Abstract

Workforce demand in project-based industries fluctuates with project awards and client schedules, and organizations increasingly adopt artificial intelligence (AI) for demand forecasting, scheduling, and skills analysis. Research on human-centered and ethical scheduling remains scattered across individual methods and planning stages. The study found no framework that organizes these commitments across the strategic, tactical, and operational horizons through which project-based organizations plan their workforces. This paper addresses this gap through a structured scoping review. A Boolean search of the Semantic Scholar Academic Graph, combined with backward and forward citation tracing of ten seed studies, yielded 804 unique records, of which 53 met the inclusion criteria and venue-indexing filter. Conceptual synthesis followed three steps. First, the three planning horizons were identified from the workforce planning literature. Second, four recurring design principles were extracted from the human-centered AI (HCAI) and algorithmic management literatures. A principle was retained only when it was anchored in sociotechnical design theory and had a counterpart in an established AI-governance requirement. Third, the two dimensions were crossed into a 3×4 design space. The four principles are augmentation over automation, meaningful human control, fairness and transparency, and capability development. Each of the twelve resulting cells states one design commitment and serves as a pre-deployment checklist item. Seven propositions were then derived from the cells, with each proposition specifying a unit of analysis, a design condition, a comparison condition, an outcome variable, and a mechanism. AI-enabled workforce planning is theorized as a sociotechnical dynamic capability whose microfoundations are represented by the four principles. The framework and propositions are conceptual and remain empirically untested. Proposition 7, concerning productivity compression between novice and experienced workers, extends evidence from a customer-service setting and is advanced as an exploratory boundary condition. Validation is proposed through expert Delphi panels and multi-project field studies.