Remote Hiring Predicted Faster AI Adoption

Firms that hired more remote workers later advertised for generative AI skills more often. In Gregor Schubert’s U.S. job-posting study, a 10-percentage-point increase in remote hiring during 2021–2022 is estimated to raise the share of postings mentioning generative AI in 2023–2024 by 0.4 percentage points across firms and 0.7 points across occupations within firms.

The proposed explanation is organizational: adapting to distributed work changed the technical and managerial capabilities firms hired for, and those capabilities could support later AI adoption. That is a qualified answer. The estimated effect concerns demand expressed in vacancies; the evidence for the precise pathway is suggestive, and the figures do not measure how many employees used AI.

What counts as adoption here

The earlier measure is the share of a firm’s job ads offering fully remote or hybrid work. The later measure is the share mentioning generative AI tools or skills. Both come from U.S. online vacancies, so the comparison follows employers’ stated hiring needs rather than observing employees at work. A posting can signal an intended role or capability without showing that the job was filled or an AI system was deployed throughout the company.

The reverse gap matters too. Employees may use generative AI in existing jobs without creating a vacancy that mentions it. The analysis includes employers with enough later postings to measure a share, which also limits how readily its estimates can be extended to firms that rarely advertise. In this article, “AI adoption” therefore means formal demand for AI-related skills in hiring, the outcome the research actually observes.

How predicted remote hiring helps isolate an effect

A simple comparison between remote-heavy employers and other firms would be hard to interpret. A company inclined to try new technology might embrace both remote work and AI independently. Some occupations are also suitable for both: their tasks can be done away from an office and can be assisted by generative AI. Either pattern could make remote and AI vacancies rise together without one change helping cause the other.

The alphaXiv research overview describes the instrumental-variable approach: it combines the remote-work suitability of jobs in the labor markets where a firm recruited before the pandemic with the suitability of the firm’s own jobs. The idea is that an employer with remotely feasible roles faced more pressure to offer that option when competing in markets full of similarly feasible jobs. Those earlier characteristics predict remote hiring without relying solely on the employer’s later choices.

The analysis then asks whether firms facing more of that predicted pressure subsequently posted more AI-related vacancies, allowing for measured differences in their jobs and industries. A comparison across occupations within the same firm provides another check: it reduces the chance that a company-wide appetite for technology accounts for the whole association. An alternative version uses pre-pandemic commuting pressure, which offers a different reason workers might value remote jobs.

This supports a causal interpretation more strongly than correlation alone, but the design has a condition that data cannot directly prove. After the controls, the predicted pressure to hire remotely must affect later AI vacancies chiefly through remote-work adoption. If the same labor markets independently gave firms access to AI talent or infrastructure, part of the estimate could reflect that route. The alternative comparisons narrow such concerns; they do not remove every possible one.

What firms may have gained from distributed work

The proposed technology ladder runs through capabilities, rather than employees’ physical location. More remote hiring was followed by greater demand for computer occupations, data-related skills, managerial roles and work involving coordination or decisions. Firms that already had stronger technical and managerial capacity were also more likely to turn the AI potential of their job mix into AI-related hiring. Together, those patterns suggest that adapting to distributed work could lower the organizational effort of introducing another technology.

UCLA Anderson’s research brief describes employers adding information-processing specialists and managers with communication and team-building skills as they adapted to remote teams. Some of those supporting roles were in-office jobs. That distinction matters to the headline: an employer could retain skills and ways of coordinating work even if its staff later spent less time at home.

The proposed sequence is plausible without implying that every remote arrangement produced it. Offering a home-work option does not automatically build data systems or management capacity. Conversely, a firm can keep capabilities developed for distributed operations after changing its location policy. Hiring patterns show a shift consistent with reusable capacity, but they cannot assign a separate causal share to each technical system, manager or workflow.

Why the effect differs across employers

At the firm level, the estimated remote-to-AI link is largest in technology and in financial activities and business services. These sectors contain substantial information and decision work, where technical support and coordination may be especially useful for introducing AI. Sector estimates use narrower samples than the overall comparison, and estimates within firms are less precise. The pattern supports the proposed mechanism without establishing an exact ranking of industries.

The firm-level link is also larger among employers identified as having return-to-office mandates. One interpretation is that firms facing difficulty coordinating remote work had an added reason to seek automation. A mandate is only a rough indicator of that difficulty, however: employers can impose office requirements for many reasons. This comparison cannot establish that bringing staff back caused their AI-related hiring.

What the result leaves open

The absolute effect is modest even though the direction is clear in the estimates. In a hypothetical group of 1,000 vacancies, 0.4 percentage points amounts to four additional ads mentioning AI; 0.7 points amounts to seven. Those are changes in the share of advertised jobs, not shares of workers replaced, AI projects completed or productivity gained.

Earlier remote hiring thus appears to have helped some firms move faster toward formal AI hiring, with accumulated technical and managerial capacity offering a credible explanation. The evidence reaches further than a raw correlation between two kinds of vacancy. It stops short of proving that any particular remote-work system caused successful AI use, or that employees had to remain at home for the later effect to appear.