Shared Prosperity Targets for the AI Industry

When AI companies develop new technologies, they should be required to perform a distributive impact assessment to ensure that inventions enhance human job opportunities rather than solely displacing human workers.

By Katya Klinova and Stephanie Bell

What is your proposal? We propose regulation requiring AI companies to measure and disclose their impact on labor demand: how many jobs have been created or eliminated, made better or worse (in terms of wages and other key quality indicators) as a result of their existence. Measuring the labor demand impact would allow regulators to incentivize the development of AI applications that genuinely complement workers, boost productivity, and support creation of good jobs, and tax/disincentivize those that do not meaningfully grow productivity, but do transfer economic power from labor to capital by cutting labor costs, impoverishing workers and devastating their communities. The presence of incentives and disincentives as well as societal pressure enabled by the transparency around AI companies’ labor demand impacts would prompt companies to adopt targets around non-destruction of good jobs, proactively anticipate the likely impacts on labor demand of their product pipeline, and adjust it to meet the targets.

What problem does your proposal address? Popular approaches to AI and the future of work fall into a series of traps. Proposals to reskill workers whose jobs are automated ignore economic research showing technology is displacing tasks faster than it replaces them. Proponents of redistribution-reliant strategies encounter two major problems: 1) today’s automation is often “so-so,” offering minimal productivity gains to redistribute, and 2) they lack feasible political paths to replace workers wages while covering for technology that destroys worker livelihoods globally. We propose “pre-distribution”: incentivize creation of genuinely worker-complementing AI that expands productivity frontiers and economic gains, while maintaining robust demand for waged work.

How does this policy proposal relate to artificial intelligence? It is not assured that AI’s gains will accrue to humanity at large as opposed to a small number of actors in the AI industry. At this early stage of AI development, we can prevent its worst economic impacts on workers—rather than forcing society to respond to mass labor market disruptions that don’t need to occur. Our proposal would ensure that AI companies do not capture windfall gains at society’s expense, and incentivize them to steer AI’s progress toward shared prosperity. Measuring and disclosing companies’ impact on labor demand would also help better differentiate genuinely human-augmenting AI from empty claims.

Visit AI and Shared Prosperity Initiative to learn more.