Structuring Human-AI Collaboration in Residential Property Valuation Using Levels of Automation

Abstract

The increasing adoption of artificial intelligence (AI) in residential property valuation has been driven largely by advances in automated valuation models and predictive analytics. However, existing research has focused predominantly on model performance and interpretability, offering limited guidance on how AI should be embedded within professional valuation practices that are judgement-intensive, regulated, and accountable. This study addresses this gap by reframing AI integration as a governance and workflow design problem rather than a purely technical challenge. Drawing on Levels of Automation (LOA) as an analytical lens, the study systematically maps human and AI roles across valuation workflows and identifies the limitations of conventional LOA frameworks in addressing professional authority and accountability. To overcome these limitations, the study introduces a Professional Governance Layer as a cross-cutting mechanism addressing decision authority, accountability ownership, override capability, and audit responsibility in AI-supported valuation. Using a systematic literature review and an expert-based Delphi approach, the study develops a Human-AI Hybrid Valuation Framework that structures human-AI collaboration while preserving non-transferable professional responsibility. The proposed framework contributes a process-oriented and professionally defensible approach to AI adoption in residential property valuation, offering practical implications for valuers, regulators, and system developers concerned with responsible and accountable AI integration.