Executive Summary
This study is motivated by the premise that the AI workforce shortage recently observed
in industrial contexts is not merely a quantitative shortage of development personnel, but
rather a structural issue arising from changes in role configurations during the diffusion of
AI across industries. As AI technologies enter a stage of maturity, workforce demand is
shifting from a focus on model development toward stages such as validation, operations,
industrial application, and field-level utilization. However, existing workforce classification
systems and policy frameworks have not adequately captured these evolving patterns.
The current AI workforce classification system is designed primarily around developers and
job functions, which constrains its ability to systematically capture the diversity of AI
utilization across industries and the transformation of role structures. As a result, the AI
workforce shortage is often reduced to a quantitative notion of a ‘developer shortage’, while
critical roles—such as validation and trust, operations (e.g., MLOps), domain-specific
application, and AI utilization in non-development functions—remain insufficiently reflected in
policy, despite constituting key bottlenecks in practice. These structural limitations not only
undermine the accuracy of workforce supply–demand projections but also exacerbate the
misalignment between talent development policies and actual industry needs.
In response, this study proposes a shift in AI workforce classification from a job- and
occupation-based approach to a framework centered on AI value chain stages and roles. By
conceptualizing the workforce across the entire AI lifecycle—including planning, data, model
development, validation, evaluation, operations, diffusion, and industrial application—this study
presents an analytical framework that diagnoses AI workforce shortages not as a matter of
aggregate quantity, but as structural imbalances across stages and roles. Furthermore, by
linking this classification system with workforce types and skill-level frameworks, the study
highlights its potential to serve as a policy infrastructure for workforce demand–supply
forecasting and the design of education and training systems.
From a policy perspective, the findings suggest the need to shift the focus of AI
workforce policy from developer-centric training toward building workforce structures that
enable industry-wide AI transformation. As AI adoption becomes more widespread, the
competencies required become increasingly differentiated across industries; therefore, it is
necessary to systematically cultivate both common foundational AI capabilities and
industry-specific specialized competencies. The transition to a value chain–based workforce
classification system represents a critical starting point for this policy shift and is expected
to serve as a key foundation shaping the future competitiveness of the Korean AI industry
and its AI-driven transformation(AX).