Generative AI reshapes online labor markets—major displacement in programming and writing roles

Generative AI systems built on large language models are reshaping online labor markets in ways that go beyond simple job elimination. New research analyzing data from a major freelancing platform documents how these tools are creating winners and losers across the gig economy, with ripple effects that extend beyond the roles most vulnerable to automation.

The contraction effect#

When skills closely match what large language models like ChatGPT can do—writing, coding, copyediting, and similar text-based work—entire submarkets experience shrinkage. Both the number of freelancers seeking work and the number of job postings decline. However, the supply-side contraction is notably milder than the demand-side collapse. The result: freelancers face stiffer competition for fewer positions in precisely the roles where AI should be most disruptive.

Skill transitions reshape the landscape#

The research uncovers a secondary effect that complicates the displacement narrative. By lowering the barrier to entry for technical work, tools like ChatGPT encourage experienced freelancers to diversify into programming tasks they might not have previously attempted. Rather than exit the platform entirely, workers are shifting sideways into adjacent fields. This skill-transition mechanism appears particularly acute among higher-skilled workers, suggesting that experience and reputation may give some freelancers the confidence and track record to branch into new territory.

Why this matters for workers and platforms#

The findings suggest that generative AI's labor-market impact is neither uniformly destructive nor confined to direct displacement. Workers in closely aligned fields face immediate competitive pressure, yet opportunities emerge for those positioned to adapt. Platform operators and policymakers must grapple with the tension between automation-driven efficiency and the need to support workforce transitions. The heterogeneous nature of these shifts—especially the ability of skilled workers to pivot—hints at complex distributional consequences that simple re-training narratives may overlook.