Experts examine the economic and social ramifications of COVID during a Stanford HAI conference.
Global disease curves have flattened. U.S. states and cities have lifted shelter-in-place restrictions and begun to reopen their economies.
Two months into the COVID-19 crisis, we are beginning to see signs of progress and hope.
But with uncertainty about everything from vaccines to workplace protocols, we are far from done with this pandemic. The question is as much about how to return to “normal” life as when.
On June 1, 2020, the Stanford Institute for Human-Centered Artificial Intelligence (HAI) held its second virtual conference on COVID, bringing together experts from across disciplines to discuss the way forward, with focus on the intersection of social, economic, political, medical, and technological domains.
“It’s one of the most difficult periods we’ve faced as a nation,” says co-moderator Rob Reich, Stanford professor of political science and HAI associate director. Fellow co-moderator and HAI associate director Russ Altman, Stanford professor of bioengineering, genetics, medicine, biomedical data science, and computer science, says, “We need to understand implications beyond the immediate crisis and create pathways out.”
Speakers addressed topics ranging from workplace COVID screening and automation to 2020 election preparation — while AI and other technologies are topics of focus, speakers emphasized the broad context and implications of the current global challenge.
Here are their key insights.
The Workplace of the Future
One major concern of this pandemic is how we return safely and productively to work, with ongoing fears of COVID transmission. “We need to provide policymakers a broad view on what’s happening in their jurisdictions with regard to all aspects of health, to help them make decisions,” says Abigail Wozniak, director of the Opportunity and Inclusive Growth Institute, Federal Reserve Bank of Minneapolis.
Specifically, leaders need to plan for large-scale absences, as those with COVID-19 diagnoses or symptoms avoid on-site work. To address this, Wozniak is co-leading a COVID impact survey of 9,500 workers across 10 states to understand symptom rates across communities. Getting a handle on these will help officials craft policies to anticipate and support those unable to work.
Among other questions, survey respondents are asked to record their temperature or self-report fevers and other symptoms. Initial results suggest that on any given day about 3 to 4 percent of the working population would be screened out based on temperature readings (>99oF) and a self-reported symptom set, with that figure rising to 12 percent when it includes those with any target symptoms.
“That’s a lot of people screened out daily,” Wozniak says, noting that not all screenings pick up the same workers. Even if it’s closer to the lower end of estimates, policies must incentivize testing and transparency, such as direct government support for those who must stay away from work for extended periods; legislation like the CARES Act provides some of this today, but more is needed.
“The bottom line is that policy should be aimed at keeping people safe,” Wozniak concludes.
Labor Market Evolution
Even if people can return to work, will there be jobs waiting for them? If so, what kind?
These are questions that have motivated research by Lisa Kahn, professor of economics at the University of Rochester. “There has been a massive spike in unemployment,” she says. “It’s unprecedented, with 40 million out of work. But we need to know more details around that.”
She’s part of a team getting a handle on the labor market’s evolution under COVID. The collaborators have used Burning Glass Technologies’ AI-based tools to scrape job-vacancy ads from Monster, CareerBuilder, and other employment sites to understand employment trends.
“Job vacancy postings are forward-looking,” Kahn says. “It’s what employers think they’ll need.” The team found that vacancy postings plummeted as unemployment claims spiked — regardless of a given region’s stay-at-home policy.
Similarly, the hardest-hit industries — those with fewer job postings — have been restaurants, nonessential retail such as clothing, and other customer-facing sectors, whereas demand has spiked for workers in essential retail like groceries. “Frontline jobs are sheltered,” Kahn says.
While reopening has resulted in slight rebounds in vacancies, including in hard-hit sectors, “overall it has made little difference,” according to Kahn. That makes clear that we can’t just “flip a switch” to return to a more productive economy.
“Getting the virus under control will help people feel more confident about hiring,” Kahn says. “But we’ll likely see widespread worker reallocation,” to the industries that need them most.
Another trend we’ll see in global workplaces is automation.
The use of machines at work is nothing new. But the trend is likely to rise exponentially, accelerated by COVID. Specifically, along with acceleration in remote work has come large-scale adoption of new technologies — there were 10 million downloads of machine-learning (ML) applications from Google’s TensorFlow open-source software library in the past month alone.
