds that can attack in a Petri dish,” says Collins. 

Two new research papers by scientists from MIT and other institutions find that AI chatbots are successful at shifting the political beliefs of voters, and that the “most persuasive chatbots are those that share lots of facts, although the most information-dense bots also dole out the most inaccurate claims,” reports Sarah Kuta for Smithsonian Magazine. “If you need a million facts, you eventually are going to run out of good ones and so, to fill your fact quota, you’re going to have to put in some not-so-good ones,” says Visiting Prof. David Rand. 

A new study by MIT researchers has found that “AI chatbots were surprisingly effective at convincing people to vote for a particular candidate or change their support for a particular issue,” reports Alex Wilkins for New Scientist. “Even for attitudes about presidential candidates, which are thought to be these very hard-to-move and solidified attitudes, the conversations with these models can have much bigger effects than you would expect based on previous work,” says Visiting Prof. David Rand. 

Researchers at MIT and elsewhere “examined how popular chatbots could change voters’ minds about candidates in the United States, Canada and Poland,” reports Will Oremus for The Washington Post. 

Prof. Kevin Chen and his colleagues have designed a tiny, insect-sized aerial microrobot that is “faster and more acrobatic than any of its predecessors,” reports Phie Jacobs for Science. The device, “which measures just 4 centimeters across and weighs less than a paperclip, flies almost five times faster and accelerates twice as quickly as existing microrobots,” explains Jacobs. “It can also execute sharp turns while enduring 160-centimeter-per-second wind gusts and—perhaps most impressively—can complete 10 consecutive somersaults in 11 seconds.” 

Prof. Sara Beery spoke at TED Radio Hour about her work developing Inquire, an AI tool aimed at supercharging ecosystem conservation that is trained on millions of photos captured by citizen scientists, reports GBH. “Under the hood, what we’re doing is we’re developing AI models that can learn and understand similarities between images and scientific language,” explains Beery. 

A new study by researchers from MIT and elsewhere describes “agentic AI as a ‘new class of systems’ that ‘can plan, act, and learn on their own,’” reports Matt O’Brien for the Associated Press.

Wall Street Journal reporter Angel Au-Yeung spotlights Anysphere, an AI startup founded by Michael Truell '21, Sualeh Asif '22, Arvid Lunnemar '22, and Aman Sanger '22. “The company makes an AI tool that learns a developer’s coding style to help autocomplete, edit and review lines of code,” writes Au-Yeung.