1 / 23

The Future of Work The Impact of Technology on Work and Consequences for LMI

The Future of Work The Impact of Technology on Work and Consequences for LMI. Steve Hine, Research Director MN DEED May 9, 2019. Consider how technology impacts employment. This topic is not something I knew a lot about! How am I ever going to fill an hour???????

lfuentes
Download Presentation

The Future of Work The Impact of Technology on Work and Consequences for LMI

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. The Future of WorkThe Impact of Technology on Work and Consequences for LMI Steve Hine, Research Director MN DEED May 9, 2019

  2. Consider how technology impacts employment • This topic is not something I knew a lot about! • How am I ever going to fill an hour??????? • Began to explore the literature, and discovered how much is out there • Including a lot of solid research and strong interest by economists • How am I ever going to fit all this into an hour???????? • My goal is to review what I have learned, and to start a conversation (and work effort?) among us to address the issue • My goal is not to tell you everything you should know • Maybe we can start with some “common ground” understanding • Three things I have learned: • There are a lot of “alarmist” claims out there, along with the more sanguine • There are ways in which future technological “disruptions” are likely to look like past episodes, but also important differences this time • Policy responses in the past have often/always been sub-optimal

  3. Technological disruption isn’t a new concern • Industrial Revolution – Remember the Luddites! • Automation disrupted lives but broader secular stagnation did not arise • “lump of labor fallacy” that there’s a fixed amount of work available and that any work done by machine necessarily reduces work available to humans. • It did bring about many changes that might be construed as negative • the transition from farm to factory, urbanization, pollution, longer hours of more intensive labor, ... • Taught economists • that machinery/capital could be either labor-saving or labor-augmenting, either a substitute or a complement • but that in the long run the impact on the productivity of labor and hence wages is positive.

  4. Concerns continued into the 1900s • “[F]or the first time since his creation man will be faced with his real, his permanent problem—how to use his freedom from pressing economic cares, how to occupy the leisure, which science and compound interest will have won for him, to live wisely and agreeably and well.” (JM Keynes, “Economic Possibilities for our Grandchildren.” 1930) • This, too, obviously did not come to pass. • “The rise in unemployment has raised some new alarms around an old scare word: automation. How much has the rapid spread of technological change contributed to the current high of 5,400,000 out of work? ... many a labor expert tends to put much of the blame on automation.” (“The Automation Jobless”, TIME, February 1961) • February ’61 also happened to be the first month of the expansion of the ‘60s which was at the time the longest ever.

  5. They were indeed scary times!

  6. Technology continues to disrupt • A well-known consequence has been the “hollowing out” of middle class jobs • Other explanations – drop in unions, declining real minimum wages, globalization – may contribute • But the automation of routine tasks often found in mid skill jobs is Suspect #1 • Topic of David Autor’s AEA Ely Lecture • Notice that this elimination of mid skill jobs has been ongoing for some time now • Recurring theme – proper policy actions have failed to meet the challenges

  7. Another noticeable recent consequence • The changing nature of working relationships • contingent arrangements versus permanent payroll-based relationships. • “Few problems in the law have given a greater variety of application and conflict than the cases arising in the borderland between what is clearly an employer-employee relationship and what is clearly on of independent entrepreneurial dealing.” NLRB v. Hearst Publications, Supreme Court 1944 in which newsboys were found to be employees of the company • So this change too is not entirely new, and its prevalence appears overstated by many • But the Gig Economy (Uber, AirBnB, TaskRabbit, Upwork, etc) driven by technology is changing relationships, and likely at increasing rates

  8. Some observations about the past • Disruptions due to technology have been substantial and frequent • Many of the fears and expectations (secular and rampant unemployment, falling incomes, increased leisure time, etc.) have not come to pass • In fact, rather than displacing workers, technology has often shifted them into other occupational areas, often at higher wages and under better work conditions • Generally accepted that technological advance, despite its disruptions, is a key requirement for an expanding economy • Many of the lingering negative consequences (e.g. income inequality) aren’t directly due to technology but rather a failure/absence of corrective policy actions • We are just not very good at predicting the future impact of technology!

