GenAI Cuts Job Postings 8% for Automatable Roles in Texas

Firms pull back hiring for AI-automatable positions

Generative AI is already reshaping which jobs Texas companies post online, according to new research from the Federal Reserve Bank of Dallas. Job openings for occupations whose tasks can be automated by tools like ChatGPT fell approximately 8 percent by the first quarter of 2025 compared to pre-ChatGPT levels, based on an analysis of millions of online job postings.

The decline isn't coming from AI-native startups or industry-wide contractions. Instead, existing firms that survived the transition are posting fewer openings and deliberately shifting their hiring mix away from roles with high AI exposure.

Dallas Fed economists Samuel Dodini and Tucker Smith measured automation exposure by mapping occupational tasks from the O*NET database to actual usage patterns on Anthropic's Claude platform. This approach captures what AI systems are cost-effectively performing today, not theoretical capabilities. The resulting metric represents the share of an occupation's tasks that generative AI can automate based on observed real-world deployment.

Why it matters

This is among the first empirical evidence showing generative AI's measurable impact on labor demand at scale. The research demonstrates that automation effects are materializing faster than many forecasts anticipated, and they're concentrated among workers with limited bargaining power—particularly recent college graduates entering the labor market. The 2.6 percent reduction in total Texas job postings the researchers estimate for 2025 may sound modest in aggregate, but it represents significant displacement for specific worker cohorts who face a narrower opportunity set precisely when they need entry points most.

Incumbent firms drive the shift

To isolate whether new AI-native companies or workforce restructuring at existing firms explained the decline, the researchers examined a balanced panel of Texas firms present throughout the entire study period. They measured each firm's AI exposure based on the occupations it posted before ChatGPT's November 2022 release.

Existing firms reduced postings for AI-exposed positions by 8 to 9 percent by early 2026—statistically indistinguishable from the occupation-level aggregate decline. Firms whose pre-ChatGPT job mix was 10 percentage points more automatable subsequently posted positions with 2 percentage points fewer automatable tasks—a nearly 50 percent reduction relative to the dataset mean.

This compositional shift indicates firms are substituting AI for certain job functions rather than simply scaling back hiring uniformly.

Software, clerical, and managerial roles most exposed

The occupations facing highest automation exposure include software development, web design, and computer-intensive roles, along with managers, clerical workers, editors, and other white-collar positions. Medical records technicians, for example, have nearly 5 percent of their tasks automatable according to the Claude usage index.

The researchers compared posting counts for more- versus less-exposed occupations within the same industry to control for sector-specific demand shifts. By the end of 2023, postings had fallen 5 percent for more-exposed positions; by Q1 2025, the gap widened to 8 percent.

Recent graduates face narrower entry points

Fewer than half of job ads in the dataset require more than two years of experience, and very few demand more than five years. Because online postings typically target entry-level candidates, reduced hiring disproportionately affects new labor market entrants.

The researchers note that recent college graduate unemployment rose to unusually high levels during this period of rapid GenAI adoption. Administrative data from Texas suggest AI automation has affected employment and earnings for recent graduates and prompted current students to adjust their course selections in response.

Overall, the researchers estimate that AI automation exposure reduced total online job postings in Texas by approximately 1.8 percent in 2024 and 2.6 percent in 2025.

The findings were first reported by the Federal Reserve Bank of Dallas in their Dallas Fed Economics series.