AI Qualitative Research: Lessons from RAND

According to Three Essays on Artificial Intelligence and Work in the U.S., AI exposure was associated with higher employment growth across higher-exposure occupations from 2005 to 2023, but the relationship differed by type of work. RAND’s qualitative interviews about medical coding found that automation can remove simpler tasks, improve some aspects of coder well-being, and weaken the training pipeline that develops future expertise.

• RAND’s analysis of U.S. occupations from 2005 to 2023 found that higher AI exposure was associated with higher employment growth, without substantial changes in mean wage.
• RAND found that AI exposure was associated with higher employment and wage growth in occupations focused on collecting, analyzing, and interpreting information.
• RAND found that AI exposure was associated with lower employment and wage growth in occupations focused on influencing and assisting people.
• RAND’s interviews with hospital leaders, medical coders, and AI vendors found that AI automated simpler coding tasks and produced both positive and negative effects on coder well-being.
• RAND concluded that automating entry-level coding tasks can make it harder for training systems to produce experienced coders, a problem the dissertation described as threatening an “already weak training pipeline.”

RAND’s dissertation studied AI and work in the United States through both quantitative and qualitative methods. The quantitative analysis examined the relationship between occupational AI exposure and labor-market outcomes from 2005 to 2023, while the qualitative analysis used interviews with hospital leaders, medical coders, and AI vendors to examine medical coding.

RAND’s medical-coding interviews show why AI qualitative research adds context that labor-market statistics alone cannot provide. The interviews examined how automation changes everyday tasks, how workers experience those changes, and how organizations develop expertise when simpler tasks are increasingly handled by AI.

The RAND dissertation states that “AI automates simpler tasks once performed by coders, ” but the dissertation also reports that automation can increase demand for expertise while complicating the systems that produce that expertise. The finding matters for UX researchers, policy analysts, workforce planners, and healthcare leaders evaluating AI adoption.

Researchers studying AI and work should analyze task changes and learning pathways alongside employment and wage outcomes. RAND’s findings show that the same technology can support employment growth in some occupational groups while creating risks for training and advancement in a specific profession.

Qualitative research teams should ask four practical questions when studying workplace AI: Which tasks are automated? Which tasks become more valuable? How do workers learn after automation? How does AI affect well-being and perceived job quality? RAND’s medical-coding interviews demonstrate that these questions can reveal effects that aggregate labor-market measures may not show.

RAND also identified gaps in existing AI-and-work research, including inadequate measures of AI activity and availability, limited task and skill data, and problems with the usability of individual employment and wage data. Qualitative researchers can help address those gaps by documenting how workers and organizations describe AI use in practice.

AI qualitative research should remain research-focused rather than diagnostic when it concerns health-sector workers or medical coding. RAND’s findings support careful inquiry into work and training systems, not medical or employment advice for individuals.

RAND’s 2005 to 2023 analysis shows that AI exposure has different relationships with employment and wage growth across types of work. RAND’s medical-coding interviews add that automation can improve some work experiences while weakening the entry-level pathways that build expertise. AI qualitative research gives teams a practical way to examine those trade-offs through worker and stakeholder evidence. Try Evidano for free to analyze interview transcripts, open-ended responses, and research documents in one workspace.