Does artificial intelligence require appraiser intelligence?
This idea opens up a whole series of questions that need to be answered. A lot. And we need to be smart about how we look at this.
Are there other forms of intelligence? What is intelligence? Are you intelligent? Does valuation require intelligence? And if so, are there certain “intelligences” that we need more than others?
Is the appraisal “process” the most intelligent way to form an opinion of value. We might even look at whether an opinion is what appraisal client-users really need? Whew.
Gotta look.
What do appraisal users really want?
There are basically just four motives:
- Avoid a loss of collateral
- Avoid failure to achieve a gain
- Avoid a bad/unfair equitable decision
- “Make the deal” regardless of justification
In each of the first three cases, the decision maker: lender, investor, or adjudicator make their decision, their opinion – based on the appraiser’s opinion. Note the definition of an appraisal — an opinion of value. Is it possible that these decision-makers do not really want an opinion to form their opinion? Is it possible that they would prefer actual factual real logical analysis and information to help them form their opinion, their decision.
In the fourth “make the deal” motive, it is only important that the opinion be credible enough to pass.
Is it possible that building an opinion on another’s opinion is not intelligent? Even if it is a ‘believable’ ‘credible’ and ‘worthy’ opinion?! Recall that the goal of the appraiser’s opinion is to be credible – “worthy of belief.”
So now the user (lender/investor/equitor) is charged with two jobs: 1) decide on the action; and 2) decide on the worthiness/believability of the appraisal.
We call this “appraisal review.” (Or alternately, “administrative review”.)
But wait, there’s more! An “appraisal review” itself requires yet two more opinions: 1) the worthiness of the appraisal opinion and its development, and 2) the report’s compliance to administrative detail.
Finally, and mercifully, the review process itself is an opinion. An opinion of another appraiser. A reviewer-appraiser who also forms an opinion of value (whether declared or not), in order to form an opinion of accuracy of that value opinion.
Artificial intelligence has difficulty mind-reading, whether the human has intelligent opinion, or not.
In coming editions, we look at the roles of human and artificial intelligence, and their interplay. We note that it seems likely that we will be forced to look more at appraisal science, rather than on appraisal art.
This is data science. In our curriculum, we call this “Evidence Based Valuation” EBV©.
EBV recognizes expert judgment. It is just sharper and crisper than the “trust me” version.