Computer Science and Artificial Intelligence Laboratory (CSAIL)

A new study by MIT researchers demonstrates how “machine-learning systems designed to spot someone breaking a policy rule—a dress code, for example—will be harsher or more lenient depending on minuscule-seeming differences in how humans annotated data that were used to train the system,” reports Ananya for Scientific American. “This is an important warning for a field where datasets are often used without close examination of labeling practices, and [it] underscores the need for caution in automated decision systems—particularly in contexts where compliance with societal rules is essential,” says Prof. Marzyeh Ghassemi.

Forbes reporter Rob Toews spotlights Prof. Daniela Rus, director of CSAIL, and research affiliate Ramin Hasani and their work with liquid neural networks. “The ‘liquid’ in the name refers to the fact that the model’s weights are probabilistic rather than constant, allowing them to vary fluidly depending on the inputs the model is exposed to,” writes Toews.

Venti Technologies, which was co-founded by MIT researchers and alumni, is working to build autonomous vehicles for industrial and global supply chain hubs, reports Bruce Rogers for Forbes. “Working with the world's leading port operator provides Venti the opportunity to bring the economics of autonomous vehicles to over 60 ports globally,” writes Rogers. “These ports operate 24/7 requiring 2-3 shifts of human drivers.”

Nature contributor David Chandler writes about the late Prof. Edward Fredkin and his impact on computer science and physics. “Fredkin took things even further, concluding that the whole Universe could actually be seen as a kind of computer,” explains Chandler. “In his view, it was a ‘cellular automaton’: a collection of computational bits, or cells, that can flip states according to a defined set of rules determined by the states of the cells around them. Over time, these simple rules can give rise to all the complexities of the cosmos — even life.”

Using techniques inspired by kirigami, a Japanese paper-cutting technique, MIT researchers have developed a “a novel method to manufacture plate lattices – high performance materials useful in automotive and aerospace designs,” reports Andrew Paul for Popular Science. “The kirigami-augmented plate lattices withstood three times as much force as standard aluminum corrugation designs,” writes Paul. “Such variations show immense promise for lightweight, shock-absorbing sections needed within cars, planes, and spacecraft." 

Prof. Yoon Kim speaks with Popular Science reporter Charlotte Hu about how large language models like ChatGPT operate. “You can think of [chatbots] as algorithms with little knobs on them,” says Kim. “These knobs basically learn on data that you see out in the wild,” allowing the software to create “probabilities over the entire English vocab.”

Graduate students Martin Nisser and Marisa Gaetz co-founded Brave Behind Bars, a program designed to provide incarcerated individuals with coding and digital literacy skills to better prepare them for life after prison, reports Morgan Radford for MSNBC. Computers and coding skills “are really kind of paramount for fostering success in the modern workplace,” says Nisser.

Prof. D. Fox Harrell writes for The Guardian about the importance of ensuring AI systems are designed to “reflect the ethically positive culture we truly want.” Harrell emphasizes that: “We need to be aware of, and thoughtfully design, the cultural values that AI is based on. With care, we can build systems based on multiple worldviews – and address key ethical issues in design such as transparency and intelligibility."

Prof. Daniela Rus, director of CSAIL, speaks with TechCrunch reporter Brain Heater about liquid neural networks and how this emerging technology could impact robotics. “The reason we started thinking about liquid networks has to do with some of the limitations of today’s AI systems,” says Rus, “which prevent them from being very effective for safety, critical systems and robotics. Most of the robotics applications are safety critical.”

MIT researchers have developed a new tool called “PhotoGuard” that can help protect images from AI manipulation, reports Ross Cristantiello for Boston.com. The tool “is designed to make real images resistant to advanced models that can generate new images, such as DALL-E and Midjourney,” writes Cristantiello.

Researchers at MIT have developed “PhotoGuard,” a tool that can be used to protect images from AI manipulation, reports Catherine Thorbecke for CNN. The tool “puts an invisible ‘immunization’ over images that stops AI models from being able to manipulate the picture,” writes Thorbecke.

A number of MIT alumni including Elaheh Ahmadi, Alexander Amini, and Jose Amich have been named to the Forbes 30 Under 30 Local Boston list.

Researchers at MIT and Dana-Farber Cancer Institute have published a paper showcasing the development of OncoNPC, an artificial intelligence model that can predict where a patient’s cancer came from in their body, reports Tony Ho Tran for The Daily Beast. This information “can help determine more effective treatment decisions for patients and caregivers,” writes Tran.

At CSAIL’s Imagination in Action event, CSAIL research affiliate and MIT Corporation life member emeritus Bob Metcalfe '69 showcased how the many individual bits of innovation that emerged from the Telnet Protocol later become the foundation for email, writes Prof. Daniela Rus, director of CSAIL, for Forbes. “Looking ahead to the future of connectivity, Metcalfe spoke of the challenges of limited network bandwidth, and the importance of keeping connectivity firmly in mind when developing any new computing technologies,” writes Rus.

Prof. Jacob Andreas explored the concept of language guided program synthesis at CSAIL’s Imagination in Action event, reports research affiliate John Werner for Forbes. “Language is a tool,” said Andreas during his talk. “Not just for training models, but actually interpreting them and sometimes improving them directly, again, in domains, not just involving languages (or) inputs, but also these kinds of visual domains as well.”