In the late 1950s and early 1960s, policymakers warned that the automation of the time would be leading to unemployment rates at 10% or higher. In this 1968 photo, a robot is tested as a replacement for car mechanics.
Mirrorpix via Getty Images
The newly formed Commission on Artificial Intelligence (AI) and the American Workforce, announced this past week by the American Enterprise Institute (AEI) and the Urban Institute, is the latest in a series of recent efforts to get in front of AI impacts on jobs. It joins the high-profile Peterson Foundation/Windfall Trust conference on AI and jobs earlier in June, the ongoing economic analyses by teams at Open AI and specialized AI providers, and the introduction of the Great American Artificial Intelligence Act in Congress on June 4.
The current efforts follow a half-century of previous “future of work” commissions and committees that were formed to map the employment future in response to technology advances. These previous commissions and committees were well-funded and staffed. Yet, their designs for the job future were quickly overtaken by the pace and nature of job growth that they failed to anticipate.
Often their analyses were at such high levels of generality as to add no guidance to policymakers or practitioners in the job training field. When the commissions did make employment projections, these projections missed badly on the job destruction and creation that did occur, while failing to identify the job occupations and structures that emerged.
Why have these past “future of work” efforts been so unsuccessful? What workforce planning efforts, if any, are worth undertaking today in relation to AI?
In California since the 1960s a series of blue ribbon commissions and committees have been established in response to fears of technology replacing jobs.
In June 1963, the state legislature established a Commission on Manpower, Automation and Technology, responding to fears that automation and technology were eliminating jobs with unprecedented speed. The Commission members and staff traveled throughout the state in 1964 gathering testimony from labor researchers, employers and worker representatives. Much of the testimony came from union officials, describing how automation was putting their members out of jobs: the Cannery and Food Processing Workers, the State Council on Carpenters, even the Milk Wagon Drivers, Local 302 (whose membership in California had dwindled to 1400 by 1964).
The Commission warned that unemployment in California could reach 10% or higher by the end of the decade. That proved to be a massive miscalculation. Automation did eliminate jobs, but also led a greater number of jobs being created. The California economy saw unprecedented job growth in the next five years. Non-farm payroll jobs totaled 5.6 million statewide in December 1964. By December 1969, this number had jumped to over 7 million.
In the early 1980s, California experienced a period of “deindustrialization” and widespread industrial plant closings. More than 900 industrial facilities closed between 1980 and 1983, in automobile manufacturing, lumber and paper mills, food processing, and steel--the Kaiser steel plant closure in Fontana California, International Paper plant in Siskiyou County, General Motors auto plant in Fremont. Governor Jerry Brown’s Administration quickly established a “California Economic Adjustment Team”, assisted by numerous task forces. Elaborate designs were prepared for government-directed and subsidized jobs in service and clean energy industries seen as replacing manufacturing,
None of the plans came to be implemented, as once again the fears of a job apocalypse proved unfounded. The California economy continued to gain jobs in the 1980s, reaching 12.5 million payroll jobs by January 1990. Further, while heavy manufacturing declined, other forms of manufacturing not envisioned by the task forces emerged. In 1980, the state had 1.82 million manufacturing jobs; by 1990 this number was up to 1.98 million. Today, despite warnings since the 1990s of the “end of manufacturing”, the state has more than 1.2 million manufacturing jobs, including a re-engineered steel plant at the former Kaiser site in Fontana, a Roseburg Forrest Products paper plant in Siskiyou and a number of semiconductor manufacturers near the former auto plant in Fremont.
Similar “future of work” miscalculations followed with subsequent employment dislocations: the downturn in the state’s aerospace and defense industry in the 1990s, the dot.com bust in the early 2000s, the Great Recession in 2008, and the pandemic. Each time, job losses were met with commissions and task forces, issuing job predictions and calls for government to plan and spur jobs in favored sectors, such as clean energy, healthcare and social services. Each time the economy rebounded in ways not at all anticipated and without government direction.
The most recent high profile “future of work” effort, the “California Future of Work in California” commission was announced in 2019 with great fanfare. Its members occupied high positions in government, business and labor, and the state’s foundations contributed hundreds of thousands of dollars for staff and consultant support.
Yet, the Commission’s final report, issued in March 2021, was quickly forgotten. Its hackneyed title, “A New Social Contract for Work and Workers”, reflected the absence of any serious research or insights into job growth in the state. Its five areas of recommendations merely repeated well-worn themes in the workforce field (“ensure jobs for everyone who wants to work”; “eliminate working poverty”; “create a 21st century safety net”; “raise the standard of quality jobs”; “future proof California with jobs and skills”) without any details on how these themes were to be implemented. Moreover, it failed to identify any of the job dynamics that have emerged in the state since: the rebound of the state’s blue collar economy, the boom in the health care workforce, the ridesharing and self-driving cars, and of course the speed of the AI take-up by employers.
The path and speed of job creation and destruction are built on numerous factors including the path and speed of technology, shifting government policies, changing consumer preferences. None of these factors are knowable in advance. Some predictions will be better than others, but all will be based on the limited information at the time of prediction.
A few years back in California, in early 1999, a new online grocery company, Webvan launched with an initial $275 million in venture capital funding—one of the biggest capital deals of the year. By November, it had more than 400 employees. It promised to “reengineer the way people shop for groceries”. Yet by July 2001, Webvan had closed down after going through $1.2 billion in investment, and the online grocery industry was widely dismissed as impractical .
Today, though, the online grocery and restaurant home delivery market in California is booming—Uber Eats, Doordash, Instacart. The technology for delivery coordination improved. State government policies shifted on worker classification and delivery regulation. Most of all, consumer preferences for home deliveries changed during the pandemic and continued after the pandemic. None of these developments would have been recognized by a future of work commission in 2001, nor were they identified by economists of the time.
Brent Orrell is the co-director of the new AEI/Urban Institute Commission on AI and Workforce, along with Elisabeth Jacobs of the Urban Institute. Orrell, a well-known scholar in the workforce field, is aware of the shortcomings and generalities of past future of work efforts. He emphasizes the Commission’s mandate to operate in real time in ways that can assist workforce practitioners and employers: tracking the impacts of AI as they emerge and providing various implementation scenarios and options.
The Commission has a one-year timetable to deliver product on eight research tracks, including both how AI adoption affects specific occupations and wage levels, and the realistic upskilling and reskilling options. Up to now, AI’s impact on jobs has been in augmentation in existing jobs, rather than job creation or destruction, but that could change.
Georgetown University Professor Harry Holzer is one of the Commission members. Holzer recently published an essay in Democracy Journal reviewing the impacts so far of AI, and how the benefits of AI might be widely shared. Like Orrell, he emphasizes the need not to get too far ahead of the technology, to avoid excessive speculation, while ensuring that workers in sectors outside of tech are able to develop AI skills.
***
The AI Future of Jobs: Who Knows?
This past week, Jeff Bezos became the latest business leader to weigh in on the job impacts of AI. In contrast to other business leaders who have warned of AI leading to large scale job displacements, Bezos argued that AI will create more jobs than it eliminates and there will be labor shortages not labor surpluses. In his optimistic scenario, he joins recent upbeat job assessments by David Solomon of Goldman Sachs, and Sam Altman of OpenAI.
Who knows. But we can learn from past “future of work” commissions on not getting ahead of ourselves in workforce policies and programs, and instead being prepared to pivot and adapt as job structures evolve.
此内容由惯性聚合(RSS阅读器)自动聚合整理，仅供阅读参考。 原文来自 — 版权归原作者所有。