What does the future of work look like?
How is AI changing the future of work? What digital-technology practices and policies will promote fairness and equality? How can government incorporate digital technologies to best serve constituents?
These and other questions took center stage at Stanford Digital Economy Lab’s inaugural event, the AI & the Future of Work Conference, on Oct. 27.
The virtual conference assembled a roster of visionary researchers, executives, and policy experts to share their perspectives on the impact of AI and other digital technologies. Stanford Digital Economy Lab director Erik Brynjolfsson hosted the conference, while executive director Christie Ko emceed the event.
Formed in June, the Stanford Digital Economy Lab focuses on four key research areas of the digital economy: the future of work, digital business models, data-driven decision making, and economy measurement. The lab is an initiative of the Stanford Institute for Human-Centered Artificial Intelligence (HAI) and is co-sponsored by the Stanford Institute for Economic Policy Research (SIEPR).
Stanford President Marc Tessier-Lavigne opened the conference by framing the university’s role in shaping the future of AI: “At Stanford, our approach is to put humans at the center of AI development to ensure that we are developing tools and policies that will serve, augment, and complement humanity, not replace or divide it.”
That’s largely why HAI launched the lab: to bring together world-class researchers, data sets, and industry partners to understand, measure, and shape the future of work.
Throughout the day, panelists painted a picture of a promising AI-influenced work landscape that relies heavily on decision makers to ensure broad, accessible benefits for all.
The Digitization of Everything
The fast-changing technology landscape served as the foundation for a number of panelist insights.
“We’ve seen the digitization of everything,” said Brynjolfsson, who is a senior fellow at SIEPR and HAI. “It has created a whole new set of economics and business models, winners and losers.”
The rise of more intelligent machines, he said, is part of a broad “AI awakening” in which machine learning and similar AI advances have become general-purpose technologies. While the “gold rush” to profitably apply AI tools has improved certain facets of life, such as shopping and productivity, it has also created disruption as machines increasingly replace human workers, especially for lower-income jobs.
“Digital processes make the economic pie bigger,” Brynjolfsson said, “but not everyone may benefit.” This is best represented by the ongoing “great decoupling,” in which AI and other technologies have improved productivity significantly. Yet only a small percentage of the population has reaped related benefits, with real median personal income remaining flat.
“We need the digital transformation to create more broadly shared prosperity,” Brynjolfsson said.
Technology and the Transformation of Work
“Technology is going through much faster iteration than in the past,” said Reid Hoffman, co-founder of LinkedIn and Greylock Ventures partner. He noted that new technologies can be harnessed to serve a much broader range of tasks and jobs, including traditionally white-collar ones. “The things that are most terrifying often create the most opportunity,” he said.
Hoffman said the goal for individuals and nations is to “speed toward technology” with everyone acting more entrepreneurially and “playing offense, not defense.” Chinese companies are embracing this mentality — including tech giant Tencent, which deployed multiple internal teams in a race to create the popular WeChat app. It’s a lesson that U.S. businesses can learn from, Hoffman said.
James Manyika, a senior partner at McKinsey, agreed that AI is driving rapid change: “About 60 percent of all occupations have about a third of tasks that are automatable, and new jobs are being created.” He also echoed the sentiment that inequality has grown, with highly skilled and educated workers doing well, employment tenure decreasing (more gig and contract work), and in many cases wages declining. The COVID pandemic has exacerbated the situation.
Both speakers see multifaceted, strategic solutions to future-of-work challenges. “We can use AI to solve problems it might create, such as providing digital-skill tutoring,” Hoffman said.
It’s also about an orientation toward learning and training. “Lifelong learning is a key theme,” Manyika said. “We need to create more incentives for on-the-job training and other human capital development, similar to those for traditional capital investment.”
Policy Challenges and Solutions
A panel of experts discussed how policy challenges and solutions posed by AI will impact the future of work. The panel was moderated by Gillian Tett, chair of the Financial Times (U.S.) editorial board.
Panelists spoke to the dramatic nature of press coverage related to AI advancements. “Headlines go to extremes about the impact of AI on work,” said Susan Athey, Stanford GSB Economics of Technology Professor, Golub Capital Social Impact Lab director, and HAI faculty associate director. “But we’ll likely see more gradual evolution. It’s not all-or-none in terms of taking over jobs.”
Mary Kay Henry, president of the Service Employees International Union (SEIU), noted that all service workers — ranging from engineers to caregivers — are affected by AI. “People can be automated out of a job but shouldn’t be automated out of an income,” she said.
The panel agreed that technology must be part of the solution. “Technology can deliver skills-training programs to people at home, rather than making them give up time,” Athey said.
Home care represents another promising domain for AI technologies. Henry noted that SEIU partnered with Apple to provide iPads to home-care workers, allowing them to work with physicians directly. As a result, hospitalizations decreased by 30 percent.
The panel also agreed on the high potential for AI to improve government work. “Government represents one third of U.S. GDP and too often maintains the status quo,” said Mark Duggan, Stanford economics professor and director of the Stanford Institute for Economic Policy Research (SIEPR). “We can leverage AI more, such as supporting more efficient distribution of social-program benefits and matching people with jobs during COVID.”
