A new study by Media Lab researchers finds that Amazon’s Rekognition facial recognition system performed more accurately when identifying lighter-skinned faces, reports Drew Harrell for The Washington Post. The system “performed flawlessly in predicting the gender of lighter-skinned men,” writes Harrell, “but misidentified the gender of darker-skinned women in roughly 30 percent of their tests.”
Verge reporter James Vincent writes that Media Lab researchers have found that the facial recognition system Rekognition performed worse at identifying an individual’s gender if they were female or dark-skinned. In experiments, the researchers found that the system “mistook women for men 19 percent of the time and mistook darker-skinned women for men 31 percent of the time,” Vincent explains.
MIT researchers have found that the Rekognition facial recognition system has more difficulty identifying the gender of female and darker-skinned faces than similar services, reports Natasha Singer for The New York Times. Graduate student Joy Buolamwini said “the results of her studies raised fundamental questions for society about whether facial technology should not be used in certain situations,” writes Singer.
New Yorker contributor Caroline Lester writes about the Moral Machine, an online platform developed by MIT researchers to crowdsource public opinion on the ethical issues posed by autonomous vehicles.
New York Times reporter Steve Lohr writes about the MIT AI Policy Conference, which examined how society, industry and governments should manage the policy questions surrounding the evolution of AI technologies. “If you want people to trust this stuff, government has to play a role,” says CSAIL principal research scientist Daniel Weitzner.
Prof. Max Tegmark speaks with Forbes contributor Peter High about his work trying to ensure that AI technologies are implemented in a way that is beneficial to society. “If we plan accordingly and steer technology in the right direction, we can create an inspiring future that will allow humanity to flourish in a way that we have never seen before,” says Tegmark.
Writing for The Boston Globe, members of the Media Lab’s Scalable Cooperation research group argue that independent oversight is needed to ensure that new AI technologies are developed in an ethical manner. “AI is the new framework of our lives,” they write. “We need to ensure it’s a safe, human-positive framework, from top to bottom.”
Writing for Forbes, Prof. David Mindell explores the concept of using work, in particular the duties a home health aide performs, as a Turing test for the abilities of AI systems. “In this era of anxiety about AI technologies changing the nature of work,” writes Mindell, “everything we know about work should also change the nature of AI.”
Prof. Daniela Rus and R. David Eldeman, director of the Project on Technology, Economy, and National Security at MIT speak with Matt Simon at Wired about working with robots. “The robots have a fixed architecture and they have a fixed vocabulary,” explains Rus. “So, people will continue to have to learn that and understand what the tool is useful for.”
Provost Martin Schmidt and SHASS Dean Melissa Nobles speak with Wall Street Journal reporter Sara Castellanos about MIT’s efforts to advance the study of AI and its ethical and societal implications through the MIT Stephen A. Schwarzman College of Computing. Schmidt says this work “requires a deep partnership between the technologists and the humanists.”
Prof. Erik Brynjolfsson speaks with Gizmodo reporter Brian Merchant about the 2018 AI Index report, which examines trends in the field of AI. Brynjolfsson says that when it comes to the impact of automation on the labor market, “developing countries are likely to be the hardest hit—they are the ones that depend most on low wages to compete in manufacturing.”
Boston Globe reporter Jessie Scanlon spotlights Prof. Regina Barzilay’s work developing machine learning systems that can identify patients at risk of developing breast cancer. Barzilay is creating “software that aims to teach a computer to analyze mammogram images more effectively than the human eye can and to catch signs of cancer in its earliest phases.”
Prof. Aleksander Madry and graduate student Anish Athalye speak with BBC News reporter Linda Geddes about how AI systems can be tricked into seeing or hearing things that aren’t actually there. “People are looking at it as a potential security issue as these systems are increasingly being deployed in the real world,” Athalye explains.
Gizmodo reporter Jennings Brown writes that researchers from the MIT Media Lab are developing a machine learning system that can develop addresses for regions of the planet that don’t have a recognized address system. Brown explains that the researchers “compared their results to an unmapped suburban region and found that their system labeled more than 80 percent of the populated portions.”
CSAIL researchers have developed a new technique to recreate paintings from a single photograph, reports John Biggs for TechCrunch. “The project uses machine learning to recreate the exact colors of each painting and then prints it using a high-end 3D printer that can output thousands of colors using half-toning,” Biggs explains.