Machine learning

Researchers from MIT and elsewhere have used quantitative and computational methods to analyze animal communication, reports Emily Anthes for The New York Times.

An analysis by MIT researchers has identified “wide-ranging instances of AI systems double-crossing opponents, bluffing and pretending to be human,” reports Hannah Devlin for The Guardian. “As the deceptive capabilities of AI systems become more advanced, the dangers they pose to society will become increasingly serious,” says postdoctoral associate Peter Park. 

MIT researchers have developed a new procedure that uses game theory to improve the accuracy and consistency of large language models (LLMs), reports Steve Nadis for Quanta Magazine. “The new work, which uses games to improve AI, stands in contrast to past approaches, which measured an AI program’s success via its mastery of games,” explains Nadis. 

MIT researchers have used advancements in machine learning and computing to help decode whale vocalizations, reports Sarah Kuta of Smithsonian Magazine. “If researchers knew what sperm whales were saying, they might be able to come up with more targeted approaches to protecting them,” Kuta explains. “In addition, drawing parallels between whales and humans via language might help engage the broader public in conservation efforts.”

A new analysis of years of vocalizations by sperm whales in the eastern Caribbean has provided a fuller understanding of how whales communicate using codas, reports Will Dunham of Reuters. Graduate student Pratyusha Sharma explained that: "The research shows that the expressivity of sperm whale calls is much larger than previously thought."

New Scientist reporter Clare Wilson writes that a new analysis by MIT researchers of thousands of exchanges made by east Caribbean sperm whales demonstrates a communication system more advanced than previously thought. “It’s really extraordinary to see the possibility of another species on this planet having the capacity for communication,” says Prof. Daniela Rus.

Researchers from MIT and elsewhere have uncovered a phonetic alphabet used by sperm whales, which provides “key breakthroughs in our understanding of cetacean communication,” reports Brain Heater for TechCrunch. “This phonetic alphabet makes it possible to systematically explain the observed variability in the coda structure,” says Prof. Daniela Rus, director of CSAIL. “We believe that it’s possible that this is the first instance outside of human language where a communication provides an example of the linguistic concept of duality of patterning. That refers to a set of individually meaningless elements that can be combined to form larger meaningful units, sort of like combining syllables into words.”

Associated Press reporter Maria Cheng spotlights a new study by MIT researchers that identifies a “phonetic alphabet” used by whales when communicating. “It doesn’t appear that they have a fixed set of codas,” says graduate student Pratyusha Sharma. “That gives the whales access to a much larger communication system.” 

Using machine learning, MIT researchers have discovered that sperm whales use “a bigger lexicon of sound patterns” that indicates a far more complex communication style than previously thought, reports Lauren Sommers for NPR. “Our results show there is much more complexity than previously believed and this is challenging the current state of the art or state of beliefs about the animal world," says Prof. Daniela Rus, director of CSAIL. 

MIT researchers have discovered that sperm whales use a “much richer set of sounds than previously known, which they call a ‘sperm whale phonetic alphabet,’” reports Carl Zimmer for The New York Times. “The researchers identified 156 different codas, each with distinct combinations of tempo, rhythm, rubato and ornamentation,” Zimmer explains. “This variation is strikingly similar to the way humans combine movements in our lips and tongue to produce a set of phonetic sounds.”

Prof. Yoon Kim speaks with USA Today reporter Eve Chen about how AI can be used in everyday tasks such as travel planning. “AI is generally everywhere,” says Kim. “For example, when you search for something – let’s say you search for something on TripAdvisor, Hotels.com – there is likely an AI-based system that gives you a list of matches based on your query.” 

 A new report by Principal Research Scientist Andrew McAfee explores the “implications of generative AI in economic growth, looking at everything from its possible effects on job skills and wages to how it may transform entire industries to its potential risks and pitfalls,” reports Sheryl Estrada for Fortune.

MIT researchers have “discovered hitherto unknown space molecule while investigating a relatively nearby region of intense star birth,” reports Robert Lea for Space.com. This discovery “revealed the presence of a complex molecule known as 2-methoxyethanol, which had never been seen before in the natural world, though its properties had been simulated in labs on Earth,” writes Lea.

MIT researchers have developed a machine-learning accelerator chip to make health-monitoring apps more secure, reports Aman Tripathi for Interesting Engineering. “The researchers subjected this new chip to intensive testing, simulating real-world hacking attempts, and the results were impressive,” explains Tripathi. “Even after millions of attempts, they were unable to recover any private information. In contrast, stealing data from an unprotected chip took only a few thousand samples.”

Bloomberg Opinion columnist Parmy Olson spotlights a new study by MIT researchers that finds AI chatbots can be highly persuasive when reinforced with facts and could potentially be used to help tackle conspiracy theories. “The scientists invited more than 2,000 people who believed in different conspiracy theories to summarize their positions to a chatbot — powered by OpenAI’s latest publicly available language model — and briefly debate them with the bot,” Olson writes. “On average, participants subsequently described themselves as 20% less confident in the conspiracy theory; their views remained softened even two months later.”