Machine learning

Prof. Tod Machover’s composition “already and not yet,” was performed by members of the Boston Symphony Orchestra along with an AI system that listens to musicians and adds improvisations, reports A.Z. Madonna for The Boston Globe. “Music is something that’s meant to plumb your soul, and have you discover things,” says Machover. The composition is “about my view of where we are, how to take a deep breath and figure out, gather strength to go somewhere else, and then what a possible future of democracy might be.”  

Prof. Tod Machover speaks to WBUR’s Andrea Shea about “already and not yet” an interactive composition he wrote for the Boston Symphony Orchestra and an AI instrument developed by Machover and graduate student Manaswi Mishra. Machover explains that they designed the AI tools so that they’re “always going to make a kind of variation, a coloration, an extension of what the musician's playing – but exactly what it's going to do I do not have complete control over. I'm trying to show that it [AI] can be integrated in a beautiful, provocative and interesting way in a serious piece of music.”  

Instead of giving AI agents a specific task, MIT researchers put 13 agents, named Eko, alone on a digital desert island for 30 hours, and witnessed the technology playing instead of simply wandering around, writes Fast Company’s Jude Cramer. “The researchers compared Eko’s behavior to common elements of play in animals, including humans,” notes Cramer. “These signs of play include creating their own rules, repeating behaviors with slight variations, setting goals for themselves, and transforming objects and events through make-believe.” 

After observing the behavior of 13 AI agents on a simulated desert island, MIT researchers suggest that allowing agents to explore “without a goal” could be an effective way to teach AI agents new skills, reports Aditya Mehta for TechCrunch. “We normally develop artificial agents by deciding in advance what they should practice, what they should optimize, and what counts as progress,” the researchers write. “Play suggests another regime: give an agent a sufficiently rich and safe world, let it choose what is interesting, and allow useful knowledge to accumulate from the history that follows.” 

Flapping Airplanes, a startup co-founded by Benjamin Spector ’22, MEng ’22, has developed a framework for AI models to run efficiently while being trained on less data and in a single graphics processing unit (GPU), writes Julie Bort for TechCrunch. The framework “can run a task that spends most of its time waiting for data to be fetched alongside one that spends most of its time doing math, so the chip is rarely sitting idle,” notes Bort. “The premise is that AI has barely scratched the surface of what GPUs can do, and they believe it can create intelligent systems ‘without ingesting half the internet.’”  

Prof. Daron Acemoglu joins “The Claman Countdown” on Fox Business to discuss the possibility that largescale deployment of AI could triple unemployment within the next decade. “There is so much uncertainty about AI,” says Acemoglu. “It’s very difficult to know exactly when we’re going to see the productivity gains, which will come, and when we’re going to see job losses, and whether we’ll actually see some job creation as well. But on the current path where AI is being prepared and developed as a labor replacing technology, I would be very worried that within the next ten years we’ll see a lot of jobs disappear.”  

Postdoctoral Associate Iman YeckehZaare speaks to El País reporter Jordi Pérez Colomé about his research on RentAHuman, a platform that allows AI agents to hire people to do physical labor. “RentAHuman has shown that many people are willing to sign up to be hired by AI agents. What it still hasn’t shown is that many AI agents are doing the hiring,” says YeckehZaare. “Listings posted by agents will increase as AI agents become capable of acting and paying on their own.”  

Prof. Jacob Andreas speaks with CNN’s Lisa Eadicicco about deciphering the chain-of-thought behind complex AI neural networks. “I think we’re still very far from being able to claim that we have any sort of fine-grained understanding of how the model is actually arriving at a solution,” says Andreas.  

Prof. Emilio Castilla discusses the factors companies must consider when deploying AI to help in the hiring process, in an article by USA Today’s Greta Reich. “We should evaluate not only whether the algorithm predicts historical hiring decisions, but whether it helps identify candidates who subsequently perform well in the job,” writes Castilla. “The most exciting possibility is not simply that AI might help us find the best talent more efficiently. It is that AI might help us recognize talent that our traditional hiring practices have systematically overlooked.”  

In The Boston Globe’s “Starting Point Newsletter,” reporter Anna Kusmer features Prof. Sherry Turkle’s insights from her latest book “Artificial Intimacy: Who We Become When We Talk to Machines,” including the risks that AI companionship poses to school-aged children. “Children are bringing the world inside; they’re bringing the people around them inside them to create an inner life,” says Turkle. “And when you talk to a chatbot, there is nothing to bring inside. There is no caring person.” 

In an interview with Wired’s Steven Levy, Prof. Sherry Turkle discusses how she witnessed chatbots with human-like features becoming “dangerously enmeshed” with people, during research for her latest book “Artificial Intimacy: Who We Become When We Talk to Machines.” “In every serious interview I do, people invoke this myth about practice—that you use chatbots to practice for talking to your boss, your lover, your teacher,” says Turkle. “But when you’re bathing in the warm soup of total acceptance, talking to chatbots becomes the endgame.” 

MIT researchers gave 13 AI agents, named Eko, free rein, on a simulated desert island for 30 hours, and observed as the agents played and learned from their surroundings like humans, writes Gizmodo’s Webb Wright. “We had this mindset to have as [few] constraints as possible, because we really wanted to see what the behavior of these autonomous agents is when you don’t really tell them what to do,” says graduate student Nathan Cloos. “Play is really about curiosity or novelty-seeking behavior, and this is essentially what Eko does,” Prof. Daniela Rus adds.  

Testing 1,000 synthetically generated client profiles against AI chatbots for investment advice, Prof. Andrew Lo and Research Affiliate Jillian Ross identified “heuristic collapse,” where chatbots based their recommendations almost entirely on self-reported risk tolerance, while ignoring income and other variables, writes Damilola Esebame for The Street. “Lo has recommended that investors use chatbots for orientation, leveraging them to frame questions, map out available options, and prepare for conversations with a human advisor rather than treating the output as a finished plan,” notes Esebame. 

Maxime Cohen PhD ‘15 speaks to Forbes’ Karl Moore about his latest book “Pricing in the Age of AI,” that explores how AI is transforming pricing decisions for businesses, consumers and society. “We may be entering an era of agent-to-agent commerce, in which AI systems negotiate directly with one another on behalf of consumers and businesses,” says Cohen. “This could fundamentally reshape how prices are determined and how transactions take place.” 

Prof. Daron Acemoglu joins CNBC’s “Closing Bell” to discuss how AI could cause mass job displacement due to companies deploying rapidly advancing technologies. “AI’s capabilities are real and powerful, but human jobs are also complex. I don’t think you want AI to take over air traffic controller jobs right now; you don’t want AI to do CEO jobs,” says Acemoglu. “What we are seeing when these models escape their sandboxes and create havoc—it's not that they are necessarily super intelligent, it is that they don’t have judgement and they are optimized for certain narrow metrics and when you let them loose, they are going to do lots of unforeseeable things. You don’t want that in many jobs.”