Intelligence is “the ability to acquire and apply knowledge and skills.” For valuation and risk decisions, we have three relevancies:
- Human intelligence applies research, market knowledge, and proper methodology skills.
- Artificial intelligence (Ai) follows instructions and context provided by the analyst.
- Human art – is an expression of ideas, intuition, and skill.
This Modernizing Appraisal series is about the analysis process, not FannieMae and FreddieMac reporting requirements. The process is identical: 1) The problem; 2) The data; 3) Predict or adjust; 4) Communicate. We continue here on #1 – defining the problem to be solved. Read the earlier parts of this series here.
Humans rely on prior established intelligence. Ai relies on prior established intelligence.
Artificial intelligence requires certain things. These are accomplished via the art of prompting. Guidelines for well-designed prompts include:
- Clear intent – the goal or result you want: like form, structure, or appearance of the result.
- Context – background, constraints, goals: “We are doing market analysis for valuation.”
- Format – the output style: language, text, tables, statistics, graphs.
- Boundaries – what is not necessary, or is to be avoided.
- Reproducibility – designed for any future reuse?
- Citations and sources – reference handling.
Generally, this means: Role framing, how to proceed, guardrails, results appearance.
This requires competence in the field of study, prompt engineering, and related ethical issues.
For appraisal, the competencies using Ai include an understanding of:
- Appraisal theory and appraisal process
- Appraisal ethics and regulatory compliance
- Minimum standards and client/user requirements
- The above six prompting guidelines, as applied to appraisal.
Automated Valuation Models (AVMs) – which fill a large part of the “market” for collateral risk decision-making – are substantially obscure as to the actual algorithms used. As such, if there is any use of Ai, it may or may not involve any human expertise in the analysis nor the prompting decisions.
Similarly, “non-appraisal” appraisals – also completed without appraiser competence, licensing, or oversight – can conceal or simply ignore any guidance or competence in the use of Ai for valuation or risk management.
On the other side, there is great benefit from the use of Ai – with good prompt engineering when based on data science principles – when applied to valuation. We call this Evidence Based Valuation (EBV)©.
EBV, as taught in the Valuemetrics.info curriculum emphasizes the use of repeatable models, computer algorithms, and visual brain-machine interaction.
Artificial intelligence, properly supervised, improves and enhances human expert judgment. Those appraisers who get to today’s technology will thrive. Others will not.