The need for workers’ perspectives on workplace AI
The need for workers’ perspectives on workplace AI
In the past decade, global investment in artificial intelligence development has soared. Private investment in AI went from under $5 billion globally in 2013 to over $90 billion in 2021, more than doubling between 2020 and 2021 alone.Daniel Zhang et al., “The AI Index 2022 Annual Report” (AI Index Steering Committee, Stanford Institute for Human-Centered AI, Stanford University, March 2022), https://aiindex.stanford.edu/wp-content/uploads/2022/03/2022-AI-Index-Report_Master.pdf. The implementation of AI products has similarly grown: In 2021, 56% of respondents to a McKinsey survey said their organizations used AI in at least one business functionMichael Chui et al., “Global AI Survey 2021,” Survey (McKinsey & Company, December 8, 2021), https://ceros.mckinsey.com/global-ai-survey-2020-a-desktop-3-1/p/1 compared to 20% of respondents in 2017 who reported using AI at scale or in a core part of their business.Jacques Bughin et al., “Artificial Intelligence: The Next Digital Frontier?,” Discussion Paper (McKinsey Global Institute, June 2017), https://www.mckinsey.com/~/media/mckinsey/industries/advanced%20electronics/our%20insights/how%20artificial%20intelligence%20can%20deliver%20real%20value%20to%20companies/mgi-artificial-intelligence-discussion-paper.ashx The positive and negative effects of this are already being felt by both formal workers (millions of whom are interacting with AI products or will soon see them incorporated into their jobs) and informal workers (who are encountering transformed market conditions due to the use of AI by businesses). For both groups of workers, the positive and negative impacts of these technologies are unevenly distributed, often following other existing axes of inequality, such as geography, race, and gender. Yet workers’ needs, well-being, and expertise are under-considered in AI research, development, and implementation.
In an earlier publication, “Redesigning AI for Shared for Prosperity: An Agenda,”Partnership on AI, “Redesigning AI for Shared Prosperity: An Agenda” (Partnership on AI, May 2021), https://partnershiponai.org/paper/redesigning-ai-agenda/ PAI highlighted the need to better understand AI’s impacts on job quality, including by engaging the workers who experience these impacts firsthand. Workers who directly interact with AI understand these systems’ benefits and harms in depth. In the best of circumstances, they experience the ways these technologies can make their work more efficient, error-free, and pleasurable or less grueling, tiring, or dangerous. Too frequently, workers also experience the downsides. These systems can restrict workers’ autonomy, invade their privacy, undercut their judgment and empathy, and push them to the point of exhaustion or injury. Companies that allocate managerial tasks to AI systems can subject workers to binding decisions that are capricious or cruel.
At a societal level, the increasing adoption of AI systems is poised to accelerate existing problems arising from economic inequality.David Autor, David A. Mindell, and Elisabeth B. Reynolds, The Work of the Future: Building Better Jobs in an Age of Intelligent Machines (The MIT Press, 2022), https://doi.org/10.7551/mitpress/14109.001.0001 AI research and product development is taking place in a highly concentrated group of countries and companies. Private AI investment in the United States in 2021 totaled $52.9 billion, over three times the investment by the next highest country, China at $17.2 billion — which in turn exceeded investment by the next nine countries combined.Daniel Zhang et al., “The AI Index 2022 Annual Report” (AI Index Steering Committee, Stanford Institute for Human-Centered AI, Stanford University, March 2022), https://aiindex.stanford.edu/wp-content/uploads/2022/03/2022-AI-Index-Report_Master.pdf The impacts of workplace AI use, however, will be felt around the world. As some companies attempt to automate work they had previously outsourced, others will adopt AI systems created in and for entirely different geographies.Lant Pritchett, “The Future of Jobs Is Facing One, Maybe Two, of the Biggest Price Distortions Ever,” Middle East Development Journal 12, no. 1 (January 2, 2020): 131–56, https://doi.org/10.1080/17938120.2020.1714347 Both between and within countries, AI’s current trajectory threatens to widen the gaps between the haves and have-nots.
