Redesigning AI for Shared Prosperity: An Agenda
Artificial intelligence is expected to contribute trillions of dollars to the global GDP over the coming decade, but these gains may not occur equitably or be shared widely. Today, many communities around the world face persistent underemployment, driven in part by technological advances that have divided workers into cohorts of haves and have nots. If AI advancement continues on its current trajectory, it could accelerate this economic exclusion.
This is not the only trajectory AI could be on: switching the emphasis from automating human tasks to genuinely complementing human workers can help raise these workers’ productivity while making jobs safer, more stable and rewarding, and less physically exhausting. Redesigning AI for Shared Prosperity: an Agenda is a foundational document of the AI and Shared Prosperity Initiative outlining practical questions stakeholders need to collectively find answers to in order to successfully steer AI toward expanding access to good jobs—and away from eliminating them. We are sharing this living Agenda with the community to inform aligned efforts and invite all interested stakeholders to partake in the work. (Read our press release on the Agenda here.)
The Agenda, developed under the close guidance of the Initiative’s Steering Committee and based on their deliberations, calls for the creation of shared prosperity targets: verifiable criteria the AI industry must meet to support the future of workers. These targets would consist of commitments by AI companies to create (and not destroy) good jobs—well-paying, stable, honored, and empowered ones—across the globe. The commitments could be adopted by the AI industry players either voluntarily or with regulatory encouragement.
To date, no metrics have been developed to assess the impacts of AI on job availability, wages, and quality. Additionally, no targets have been set to ensure new products do not harm workers, either in aggregate or by category of potential vulnerability. Without clear metrics and commitments, efforts to steer AI in directions that benefit workers and society are susceptible to unbacked claims of human complementarity or human augmentation. Currently, such claims are frequently made by organizations that, in reality, produce job-displacing technology or employ worker-exploiting tactics (such as invasive surveillance) to produce productivity gains. We expect that organizations genuinely seeking to complement and benefit workers with their technology, would be most interested in measuring and disclosing their impact on availability of good jobs, helping differentiate themselves from industry actors seeking to sell worker exploitation-enabling technologies masked as “worker-augmenting AI”.
The success of the targets to be developed relies on their support by critical stakeholders in the AI development and implementation ecosystem: workers, private sector stakeholders, governments, and international organizations. Support within and across multiple stakeholder categories is particularly important given the diffuse nature of AI’s development and deployment: technologies are often created in separate companies and separate geographies than where they are implemented. Directing AI in service of expanding access to good jobs offers opportunities as well as complex challenges for each set of stakeholders. The Agenda outlines questions that need to be resolved in order to align the incentives, interests, and relative powers of key stakeholders in pursuit of a shared prosperity-advancing path for AI.
As an immediate next step, the Initiative is working to conduct thorough research on workers’ experiences of AI in the workplace. The research aims to identify key categories of impact on job quality to be included in the shared prosperity targets, as well as the most effective ways to empower workers throughout the AI development and deployment process. If you are an employer or worker organizing group who would potentially be interested in participating in this research, please get in touch to learn more about our research and how you can contribute.
It is our hope that this Agenda will catalyze the research and debates around automation, the future of work, and the equitable distribution of the economic gains of AI, and specifically on steering AI’s progress to reduce inequality and support sustainable economic and social development. PAI also enthusiastically invites collaboration on the design of shared prosperity targets. For more information on the AI and Shared Prosperity Initiative and how to get involved, please visit shared-prosperity-initiative.
To read the Agenda’s Executive Summary, click here. To read “Redesigning AI for Shared Prosperity: an Agenda” in full, click here.