Workers Fleeing AI-Exposed Jobs Face Steep Pay Cuts, Study Finds

The Hidden Cost of Automation Insurance

Workers seeking to reduce their exposure to artificial intelligence by changing occupations are discovering an uncomfortable reality: the safer the destination, the steeper the pay cut. A new analysis from the Bipartisan Policy Center tracking 595,000 job transitions between 2019 and 2025 reveals that moving from highly AI-exposed roles to positions with lower automation risk frequently comes with significant financial penalties.

The research, authored by Daniel Plaisance, examined workers leaving occupations with high AI exposure—defined as roles where more than 50% of tasks could potentially be automated. The findings challenge simplistic narratives about workforce adaptation to technological change.

A Stark Tradeoff Between Safety and Income

The data exposes a roughly linear relationship between AI exposure reduction and wage outcomes. Destinations that reduce AI exposure by approximately 20 percentage points—such as management, financial, and professional roles—typically pay nearly 50% more than workers' origin occupations. However, these higher-paying pathways absorb relatively few workers.

The largest volume of transitions flows into office and administrative support roles, which account for nearly twice the volume of any other destination cluster. Yet these positions deliver minimal protection: only an 11.4 percentage point reduction in AI exposure on average, accompanied by a 7% pay decline.

At the opposite end, manual and service occupations like food preparation, personal care, and cleaning offer substantial reductions in AI exposure but frequently impose wage cuts of 25% or more. For workers already earning higher incomes in knowledge work, the same lateral move that might work for a retail employee represents a significant economic setback.

Why it matters

This research reframes the workforce policy challenge around AI. The question is no longer simply whether displaced workers can find new jobs, but whether those jobs allow them to maintain their standard of living. With upward mobility from AI-exposed occupations falling from 57% of transitions in 2023 to just 41% in 2025, the economic cost of automation is becoming more concrete—and it's being borne disproportionately by individual workers rather than distributed across society.

The Narrowing Window for Upward Mobility

Perhaps most concerning is the trend line. Workers who left highly AI-exposed occupations in 2023 achieved income gains 57% of the time, compared to 24% who experienced losses—a 33-point advantage. By 2025, that margin had collapsed to just 11 points, with upward mobility dropping to 41% while downward mobility held steady.

The timing coincides with the widespread deployment of large language models across professional workflows, though the analysis notes that concurrent factors including tech sector layoffs and federal workforce reductions make it impossible to isolate AI's specific impact.

Importantly, the decline in favorable outcomes stems not from more workers taking pay cuts, but from more transitions becoming lateral—neither gaining nor losing significant income.

Policy Implications Beyond Reskilling

The findings complicate two dominant policy responses to AI-driven displacement. The first approach—reskilling workers into higher-paying technical roles—faces the reality that these destinations, while income-preserving, are difficult to access and limited in capacity. The second assumption—that displaced workers can simply shift into manual and service work less susceptible to automation—ignores the steep and often permanent wage penalties these moves entail.

The research suggests workforce policy must extend beyond training programs to address the quality and accessibility of destination occupations, the barriers preventing workers from reaching higher-wage pathways, and the supports needed to prevent occupational transitions from becoming economic decline.

These findings were first reported by the Bipartisan Policy Center as part of their ongoing "Trapped Workers" series examining labor market mobility in the age of AI.