Moreover, workers are uniquely positioned to understand how to avoid these harms and contribute ideas to improve their employers’ bottom lines. Using AI to increase job quality (or at least not decrease it) would enable employers to reap the benefits of a more engaged and satisfied workforce. Higher job quality and employee satisfaction increases productivity of existing workers, reduces turnover (retaining experience and expertise), and fosters the ability to recruit higher-caliber talent in competitive labor markets.James K. Harter, Frank L. Schmidt, and Theodore L. Hayes, “Business-Unit-Level Relationship between Employee Satisfaction, Employee Engagement, and Business Outcomes: A Meta-Analysis,” Journal of Applied Psychology 87, no. 2 (2002): 268–79, https://doi.org/10.1037/0021-9010.87.2.268 Decades of research on innovation in domains as diverse as manufacturing,Kaoru Ishikawa, What Is Total Quality Control? The Japanese Way, trans. David John Lu (Englewood Cliffs, N.J.: Prentice-Hall, 1985)Gary P. Pisano, The Development Factory: Unlocking the Potential of Process Innovation (Harvard Business Press, 1997) hospitality,Terje Slåtten and Mehmet Mehmetoglu, “Antecedents and Effects of Engaged Frontline Employees: A Study from the Hospitality Industry,” in New Perspectives in Employee Engagement in Human Resources (Emerald Group Publishing, 2015)Kayhan Tajeddini, Emma Martin, and Levent Altinay, “The Importance of Human-Related Factors on Service Innovation and Performance,” International Journal of Hospitality Management 85 (February 1, 2020): 102431, https://doi.org/10.1016/j.ijhm.2019.102431 and government service provisionSergio Fernandez and David W. Pitts, “Understanding Employee Motivation to Innovate: Evidence from Front Line Employees in United States Federal Agencies,” Australian Journal of Public Administration 70, no. 2 (2011): 202–22, https://doi.org/10.1111/j.1467-8500.2011.00726.x has underscored the unique insights and innovative potential of frontline workers and other individual contributors. Workers are afforded intimate knowledge of crucial aspects of their work that managers and leaders only see from a distance. They are experts in things like the nuances of how to create the conditions for customer satisfaction or the levels of care that need to be taken in moving objects of different fragility through a warehouse. This deep knowledge of the ins and outs of completing core tasks makes workers an underutilized source of expertise on issues and problems where AI could be a powerful tool or assistant.
Finally, pursuing collaborative workplace AI that draws on the unique strengths of humans and technology enables businesses to expand the production frontier. Many current integrations of AI into human workflows are designed around the limited capabilities of the AI systems. This, in turn, circumscribes the range of talents and skills of the people who work with them. Starting from the opposite premise — that AI should be integrated into workplaces in a way that enables human skills and talents to flourish — is undeniably harder. The reward for the achievement, however, is far greater for both workers and their employers.
Strict rationalist economic theory would predict that workers will receive sufficient wages to compensate for technologically driven harms.Edward P. Lazear, “Compensation and Incentives in the Workplace,” Journal of Economic Perspectives 32, no. 3 (August 2018): 195–214, https://doi.org/10.1257/jep.32.3.195 However, employers and workers alike lack the perfect information required for this effect. Additionally, this theory presumes robust competition for labor, and workers who possess a genuine ability to choose between different employment options.
In many labor markets, employment options are relatively concentrated, enabling companies to treat workers worse than they would in more competitive environments.Joan Robinson, The Economics of Imperfect Competition (Springer, 1969)José Azar, Ioana Marinescu, and Marshall I. Steinbaum, “Labor Market Concentration,” Working Paper, Working Paper Series (National Bureau of Economic Research, December 2017), https://doi.org/10.3386/w24147Alan Manning, Monopsony in Motion: Imperfect Competition in Labor Markets, Monopsony in Motion (Princeton University Press, 2013), https://doi.org/10.1515/9781400850679 Steps to increase information and awareness can reduce the likelihood that workers unwittingly accept poor working conditions without a sufficient compensating wage. Regulation and increased unionization can reduce the negative effects of concentrated labor markets. However, the insufficiency (as well as improbability) of these solutions point to a need for direct attention to AI’s effects on job quality.
The contributions of this report
The contributions of this report
Past research and discussions on AI’s impacts on workers have frequently taken one of three forms, with the first two especially common in popular and business discussions:
- Predictions about AI’s impacts on job availability (i.e., how many jobs AI will eliminate and which ones).
- Aspirational discussions of how AI will improve work for humans by automating “dull, dirty, and dangerous” work.
- Targeted research by academics and civil society groups on the negative impacts of AI focused on specific technologies or groups of workers.
