The one piece of data that could actually shed light on your job and AI

The prevailing narrative of an AI-fueled jobs apocalypse, echoed by figures like Anthropic CEO Dario Amodei, is based on incomplete data. Current analysis relies heavily on measuring 'AI exposure'—the percentage of tasks within a job that AI could perform. Researchers at OpenAI and Anthropic have used the US government's O*NET task catalog to calculate these scores, finding, for instance, that a real estate agent's job is 28% exposed. However, University of Chicago economist Alex Imas argues that 'exposure alone is a completely meaningless tool for predicting displacement.' Knowing which tasks AI *can* do tells us little about whether it *will* replace a human, as cost, quality, and the need for human oversight are critical factors.

Imas highlights a more pressing and complex economic question: how does AI-boosted productivity affect hiring? For example, if a coder uses AI tools to triple their output, will their company hire more engineers to scale new products or lay off workers because less labor is needed? The answer depends on how much consumer demand increases in response to potentially lower prices or better products—a dynamic that varies wildly by industry. Imas issues a 'call to arms' for economists to start collecting this specific data on productivity, demand elasticity, and hiring outcomes. Without it, policymakers are operating in the dark, unable to craft coherent plans for the workforce's future, leaving workers in a state of unnecessary panic and lawmakers without a roadmap.