“ML can’t do everything,” he says, but what it can do, they found, is fill in for a disproportionate number of lower-wage jobs, like cashiers. Some higher-paid jobs will also be affected, for example, pilots, and we’re more likely to see ML applied in manufacturing and retail than education services. Workplaces should be thinking now about how to reinvent and evolve jobs to be less vulnerable to takeover by ML, says Brynjolfsson.
Technology will also enable more remote work. A survey study Brynjolfsson co-led showed about half of Americans are working from home — up from about one-sixth pre-COVID. “Many will continue working from home after the pandemic,” he says. “Executives are happy with the productivity and likely to reassess their leases.”
“The future is already here,” Brynjolfsson suggests in summary. “The winners will be those who think ahead and get the right skills in place” — and capture the most value from remote work.
What Makes an Effective COVID Response
The U.S.’s strategy to address the pandemic has been far from optimal, many speakers agreed. Several framed the big-picture COVID challenge that the government and other organizations continue to face.
“We are witnessing a legitimacy crisis,” says Danielle Allen, Harvard professor and director of the school’s Edmond J. Safra Center for Ethics. She points to the recurrent failure of public leadership and institutions since the pandemic began and lays out a set of prerequisites for an effective response.
“The single most important feature is a valid and viable social contract,” Allen says. “That includes commitment of citizens to one another and of the government to the entirety of its citizenship.” She says that we as a country should not accept abandoning segments of the population, such as the elderly or essential workers.
Upholding the social contract is part of a broad effort to “deliver safety and happiness to all,” Allen says. She emphasizes development of more integrative protection of “lives, liberties, and livelihoods” under COVID, pointing to the successful examples of fellow democracies Germany and Singapore.
More partnerships with community organizations — such as to support testing and contact-tracing — should be part of the broad strategy of disease suppression. “Such organizations can work with communities of affinity and build trust,” Allen says. “It’s ultimately about empowering communities and abandoning no one.”
Preparing for the 2020 Elections
That sentiment applies well to the upcoming presidential election.
“Elections are tests of a nation’s ability to communicate messages of legitimacy,” says Nathaniel Persily, Stanford professor of law. “It’s the absolute worst time for a pandemic.”
Because COVID has hindered “analog” means of campaigning and voting, we will rely on more remote, technology-enabled means, he notes; “That requires shifting tens of millions of voters from the old way to a new, unfamiliar one.”
Broad solutions will also include increased voting by mail and retrofitting polling stations to promote social distance. The process will inevitably be complex and controversial. Part of the challenge is a highly decentralized electoral system, with many responsibilities held by individual jurisdictions — “a patchwork quilt of regulations and opportunities,” Persily says.
Beyond governance issues, there’s potential for racial bias; for example, mail-in votes tend to be used more by white, highly educated voters. Moreover, challenges include providing the people (poll workers are generally older volunteers more at risk for COVID), the places (senior centers are typical polling stations), and the things (voting machines, ballots); states may even compete for resources.
“We need to pursue education and outreach,” along with exploring novel approaches including curbside voting options. Persily is working on such ideas as part of a collaboration between Stanford Law School and MIT.
The Government and Privacy Protection
Whoever wins in 2020 will have to assume key obligations for COVID-management; chief among these is protection of citizens’ privacy.
“The pandemic is a litmus test for democracies to practice what they preach,” says Marietje Schaake, international policy director of Stanford’s Cyber Policy Center and a former European Parliament member. “They have to be extra mindful of civil liberties when there is such urgency to act fast.”
She points to increasing worldwide proposals of technology-enabled solutions, such as contact-tracing apps. These apps should ensure privacy, she notes, and we need to understand their longer-term impact. In line with this, Schaake was among legislators and others who signed a letter urging the Dutch government to assess contact-tracing app implications fully.
For example, she suggests such apps should have a “sunset clause”: “We can’t let their scope creep from supporting doctors to deciding who gets on plane or gets insurance.” Similarly, use of the apps should be optional. Alternative models such as manual, diary-based tracing are also worth considering, in part for the economic boost they represent with the hiring of human tracers.
“Trust is key,” Schaake says. “Trust in the technology but also in the government. This will continue to be tested and scrutinized. Technology can help, but it’s not a panacea.”