  9. A Small Set of Examples of This Difficulty 1876: "The Americans have need of the telephone, but we do not.  We have plenty of messenger boys." — William Preece, British Post Office. 1876: "This 'telephone' has too many shortcomings to be seriously considered as a means of communication." — William Orton, President of Western Union. 1889: “Fooling around with alternating current (AC) is just a waste of time.  Nobody will use it, ever.” — Thomas Edison 1903: “The horse is here to stay but the automobile is only a novelty – a fad.” — President of the Michigan Savings Bank advising Henry Ford’s lawyer, Horace Rackham, not to invest in the Ford Motor Company. 1921: “The wireless music box has no imaginable commercial value.  Who would pay for a message sent to no one in particular?” 1946: "Television won't be able to hold on to any market it captures after the first six months.  People will soon get tired of staring at a plywood box every night." — Darryl Zanuck, 20th Century Fox. 1955: "Nuclear powered vacuum cleaners will probably be a reality within 10 years." — Alex Lewyt, President of the Lewyt Vacuum Cleaner Company. 1959: "Before man reaches the moon, your mail will be delivered within hours from New York to Australia by guided missiles.  We stand on the threshold of rocket mail." — Arthur Summerfield, U.S. Postmaster General. 1961: "There is practically no chance communications space satellites will be used to provide better telephone, telegraph, television or radio service inside the United States." — T.A.M. Craven, Federal Communications Commission (FCC) commissioner. 1966: "Remote shopping, while entirely feasible, will flop.” — Time Magazine. 1981: “Cellular phones will absolutely not replace local wire systems.” — Marty Cooper, inventor. 1995: "I predict the Internet will soon go spectacularly supernova and in 1996 catastrophically collapse." — Robert Metcalfe, founder of 3Com. 2005: "There's just not that many videos I want to watch." — Steve Chen, CTO and co-founder of YouTube expressing concerns about his company’s long term viability. 2006: "Everyone's always asking me when Apple will come out with a cell phone.  My answer is, 'Probably never.'" — David Pogue, The New York Times. 2007: “There’s no chance that the iPhone is going to get any significant market share.” — Steve Ballmer, Microsoft CEO.

  10. Is this time different? • The nature of the technology change is, our (in)ability to predict its impact isn’t • Our ability to formulate appropriate policy is to be seen (but I’m not optimistic!) • Automation in the past generally applied machinery to tasks that required lots of muscle (plowing a field) or were routine in nature (weaving a blanket) or both.

  11. But AI is different than steam power • The development and application of the following allows the automation of tasks that look a lot more cognitive and non-routine than in the past: • the application of improved algorithms • substantial computing capacity and speed • large sets of data • And this development is happening at a pace far faster than any previous introduction of labor-replacing technology • Moore’s Law (Intel co-founder 1970) – computing capacity doubles every two years • This has already impacted areas thought safe not too long ago • medical diagnosis, financial management, lawyering, truck driving, telemarketing, journalism, restaurant cooking and serving

  12. What about from here on out? • This observation that many more occupations are being impacted often leads to headlines like this – “Robots may replace 800 million workers by 2030.” • Although the actual McKinsey study states “our new research estimates that between almost zero and 30 percent of the hours worked globally could be automated by 2030.” • Too often these translate automation into “safe” and “at risk” occupations, with those at risk meaning eliminated • Most serious and data-based analysis by economists are more sanguine about the impact of technology in the years ahead • “Automation doesn’t generally eliminate jobs. Automation generally eliminates dull, tedious, and repetitive tasks. If you remove all the tasks, you remove the job. But that’s rare.” Hal Varian, Google Economist

  13. So the Alarmists are right, right? • Researchers point out that occupations can be thought of as a collection of tasks – think O*Net tasks – that can vary in their nature • Specifically, in their susceptibility to automation. Tasks that are susceptible to automation are those that: • Map well-defined inputs into well-defined outputs – e.g. the autocoder • Have large datasets of input-output combinations that enable learning • Have clearly defined goals and clear metrics of success/failure • Do not rely on long and complex reasoning – Andrew Ng's 1-second Doctrine goes: "If a typical person can do a mental task with less than one second of thought, we can automate it using AI …" • Don’t require detailed explanations for how a decision is made • As AI works probabilistically, tasks that allow for some margin of error • Are not required to change rapidly over time relative to the acquisition of new learning data • Do not require special dexterity or mobility (for now)

  14. Interesting analysis • Brynjolfsson, Mitchell and Rock (AER P&P 2018) use O*Net tasks to evaluate the “suitability for machine learning” (SML) of occupations based on the various tasks performed by those in a given occupation • They developed a 23-question rubric to score each of 18,156 O*Net tasks • They use CrowdFlower, a human intelligence task crowdsourcing platform to produce these scores • They then weight these SML scores by their importance to each SOC to calculate a set of SOC-level SMLs • “[W]e find that 1) most occupations in most industries have at least some asks that are suitable for machine learning (SML), 2) few if any occupations have all tasks that are SML and 3) unleashing ML potential will require significant redesign of the task content of jobs, as SML and non-SML tasks within occupations are unbundled and rebundled.

  15. In addition • So rather than eliminating all jobs within any given occupation, jobs will be redesigned. • “The focus of researchers, as well as managers and entrepreneurs, should be not (just) on automation, but on job redesign” • Brynjolfsson, Mitchell and Rock also find that • The variance in occupation-level SML scores is much lower than that for the tasks, suggesting that AI will tend to have a greater impact on how tasks are “bundled” into occupations than on the elimination and creation of occupations themselves. • Correlations between SMLs and measures of wages suggest that automation through AI will be (has the potential to be) more evenly distributed across the income spectrum than was much of the earlier automation of routine cognitive tasks often found in middle-income occupations

  16. Consequences for our labor markets • AI will disrupt labor markets just as steam power, electrification, internal combustion and the microprocessor have in the past, and it will also bring the potential of productivity and living standard improvements • This disruption appears to be happening with greater speed than previous technological advances, but perhaps slower than some predict and in ways we can’t predict • Likely experience rapid changes in the tasks that comprise and the skills needed to perform well within an occupation even as those occupations continue to exist (in altered form) • This will likely occur across a broader spectrum of occupations and income ranges than previously experienced • Impact will be felt more intensely in areas with less occupational diversity and concentrations of high SML occupations

