A Review of the Methods, Applications, and Challenges of Adopting Artificial Intelligence in the Property Assessment Office

In early 2020, IAAO President Amy Rasmussen created the Artificial Intelligence Task Force with the goal of developing a white paper describing the impact and uses of artificial intelligence (AI) in government valuation offices. The COVID-19 pandemic in early 2020 forced government valuation offices to adapt overnight. Many jurisdictions rapidly virtualized tasks and duties, which accelerated ongoing efforts to utilize office automation and implement intelligent software solutions. More and more, workflows incorporating digital information and multiple sources of data are processed and analyzed using software and integrated applications. The fully integrated workflows facilitate the increased usage of AI in operations, assessment, and valuation. This white paper delivers an introduction and overview of AI through case and pilot studies and review of relevant analytic methods while touching on possible organizational impacts. The paper looks at the changing role of valuers and assessment administrators and the evolution of valuation offices where AI will be used to improve operations, value estimates, and administration. It provides illustrative examples of AI use in the conduct of tax assessment, including the administrative aspects not directly involved in valuation. While there is substantial fanfare around valuation with AI, many of the benefits to be realized from the technology are in areas of administration, validation, and oversight. This is reflected in the case studies included, with more than half involving AI applications outside of the explicit valuation function. The introduction provides a definition and brief history of AI. It also helps disentangle the raft of AI methods with how they are used and provides a concrete list of which assessment activities may benefit from those general classes of algorithms. More importantly, the first section helps put into context why AI is becoming more widespread and what that means for organizations from both staffing and administration standpoints. After the introduction’s overview of what AI is, why it has captured professional imagination, and the organizational changes it portends, we provide examples of current uses by assessing organizations and their partners. The first case study is about the Property Valuation Services Corporation’s (PVSC) foray as the first organization in Canada to publish a tax assessment roll using AI-based valuations. This case study highlights the multiyear process leading the organization to that accomplishment and the lessons it learned along the way. The second use case is a pilot study by PVSC. The section discusses the success of AI, particularly machine-learning methods, for the valuation of residential properties in the Netherlands. The third use case is from BC Assessment (BCA) and describes how valuations of manufactured homes were conducted using AI methods. For successful adoption of AI in an assessment office, this case study highlights the importance of communication and feedback from appraisers and integration of AI-modeled values with the computer-assisted mass appraisal (CAMA) system. The fourth case study comes from the City of New York and showcases applications using AI to better manage form intake and processing. Using optical character recognition (OCR), it is possible to process the volumes of senior exemption applications and condominium declaration forms received in paper and PDF formats. As with the other case examples, the results still require human oversight but provide a significant improvement over the existing process. The fifth use case also comes from the City of New York. This section discusses how geospatial data and AI methods are being integrated and leveraged to determine land use, detect building changes, and extract parcel data from images and may be used to automate data collection. This section also gives background on the geospatial data required to leverage land use and building change detection applications, which are growing increasingly familiar and important to tax assessment organizations. The sixth application involves integrating AI-powered valuation as a feature within CAMA. To illustrate the potential of AI to automate sales-based valuation models, this study examines Tyler Technologies’ experience trying to provide an AI-powered valuation option for its users. It also clarifies the technology’s perceived limitations, which create headwinds for widespread adoption. This section ends with a discussion of international adoption of AI in property assessment offices in four African nations: Rwanda, Nigeria, Uganda, and Zambia. The full digitalization of their records and workflows using imagery and modern technology allows them to modernize their systems without going through and updating legacy records and operational processes found in more established assessment jurisdictions. Following the case studies, the reader will find a section delving deeper into the core machine learning (ML) and AI methods underpinning these applications. ML is covered in the first part of this section. Other methods discussed cover key concepts in artificial neural networks and search and optimization, which underpin virtually every AI application. Finally, the paper closes with recommendations. Key takeaways are that some tax assessment organizations and their partners are already cautiously adopting AI. The technology’s adoption will grow more widespread and touch every tax assessment organization. As such, familiarity with how it is being used, a basic understanding of what is driving these changes, and what they mean for your organization are important.

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.