In this last category, groundbreaking research has illuminated the harms of specific AI technologies and use cases,Caitlin Lustig et al., “Algorithmic Authority: The Ethics, Politics, and Economics of Algorithms That Interpret, Decide, and Manage,” in Proceedings of the 2016 CHI Conference Extended Abstracts on Human Factors in Computing Systems, CHI EA ’16 (New York, NY, USA: Association for Computing Machinery, 2016), 1057–62, https://doi.org/10.1145/2851581.2886426 including monitoring and surveillance,Aiha Nguyen, “The Constant Boss: Work Under Digital Surveillance” (Data and Society, May 2021), https://datasociety.net/library/the-constant-boss/Matt Scherer, “Warning: Bossware May Be Hazardous to Your Health” (Center for Democracy & Technology, July 2021), https://cdt.org/wp-content/uploads/2021/07/2021-07-29-Warning-Bossware-May-Be-Hazardous-To-Your-Health-Final.pdf algorithmic decision-making,Mary L. Gray and Siddharth Suri, Ghost Work: How to Stop Silicon Valley from Building a New Global Underclass (Houghton Mifflin Harcourt, 2019)Alexandra Mateescu and Aiha Nguyen, “Algorithmic Management in the Workplace,” Explainer (Data and Society, February 2019), https://datasociety.net/wp-content/uploads/2019/02/DS_Algorithmic_Management_Explainer.pdf shift-scheduling,Daniel Schneider and Kristen Harknett, “Schedule Instability and Unpredictability and Worker and Family Health and Wellbeing,” Working Paper (Washington Center for Equitable Growth, September 2016), http://cdn.equitablegrowth.org/wp-content/uploads/2016/09/12135618/091216-WP-Schedule-instability-and-unpredictability.pdf and platform work software and applications.V.B. Dubal. “Wage Slave or Entrepreneur?: Contesting the Dualism of Legal Worker Identities.” California Law Review 105, no. 1 (2017): 65–123, https://www.jstor.org/stable/24915689Ramiro Albrieu, ed., Cracking the Future of Work: Automation and Labor Platforms in the Global South, 2021, https://fowigs.net/wp-content/uploads/2021/10/Cracking-the-future-of-work.-Automation-and-labor-platforms-in-the-Global-South-FOWIGS.pdf Researchers have also explored particular types of impacts on workers, including worker health and safety,Phoebe V. Moore, “OSH and the Future of Work: Benefits and Risks of Artificial Intelligence Tools in Workplaces,” Discussion Paper (European Agency for Safety and Health at Work, 2019), https://osha.europa.eu/en/publications/osh-and-future-work-benefits-and-risks-artificial-intelligence-tools-workplaces data collection and privacy,Frank Pasquale, The Black Box Society: The Secret Algorithms That Control Money and Information (Harvard University Press, 2015)Ifeoma Ajunwa, “The ‘Black Box’ at Work,” Big Data & Society 7, no. 2 (July 1, 2020): 2053951720966181, https://doi.org/10.1177/2053951720938093Isabel Ebert, Isabelle Wildhaber, and Jeremias Adams-Prassl, “Big Data in the Workplace: Privacy Due Diligence as a Human Rights-Based Approach to Employee Privacy Protection,” Big Data & Society 8, no. 1 (January 1, 2021): 20539517211013052, https://doi.org/10.1177/20539517211013051 and reproductions of carceral power.Andrea Dehlendorf and Ryan Gerety, “The Punitive Potential of AI,” in Redesigning AI, Boston Review (MIT Press, 2021), https://bostonreview.net/forum_response/the-punitive-potential-of-ai/ Previously, PAI itself conducted a landscape review of AI’s demonstrated and potential impacts on worker well-being.Partnership on AI, “Framework for Promoting Workforce Well-Being in the AI-Integrated Workplace” (Partnership on AI, August 2020), https://partnershiponai.org/paper/workforce-wellbeing/
This report builds on this foundational work by bringing in the perspectives and experiences of frontline workers at the frontier of workplace AI implementation around the world. It shares their stories of how their jobs have been transformed by AI (for better and for worse) and highlights their oft-neglected expertise on challenges and opportunities in their work where they welcome AI assistance. It also synthesizes this primary research with the existing literature to offer implications and opportunities for key stakeholders on how they can take action to ensure the category of technological products commonly referred to as AI improves — not worsens — the experience of workers. Finally, it offers areas in need of further exploration in future research or implementation case studies.
Through their comments and stories, workers surfaced five key themes about their experiences of AI in the workplace. These five themes point the way toward a better future for workplace AI, one that maintains or increases companies’ profitability and revenue while also maintaining or increasing job quality. Getting there will require many decision-makers and stakeholders to do things differently than they have in the past. In some instances, the needed changes are substantial and complex. At the end of this report, we offer initial recommendations for all of the major stakeholders in this space: AI-using companies, AI-creating companies, workers and the organizations such as unions that represent them, policymakers, and investors.