New York City transit officials are exploring a controversial plan to use artificial intelligence software to track how many subway riders are wearing face masks, and where.
The technology, which is currently being used in Paris, was among a host of ideas presented in a consultant’s report released to the public on Monday that could help transit authorities measure the level of face mask compliance at specific subway stations. Commissioned by the Metropolitan Transportation Authority in May, the 41-page document details the best practices from transit systems around the world in combatting the spread of COVID-19. The list includes several high-tech tools like thermal-scanner temperature checks, which has been adopted in Canada and Singapore, as well as UV lamps and robots that China has deployed on buses to kill the viruses on surfaces.
“We’re exploring the feasibility of a wide range of tools and approaches for helping keep our employees and customers safe,” said Andrei Berman, a spokesman for the MTA, in a statement. “AI is one of those tools and we’ll continue to research whether it might be effective, and if so, how it might be deployed in an appropriate manner to continue ensuring best public health practices are followed for the safety of our customers and employees.”
Given its potential to be weaponized, the use of artificial intelligence to scan rider footage at subway stations would most likely rekindle a battle around privacy and transparency. But the latest debate over AI is more complicated, as New Yorkers struggle to resume their lives amid the shadow of a pandemic that killed more than 22,000 New York City residents.
“We have to juggle legitimate privacy concerns and concerns about public safety and public health,” said Nick Sifuentes, the executive director of Tri-State Transportation Campaign.
According to a recent report produced by the group, Paris, Tokyo and Vienna have all seen a high level of face mask compliance among riders. At the same time, there have not been any outbreaks in those cities linked to mass transit. According to Sifuentes, the evidence suggests that the MTA needs to prioritize mask enforcement over arguably more expensive and unreliable technology like temperature scans.
“The goal is 100 percent mask compliance,” he said, adding, “The public health threat that you pose is certainly much greater than the privacy threat as a noncompliant individual.”
The report from Tri-State Transportation Campaign recommends that the MTA use its current CCTV system to manually track mask use by passengers, but Sifuentes said installing AI software would likely be more cost-effective.
Supporters say that concerns about privacy could be addressed by the MTA using anonymized data, in which personally identifiable information is removed. Such is the case in the Paris metro system, where the software is not used to punish individual riders, but rather to collect data that will help city officials anticipate future outbreaks. The technology can also be used to measure the level of mask wearing at specific locations, allowing transit officials to direct resources to stations with low compliance.
“The goal is just to publish statistics of how many people are wearing masks every day,” Xavier Fischer, the CEO of the maker of of the French software system DatakaLab told the Verge. “We never sell for security purposes.”
Tensions over surveillance have arisen with other coronavirus technologies, most notably digital tracking apps that have been used in countries like China and South Korea to monitor infected individuals and trace their contacts. However, the fact that the French, who have strong notions of civil liberties, have accepted an AI-based system has signaled to some that there can be an acceptable form of surveillance in the service of protecting the public.
But critics of surveillance remain skeptical.
In 2019, the MTA acknowledged in a Wall Street Journal story that it was running a pilot program to record and identify faces of drivers driving through the Robert F. Kennedy Bridge, the initial efforts of which were not successful. A spokesman said the data was only being used for security.
“The MTA really has a trust gap here,” said Albert Fox Cahn, the founder and executive director of Surveillance Technology Oversight Project (STOP), a nonprofit advocacy group that has fought to get city agencies to disclose how they are using facial recognition.
STOP recently sued the MTA over its refusal to provide information about a video monitor installed in Times Square to deter fare evasion. MTA officials maintained that the camera was not using facial recognition technology, but Cahn demanded to see internal agency documents about the camera’s installation. In May, a state Supreme Court judge ruled that the agency had wrongly denied its Freedom of Information Law request without explanation.
Cahn warned that the use of surveillance technology would give way to a predictable result: more police encounters as well as needless arrests and violence. He pointed to how the MTA’s cameras were being used by NYPD to locate and remove homeless individuals in subway stations.
“Excessive surveillance is a matter of life and death,” he said. “I’m terrified that we will see riders of color singled our by AI and arrested for not wearing masks.”
Early on, data from the city showed large racial disparities in the social distancing and mask wearing policing of New Yorkers. In mid-May, Mayor Bill de Blasio announced that the NYPD would no longer enforce the face mask rule after a video of a Brooklyn mother getting handcuffed in a subway station for allegedly not wearing her mask went viral.
Danny Pearlstein, the policy and communications director for the advocacy group Riders Alliance, worried that the use of AI software would only make riders feel “less welcome, less inclusive and ultimately less safe.”
He also argued that an overwhelming majority of New Yorkers have in fact embraced face masks, which he said was borne out by the data. Over the last week, New York City’s coronavirus positivity rate has remained steady at two percent.
But the challenge for the MTA will be in managing a growing number of mass transit users as the city continues to reopen. Subway ridership surpassed 1 million on Tuesday, June 23rd, a more than 150 percent increase since April. Prior to the pandemic, weekday subway ridership was around 5.5 million.
Not surprisingly, MTA officials have stressed mask wearing over social distancing.
“The key is going to be mask vigilance,” Sarah Feinberg, interim head of the New York Metropolitan Transportation Authority, recently said in an interview. Earlier this month, in response to the agency’s request for extra police, the mayor deployed 800 school safety officers to hand out masks at subway stations.
Pearlstein nonetheless has urged the MTA to look more broadly at other solutions, like increasing service frequency, especially during off-peak hours.
In terms of safety and relieving tensions, he said that giving passengers even just a little extra breathing room would be better than resorting to surveillance technology.
Ultimately, he argued, “Using machines that can profile riders will make people less comfortable than giving them an extra six inches for their elbow.”
