Why we must stop talking about artificial general intelligence — and instead build ‘pro-worker’ AI

Artificial intelligence is transforming the world rapidly, prompting concerns beyond typical technological upheaval. The direction of AI development, dominated by a singular pursuit of artificial general intelligence (AGI), raises critical questions. AGI refers to systems capable of matching or surpassing human capabilities across most economically valuable tasks.

Should this scenario unfold, automation would not be confined to specific industries or mundane jobs; rather, it could encompass a broad spectrum of work, including highly skilled professions. The potential displacement of human labor would be unprecedented.

To highlight this pressing policy issue, I joined thousands of economists and AI researchers in endorsing a statement urging policymakers and technology leaders to steer AI towards complementing human labor instead of replacing it. Research on the US labor market indicates that since 1980, automation has significantly contributed to widening income inequality by diminishing demand for workers engaged in routine tasks (Acemoglu and Restrepo, Econometrica 90, 1973–2016; 2022).

A pertinent example is the proliferation of industrial robots in manufacturing. Despite heightened productivity, many communities most impacted by automation have witnessed declines in employment and wages (Acemoglu and Restrepo, Journal of Political Economy 128, 2188–2244; 2020). The automation trend observed with industrial robots is expected to accelerate as AI adoption proliferates.

The stakes are not merely equitable prosperity; they also involve the stability of our democratic system. Societies risk destabilization if significant segments of the population are excluded from economic opportunities. However, this grim outcome is not inevitable. While technology does not inherently possess a predetermined social purpose, AI's trajectory is malleable.

Its current emphasis lies on automation - the replacement of human labor wherever feasible. Yet, there exists an alternative path. AI capabilities can be harnessed to augment workers, elevating their skills and productivity rather than substituting them. This pro-worker approach could yield more equitably distributed gains, potentially fostering broader productivity growth.

Consider education as an example. Instead of supplanting teachers, AI tools could assist in identifying students' areas of difficulty, facilitating personalized support on a larger scale. Policymakers can foster such worker-enhancing innovation through regulation and strategic incentives.