AI, appraisals and appraisers - by William Pastuszek

 

It’s not a generalization to say that appraisers and the appraisal profession are deliberative, and can be hesitant to change without proven reasons to do so. But change is here. Appraisers are now faced with evaluating AI technology and how the technology can-and-will-impact appraisal theory and practice. This technology has evolved in the past couple of years to have capabilities that appraisers can use to their advantage. As USPAP says, “It is not sufficient for appraisers to simply maintain the skills and the knowledge they possess when they become appraisers. Each appraiser must continuously improve his or her skills to remain proficient in real property appraisal.” AI provides the challenge to improve and remain proficient. 

What are some responsibilities appraisers have in working with AI? A USPAP Advisory Opinion (#37) advises appraisers, “when using any of these analytical tools, the appraiser is responsible for the accuracy of the output. Thus, the appraiser must have confidence that the technology uses data that is relevant and that the output is mathematically correct and sufficiently reliable for use in the assignment.” It goes on to say “the appraiser is responsible for the entire analysis and …. should use sufficient care to avoid errors that would significantly affect their opinions and conclusions. Diligence is required to identify and analyze the factors, conditions, data, and other information that would have a significant effect on the credibility of the assignment results.”

Utilizing AI can be an important time saver. A great use is allowing the program to “edit” - and possibly “enhance” - written material. But just like taking information from a market survey or a media article requires attribution to avoid the suggestion of plagiarism, the sources and data AI utilizes need attribution. There is also a clear obligation to verify that the data and analysis offered by AI are correct. Appraisers need to remain skilled in data searches and in determining what data is the most relevant and reliable. AI search and analytical results might also require that an appraiser needs some skills in other analytic software, such as Excel or Argus, in order to be able to undertake this back-checking. 

At this point in time, AI platforms are very good at some things - great meeting summaries, for example - and not so good at other things - evaluating small data sets qualitatively. Appraisers have to sort out what works and what doesn’t on an individualized basis. Using AI without applying judgment and critical thinking is a recipe for disaster.

Jim Amorin has written a book - The Generative Shift: Preparing Appraisers for Artificial Intelligence Models Like ChatGPT - that helps real estate appraisal professionals begin to understand and work in “Generative AI.” This volume appears to be an excellent introduction into the complicated evolution of AI. The book deals more specifically with generative large language models, and their practical applications in real estate. Someone reading this book can gain understanding of how AI tools, including how ChatGPT, Google’s Gemini, Anthropic’s Claude, and Microsoft’s CoPilot, can assist in appraisal. More books, guides, courses, and seminars are on the way.

Appraisal education and educators also need to incorporate into their courses and teaching how appraisers can (and should not) use AI. It is and will remain a basic and growing tool that appraisers and their clients need to understand. We will all need to understand how AI works – to recognize when AI has been used in a report – and to judge the quality and accuracy of the results of an AI “hand” in our work.

Appraisers can tend to be slow adopters, confident that current techniques are “tried and true.” But not understanding AI - or choosing not to - may be as bad as rushing blindly into it for the sake of assumed efficiency and expediency. There are powerful AI tools out there. They should not be ignored. On the other hand, believing that AI can replace critical thinking and human judgment – factors integral to excellence in appraisal – is a wrong direction as well. Be deliberate and prudent but don’t stampede with the herd. Facing change honestly and with facts in hand will allow for a methodical, logical, and efficient assimilation of these technologies.

AI does not exist to be a simple a timesaver. For appraisers, it can help with identifying, gathering, and sorting data. It can help edit written narrative material. It can create charts and graphs. It can perform calculations. But AI cannot automatically provide appraisers with correct discussions, analyses, or conclusions. That responsibility belongs to the ever-present human factor-each of us. If anything, using AI should make us as appraisers more diligent than ever to make sure our presentation and conclusions are reliable, accurate, and supported. 

Next Month: Appraisers’ obligations when using artificial intelligence.

William Pastuszek, Jr., MAI, ASA, MRA heads up Shepherd Associates LLC, Needham, Mass.