Complementarity
AI supplies information, pattern recognition, synthesis, or decision support that increases the value of distinctly human expertise, accountability, judgment, or relationships.
Strategic workforce development
Forecasting AI’s impact on public service.
A retrieval-augmented framework for evaluating how generative AI may change tasks, skills, occupational resilience, and workforce strategy over a five-to-ten-year horizon.
The research problem
Government needs more than a list of jobs “at risk.”
AI changes work through different mechanisms. It can strengthen human judgment, accelerate specific tasks, reorganize workflows, or reduce the need for human effort in bounded activities.
CLEAR’s approach evaluates occupations against a structured inventory of AI capabilities and occupational knowledge. The goal is not a deterministic headcount forecast. It is a transparent basis for asking where governments should invest in skills, redesign work, update safeguards, and preserve essential human expertise.
Research objective
Anticipatory workforce strategy
Identify occupations and task bundles likely to change, explain why, and convert those signals into actionable questions about training, hiring, job design, governance, and service delivery.
- Compare occupations consistently.
- Retain evidence and model justifications.
- Separate task change from job elimination.
- Support scenario planning rather than certainty theater.
The CAS framework
Three dimensions of AI–work interaction.
Each occupation receives separate, reasoned assessments. The combination matters more than any single score.
Augmentation
AI improves the speed, scope, consistency, or quality of work while people remain responsible for defining goals, interpreting context, and acting on outputs.
Substitutivity
AI can perform bounded tasks with less human input, potentially changing staffing, workflow, entry pathways, supervision, or the mix of skills needed.
Illustrative occupation assessment
Program management series
CAS profile
Scores are stored with evidence, rationale, model version, and retrieved occupational context.
4.4Complementarity
4.1Augmentation
2.2Substitutivity
12345
How it works
Retrieval-augmented occupational assessment.
- Occupational knowledge base. Tasks, knowledge, skills, responsibilities, and public-sector context are assembled for each occupation.
- AI capability knowledge base. A structured inventory describes what current generative systems can do, with boundaries and evidentiary context.
- Reasoned scoring. A language model evaluates each CAS dimension on a common scale and produces a justification grounded in the retrieved material.
- Validation and comparison. Outputs are reviewed, versioned, compared across models and time, and aggregated to agencies, job families, or workforce segments.
From scores to strategy
Questions public leaders can act on.
The framework is intended to guide workforce development and responsible implementation—not to automate personnel decisions.
Where should we train?
Identify occupations with high augmentation potential and specify the technical, managerial, and judgment skills needed to capture it.
Which expertise must be protected?
Locate roles where complementarity depends on tacit knowledge, public accountability, field experience, or trusted human relationships.
Which workflows should change?
Use task-level substitutivity to redesign processes, quality controls, supervision, and entry-level developmental pathways.
Where are safeguards essential?
Connect use cases to privacy, bias, security, due process, records, procurement, transparency, and human-review requirements.
How might occupations evolve?
Build scenarios for task bundles, career ladders, staffing mix, contractor dependence, and new specialist roles.
How do models change?
Re-score occupations as capabilities, retrieval corpora, and model reasoning evolve, rather than treating one assessment as permanent.
Research and education
A living laboratory for public-sector AI.
The project links computational research to workforce strategy, public management, and student learning.
Students and collaborators can contribute to occupational data engineering, capability taxonomy development, validation, model comparison, visualization, agency case studies, and practitioner-facing guidance.
Interpretation matters. A high substitutivity score for some tasks does not imply that an occupation disappears. Tasks are bundled, missions are public, accountability cannot always be delegated, and new technology often changes demand as well as productivity.
Collaborate
Contact CLEARApply the framework to an agency, occupation, or workforce question.
CLEAR is developing research, comparative analyses, and practitioner applications around AI and public-service work.