Former U.S. Secretary of State and current director of Stanford’s Hoover Institution Condoleezza Rice described the challenge for improving public-sector AI applications. “The problem is incentives,” she noted. “Elected officials and cabinet members have little time [due to term limits] to drive change, and career service people are used to doing things a certain way.”
AI-related government regulation and frameworks are important parts of the conversation. “In China, the state decides how to use AI, often in Orwellian ways,” Rice said. “But it can improve government’s ability to respond to citizens’ needs without bureaucracy, like India’s biometric-ID program, which improves social services while reducing corruption.”
Henry added that it will take the collaboration of government, businesses, and unions to introduce and implement AI in a way that encourages innovation at every level. She noted that U.S. companies could take a page from countries like Sweden, where truck drivers are working side by side with engineers to design AI-driven trucks.
“We need public-private partnerships,” Rice agreed. Multiple speakers noted that public and private organizations also need to work together to mitigate worker fears about AI-based surveillance.
The panel agreed that more strategic media coverage of AI could promote better understanding, including positioning AI as “augmented intelligence” to emphasize human-machine collaboration. One proposed tactic: provide more positive examples of AI and collaborative work in the realms of manufacturing, health care, and other domains.
Promoting Equality with Digital Technology in Rhode Island
In a virtual fireside chat, Stanford computer science professor and HAI co-director Fei-Fei Li discussed government’s role in deploying AI-based solutions for constituents with Rhode Island Governor Gina Raimondo.
“COVID has exposed America’s deep inequality, from criminal justice to health care,” Raimondo said. “We need to rebuild the post-pandemic economy by leveraging technology to make it more equal, such as for women and immigrants.”
Her state is working toward that goal in several ways, including by leveraging AI and machine learning. One recently formed initiative, Back to Work RI, connects Rhode Islanders who have been displaced by COVID-19 with job opportunities and training. The program will employ tools, such as Google Workspace and a first-of-its-kind tool bot, to match people to job opportunities. When it comes to promoting diversity in tech-related work, Raimondo believes the public sector has the responsibility to lead. She advocates free public-education options for community college and post-high school credentials, with a specific focus on digital and analytical skills. The Computer Science for Rhode Island (CS4RI) initiative works with Microsoft to offer computer-science learning in every public-school grade.
The governor echoed the idea of educating people on the positive impact of AI. “We need to share examples, such as how technology can keep people in their homes through home-care solutions,” she said. “We need to dial down the fear and get to work on solutions.”
AI and Robotics Frontiers
Oussama Khatib, Stanford computer science professor and director of Stanford Robotics Lab, and Affectiva CEO Rana el Kaliouby shared their insights about the future of AI and robotics.
“Human-robot collaboration is everywhere today — in hospitals, warehouses, remote environments, and other places,” said Khatib. In the health care sector, he noted, robots will soon perform several tasks normally carried out by medical assistants, while physicians will be able to execute intricate surgical procedures via a haptic interface.
Khatib also cited the success of Ocean One, a human-to-robot haptic interface that supported a 2016 mission to successfully recover a 17th-century Catalan vase from the seabed of the Mediterranean. The next-gen version of Ocean One can go even deeper, reaching depths of 1,000 meters.
In the private sector, el Kaliouby is also working to optimize the human-AI intersection, largely by capturing nonverbal-communication nuances. “Machines will increasingly interact with us, like a human to another human, through greater perception and empathy,” she said.
Affectiva has made strides in applying AI to the automotive space, including cabin-sensing technologies that alert drivers and intervene when eye closure, yawning, and other signs of drowsiness are detected. el Kaliouby notes that companies can repurpose these technologies for use in hospitals, retail, and homes. “It’s about augmenting human EQ, not replacing it.”
The Road Ahead
In the second fireside chat of the conference, Brynjolfsson spoke with Eric Schmidt about AI trends and challenges. Schmidt is the former CEO and executive chair of Google and co-founder of Schmidt Ventures.
“Fundamental shifts are occurring through AI, including better drugs and automotive materials,” he said. “Chemistry is due for a redo.” Schmidt pointed to a collaboration of researchers and machines at MIT that developed halicin, an antibiotic that can kill many species of antibiotic-resistant bacteria.
Schmidt believes that governments must also realize the value of AI. “We need to pursue more government solutions that are state-of-the-art,” he said. Adding AI experts in key roles could help governments bridge the public-private-sector gap and increase detection of Medicare and other low-level fraud.
A top concern surrounding AI is the use of personal data, which Schmidt believes can be addressed by regulation. However, he noted that GDPR, the EU regulation on data protection and privacy, had the unintended impact of making it harder for EU firms to compete with American ones. “Everyone wants ‘light’ regulation but no one agrees what that is,” he said.
Schmidt also questioned if the future is two internet platforms — one for China and one for the rest of the world — in light of recent policy challenges with China-based companies Huawei and TikTok. He noted that the more decoupled China and the U.S. become, the harder it will be to collaborate and agree on policy and other matters. A more integrative approach, he believes, would be of value.
Sachin Waikar is a freelance writer.