  17. “Typical” policy recommendations from the Aspen Institute • Encourage Employers to Lead a Human-Centric Approach to Automation • CHALLENGE: Automation changes workforce skill needs, yet employer investment in workforce development has declined. • Solution: Promote employer engagement and investment through a worker training tax credit, expansion of apprenticeships, and new sector and regional workforce partnerships. • CHALLENGE: Employers are making decisions about adopting automation, but may not take into account potential impacts on workers and communities. • Solution: Encourage employers to adopt a multi-stakeholder approach to automation decisions by promoting new forms of worker voice and ownership and developing proactive strategies to identify and address impacts in advance. • Enable Workers to Access Skills Training, Good Jobs, and New Economic Opportunities • CHALLENGE: The labor market is constantly evolving, with automation contributing to changing jobs and skill needs, but supports for worker training and adult education are limited. • Solution: Improve access to effective and affordable skills training and develop a culture and system of lifelong learning. • CHALLENGE: Many workers struggle to make ends meet, and while automation has the potential to improve job quality, it also may lead to more low-wage jobs and greater economic insecurity. • Solution: Increase wage subsidies and the minimum wage, while creating more economic opportunities by improving labor market flexibility and promoting entrepreneurship.

  18. Aspen Institute continued • Help People and Communities Recover from Displacements • CHALLENGE: Workers displaced by automation face significant economic challenges. • Solution: Strengthen supports for unemployed workers through retraining, reemployment services, and Unemployment Insurance to help displaced workers transition to new jobs and careers. • CHALLENGE: Communities that are severely impacted by automation require targeted and comprehensive strategies to recover and transition. • Solution: Support local economic development and improve regional competitiveness through sector-based development strategies and investment in digital infrastructure. • Understand the Impact of Automation on the Workforce • CHALLENGE: Policymakers, communities, workers, businesses, educators, and other stakeholders struggle to understand how automation is changing the economy because federal, state, and local data on the impact of technology on work is inadequate. • Solution: Provide key stakeholders with better information on the impact of automation by collecting data on technological advancements, adoption rates, and workforce impacts.

  19. Implications for policy • These recommendations are representative of most that are out there as most include: • Policies to improve worker flexibility and mobility • Policies to increase employer/worker engagement in training and decision-making • Policies to improve training opportunities • Policies to ensure the equitable distribution of the gains to technology • Policies to assist workers and communities impacted by technology • And most importantly ........ • Policies to ensure necessary information to effectively guide all these other policies • Note that much (not all!) of the structure is already in place to implement many of these policy recommendations

  20. More on the “necessary information” policy » Enrich state UI wage records: States should include additional data elements in UI wage records, such as occupational title (using standardized occupational codes), hours worked, and work sites, to provide a more accurate picture of career pathways. DOL should work with states to establish strategies and processes to promote the enhancement of wage records, and provide one-time resources to upgrade states’ data collection systems and train employers on them. Louisiana, Oregon, Washington, and Alaska currently collect additional data elements, such as hours and occupational title. » Increase funding for state labor market information systems: The federal government provides funds to state agencies to produce, disseminate, and analyze state and local labor force statistics, including the identification of “in-demand occupations and industries,” but these funds have been cut by 45 percent since 2002. In accordance with the WIAC recommendations, federal policymakers should double the amount of funding for state agencies. In addition, state policymakers should invest more heavily in their own labor market information systems. • Among its LMI proposals, Aspen highlights two WIAC recommendations: • They go beyond simply enhancing wage records and doubling the WIGs • Not that that’s not a good start!

  21. Some additional proposals for LMI • Improved measurement of technology’s impact on work, especially on occupational data (including characteristics; KSAs, tasks • If the impact manifests as rapid changes in tasks and skills, can we measure? • Suggestions from NAS include a Technology Progress Index, technology Diffusion Index, and Specific Technology Advancement Indices • Fund studies of AI impact on work (e.g. the AI Jobs Act of 2019) • Adjustments to occupational projections that incorporate projected technology, task rebundling, and range of potential outcomes into BLS methods • I’ll let BLS read and respond, but there was an amendment introduced “to require the Bureau of Labor Statistics to submit an estimate of the resources needed to model for various changes in the workforce composition because of technological displacement.” • Provide for localized data through expanded administrative data sets and partnership with private sector (e.g. Google, LinkedIn)

  22. Last Thoughts • Should we be thinking about the future of LMI more in terms of the driving forces behind AI (the application of improved algorithms; substantial computing, capacity and speed; and large sets of data) • Does not including each of these in our decisions risk us becoming irrelevant? • That WIAC proposal to double the WIGs – is this not the context in which such a proposal can be convincingly justified? • I worry that we’ll be asked what we’ll spend it on and not have a coherent response. This may be it. • As I said at the start, I am no expert on AI (other than being good at faking my own intelligence), but maybe this can start a new and productive conversation. • Besides, I’m getting a little tired of workforce shortages as a topic!

More Related