In an article for Gizmodo, Dell Cameron writes that graduate student Joy Buolamwini testified before Congress about the inherent biases of facial recognition systems. Buolamwini’s research on face recognition tools “identified a 35-percent error rate for photos of darker skinned women, as opposed to database searches using photos of white men, which proved accurate 99 percent of the time.”
Wired reporter Lily Hay Newman highlights graduate student Joy Buolamwini’s Congressional testimony about the bias of facial recognition systems. “New research is showing bias in the use of facial analysis technology for health care purposes, and facial recognition is being sold to schools,” said Buolamwini. “Our faces may well be the final frontier of privacy.”
Popular Science reporter Rob Verger writes that MIT researchers have developed a new mechanical system that can help humans lift heavy objects. “Overall the system aims to make it easier for people and robots to work together as a team on physical tasks,” explains graduate student Joseph DelPreto.
MIT and the U.S. Air Force “are teaming up to launch a new accelerator focused on artificial intelligence applications,” writes Danny Crichton for TechCrunch. The goal is that projects developed in the MIT-Air Force AI Accelerator would be “addressing challenges that are important to both the Air Force and society more broadly.”
Will Knight writes for MIT Technology Review about the MIT-Air Force AI Accelerator, which “will focus on uses of AI for the public good, meaning applications relevant to the humanitarian work done by the Air Force.” “These are extraordinarily important problems,” says Prof. Daniela Rus. “All of these applications have a great deal of uncertainty and complexity.”
The new MIT-Air Force AI Accelerator “will look at improving Air Force operations and addressing larger societal needs, such as responses to disasters and medical readiness,” reports Breanne Kovatch for The Boston Globe. “The AI Accelerator provides us with an opportunity to develop technologies that will be vectors for positive change in the world,” says Prof. Daniela Rus.
MIT researchers have identified a method to help AI systems avoid adversarial attacks, reports Matthew Hutson for Science. When the researchers “trained an algorithm on images without the subtle features, their image recognition software was fooled by adversarial attacks only 50% of the time,” Hutson explains. “That compares with a 95% rate of vulnerability when the AI was trained on images with both obvious and subtle patterns.”
Inside Higher Ed reporter Lindsay McKenzie writes that a new AI system developed by MIT researchers to summarize the findings of technical scientific papers could “be used in the near future to tackle a long-standing problem for scientists -- how to keep up with the latest research.”
Bloomberg News reporter Carol Massar spotlights how MIT researchers have developed a robot that can identify and sort recyclables. “The system includes a soft Teflon hand that uses tactile sensors to detect the size of an object and the pressure needed to grasp it,” Massar reports. “From there it can determine if it’s made of metal, paper or plastic.”
Researchers at MIT have found that adversarial examples, a kind of optical illusion for AI that makes the system incorrectly identify an image, may not actually impact AI in the ways computer scientists have previously thought. “When algorithms fall for an adversarial example, they’re not hallucinating—they’re seeing something that people don’t,” Louise Matsakis writes for Wired.
A study by MIT researchers examining adversarial images finds that AI systems pick up on tiny details in images that are imperceptible to the human eye, which can lead to misidentification of objects, reports Louise Matsakis for Wired. “It’s not something that the model is doing weird, it’s just that you don’t see these things that are really predictive,” says graduate student Shibani Santurkar.
Scientific American reporter Jeremy Hsu highlights how CSAIL researchers have developed a robot that can automatically sort recycling. The robot “uses soft Teflon ‘fingers,’ which have fingertip sensors to detect object size and stiffness,” Hsu explains.
MIT researchers have identified a new method to engineer neural networks in a way that allows them to be a tenth of the size of current networks without losing any computational ability, reports Avery Thompson for Popular Mechanics. “The breakthrough could allow other researchers to build AI that are smaller, faster, and just as smart as those that exist today,” Thompson explains.
WBUR reporter Pamela Reynolds highlights graduate student Joy Buolamwini’s piece, “The Coded Gaze,” which is currently on display as part of the “Avatars//Futures” exhibit at the Nave Gallery. Reynolds writes that Buolamwini’s piece “questions the inherent bias of coding in artificial intelligence, which has resulted in facial recognition technology unable to recognize black faces.”
In this video, Reuters explores how MIT researchers have developed a robot that can automatically sort recycling. The robot uses a pressure sensor to squeeze items to determine how they should be sorted.