Forbes reporter Tracey Follows spotlights the MIT Media Lab’s Advancing Humans with AI (AHA) project, a “new research program asking how can we design AI to support human flourishing.” At the launch event, Prof. Sherry Turkle “raised a specific and timely concern: what is the human cost of talking to machines that only pretend to care?”
Researchers at MIT have concluded that AI does not develop “value systems” over time, reports Kyle Wiggers for TechCrunch. “For me, my biggest takeaway from doing all this research is to now have an understanding of models as not really being systems that have some sort of stable, coherent set of beliefs and preferences,” says graduate student Stephen Casper. “Instead, they are imitators deep down who do all sorts of confabulation and say all sorts of frivolous things.”
Graduate students Chuck Downing and Zhichu Ren PhD '24 highlight the potential uses of AI in the research process, reports Amanda Heidt for Nature. “I didn’t know much going in, but I learnt quite a bit, and so I use these deep dives all the time now,” says Downing on using deep-research tools when approaching unfamiliar topics. “It’s better than anything else I’ve used so far at finding good papers and in presenting the information in a way that I can easily understand.”
Prof. Pulkit Agrawal speaks with NPR Short Wave host Regina Barber and science correspondent Geoff Brumfiel about his work developing a new technique that allows robots to train in simulations of scanned home environments. “The power of simulation is that we can collect very large amounts of data,” explains Agrawal. “For example, in three hours' worth of simulation, we can collect 100 days' worth of data.”
Prof. David Sontag, Monica Agrawal PhD '23, Luke Murray SM '22, and Divya Gopinath '19, MEng '20 co-founded Layer Health - an AI healthcare startup that is applying large language models (LLMs) to help clinicians with medical chart reviews and data abstraction, reports Seth Joseph for Forbes. “The same chart review problem we’re solving with our clinical registry module is faced by clinicians at the point of care,” says Sontag. “For example, one of our next modules will focus on real-time clinical decision support to help automate clinical care pathways, leading to more reliable, high-quality care."
A new study by researchers at MIT explores how AI chatbots can impact people’s feelings and mood, reports Matthew Gault for Gizmodo. “One of the big takeaways is that people who used the chatbots casually and didn’t engage with them emotionally didn’t report feeling lonelier at the end of the study,” explains Gault. “Yet, if a user said they were lonely before they started the study, they felt worse after it was over.”
Researchers at MIT and elsewhere have found that “heavy users of ChatGPT tend to be lonelier, more emotionally dependent on the AI tool and have fewer offline social relationships,” reports Rachel Hall for The Guardian. “The researchers wrote that the users who engaged in the most emotionally expressive personal conversations with the chatbots tended to experience higher loneliness – though it isn’t clear if this is caused by the chatbot or because lonely people are seeking emotional bonds,” explains Hall.
Graduate student Cathy Fang speaks with CBS News reporter Lindsey Reiser about her research studying the effects of AI chatbots on people’s emotional well-being. Fang explains that she and her colleagues found that how the chatbot interacts with the user is important, “but also how the user interacts with the chatbot is equally important. Both influence the user’s emotional and social well-being.” She adds: “Overall, we found that extended use is correlated with more negative outcomes.”
Researchers at MIT and elsewhere have found “that frequency chatbot users experience more loneliness and emotional dependence,” reports Beatrice Nolan for Fortune. “The studies set out to investigate the extent to which interactions with ChatGPT impacted users’ emotional health, with a focus on the use of the chatbot’s advanced voice mode,” explains Nolan.
[Originally in Spanish] MIT researchers have developed a new technique to educate robots by increasing human input, reports Uriel Bederman for TN Tecno. “We can’t expect non-technical people to collect data and fine-tune a neural network model," explains graduate student Felix Yanwei Wang. "Consumers will expect the robot to work right out of the box, and if it doesn’t, they’ll want an intuitive way to customize it. That’s the challenge we’re addressing in this work."
Prof. David Autor speaks with Financial Times reporter Tom Davis about the impact of dataism – the belief that through gathering increasing amounts of information businesses can make the right decisions and create value - and automation in business management. “You can think of automation as a machine that takes a job’s inputs and does it for the worker,” says Autor. “[And] augmentation as a technology that increases the variety of things that people can do, the quality of things people can do, or their productivity”.
Prof. Eric So speaks with Financial Times reporter Seb Murray about the use of AI in business programs. “I’ve seen a mixture of surprise, enthusiasm, concern and trepidation,” explains So. “It’s quite difficult to design assignments that can be done without AI. I suspect much of our curriculum will be redesigned from the ground up.”
Prof. Behnaz Farahi and her team have created “Gaze to the Stars,” an art installation that features video projections of eyes onto the MIT Dome while sharing stories of aspiration, struggle, longing, and hope, reports Emily Maher for WCVB. “Farahi and her team created a space, a pod, where people looked into a screen of stars as their eyes were scanned,” explains Maher. “Next, an AI voice began encouraging them to share their stories.”
A new proposal by graduate student Shayne Longpre and other AI researchers suggests “a new scheme supported by AI companies that gives outsiders permission to probe their models and a way to disclose flaws publicly,” reports Will Knight for Wired. “The authors suggest three main measures to improve the third-party disclosure process: adopting standardized AI flaw reports to streamline the reporting process; for big AI firms to provide infrastructure to third-party researchers disclosing flaws; and for developing a system that allows flaws to be shared between different providers,” writes Knight.
Writing for Fast Company, graduate student Sheng-Hung Lee and Devin Liddell of Teague highlight four types of AI technologies that could aid senior citizens in their homes. “To better understand how seniors want AIs and robots to help in their homes, we asked them,” they write. “We recruited seniors from the MIT AgeLab’s research cohort—each around 70 years old and in the early stages of retirement—and then engaged in wide-ranging conversations about their aspirations and fears about these technologies.”