Journal of Property Tax Assessment & Administration


What is artificial intelligence? Although many definitions are available, in the context of property valuation, IPTI defines artificial intelligence (AI) as machine learning designed to predict an outcome or provide an estimate, e.g., most probable sale price. AI, at least in our use of the term, is based on pattern and image recognition. It has the ability to process large volumes of data and requires intensive computer power of the type available in today’s higher-end PCs and cloud services. For purposes of this white paper, AI does not include standard statistical algorithms — most prominently multiple regression analysis (MRA) — in which the user specifies and calibrates a prediction model. Although users specify the dependent and independent variables, AI models produce no tangible equation. This white paper is therefore distinct from the International Association of Assessing Officers (IAAO) standards on Mass Appraisal of Real Property and Automated Valuation Models, which focus on equation-based applications of the three approaches to value in mass appraisal. The vision of this white paper is to provide a framework or first step toward the production of a standard on the use of AI in property assessment administration. While existing mass appraisal tools can be highly effective and produce excellent performance results, AI offers another viable tool that, if used properly, can efficiently produce equally or, arguably, more accurate valuations for many jurisdictions. Therefore, the authors believe the time has come for serious consideration of AI by the assessment community, and we hope to see the guidance offered in this paper considered and debated as part of a process to have AI adopted on a more formal basis, either as a stand-alone standard or as an addition to an existing standard.


Artificial intelligence; Data processing in assessment