While AI is a sweeping topic, we need to look back.  Look back at the fundamentals of appraisal theory and what makes a good valuation different from a bad valuation.  In exploring this “goodness,” we must consider all the topics and concepts that have entered the world of appraisal – Automated Valuation, Data Science, and the “measurement” of goodness.

I suspect that this melded topic will begin another series of blogs.  A series which will put each topic into its conceptual box, but also show connections, relationships, and conflicts.  In the end, I hope this path of thinking, exploring, and feeling our way — will lead to possible ideal solutions and a place for the valuation function.  The place is public policy; the overarching issue is control.

A first thought of why public policy is important, is the reality of today’s context of regulation.  Real property, (my home) reaches from government control to the overarching status of citizenship and fairness.  The ultimate question:  Where is the balance between “to each their need” (a communistic concept), and the ownership of the resources which go beyond “need”?

We start with these concepts around wealth measurement (AI, AV, DS, BS).

To form good policy — to form “fair” policy – we must first establish an understandable and consistent manner of its measurement.

Here, we start by defining and describing our four topics:

• AI is artificial intelligence. A broad topic, encompassing all human knowledge and beyond.  Our main concept here is that AI enables dramatic improvement in the quality of the four basic steps of research and analysis of valuation.
• AV comprises automated valuations, commonly called AVMs. AVM is really more an industry model than a specific value measurement model.  This is interesting because the process — the sequence of analytic steps – is identical to traditional appraisal.  The main difference is who makes the judgment decisions given the visibility of the algorithm is zero.  The AVM industry requires algorithmic trade secrets.  If one provider discovers an improved model/algorithm, it must be kept secret, lest a competitor immediately steal the better process.
• DS is data science. Data science, applied to valuation, we call EBV (Evidence Based Valuation)^©.  The shift from traditional legacy appraisal thinking is simple:  1) Clarify and quantify the problem parameters (scope of work);  2) Define and apply all relevant competitive data (not just a handful of “good judgment” comps);  3) Apply data adjustment and predictive methods to that CMS (Competitive Market Segment)^©, and 4) Provide visual and interactive understanding of the inclusive market analysis resulting from the use of the CMS.
• BS is the outgrowth of the current requirements of USPAP (Uniform Standards of Professional Appraisal Practice), which mandates review/audit be based on one word: “credible.” Credible is defined by the quasi-governmental Appraisal Foundation as “worthy of belief.”  Thus, any idea of measuring appraisal quality will require a “Believability Score” (BS).  Per the relevant review standard, the score (also an “opinion”) would need to address “completeness, accuracy, adequacy, relevance, and reasonableness” of the reviewed report.  (Quite an assignment!)

We hope we can encourage attention to these four imperative topics, and even to the extent better solutions might fairly present issues of fairness, analytic biasedness, economic meltdowns, and even homelessness.