Intelligence Appraisal uses human intelligence combined with artificial intelligence.

The human intelligence role combines art, knowledge, and skills.

• Art involves how to interface with the AI agent, knowing the underlying theory of the analysis. This can be creative, and recognizes emotional aspects of human behavior.
• Knowledge means both the subject matter, and the right use of AI.
• Skills, for asset counseling, require ability to interface with the AI agent, by building on the theory and the subject matter science: the models, the logic, and algorithms.
• Finally, if the work is for a user/client, effective communication requires ability to explain in words, in visuals (graphs), and the clear use of math and logic — including the implications of uncertainty and risk in any such analysis.

Intelligence is reasoning, problem-solving, learning (from experience), and adjusting to new situations.

Human intelligence is cognitive.  It is perception, memory, and organizing — to process data into information.  Artificial intelligence simulates human intelligence.  The computer enables its learning, reasoning, problem solving, and making decisions.  Predictive AI focuses on specialized tasks, like identifying a good data set (like a Competitive Market Segment – CMS), or predicting/adjusting features of a “comparable” sale.  Generative AI can create new, original content of all “artsy” types.  This includes text, pictures, videos, music, and even software code.  SOFTWARE CODE – wow.

This last genre of AI, sometimes called “vibe coding,” is the future of the appraisal profession.  Vibe coding means the analyst speaks or types, using natural language. This applies to problem identification, data selection, prediction/adjustment, and brain/machine interaction.  All the key parts of valuation.

So the analyst must have skills and specialized knowledge, including the underlying economic theory and data-centric models.

This brings us to the most important concept and rule of intelligence valuation: AI does not do well with vague, poorly defined, or sweeping instructions.  Like “pick some comps, adjust them – and give me a number for an opinion.”  My opinion.  Oh yeah, also help me figure out how to “support” that opinion!

Instructions and “prompts” work best with clear, logical, and defined words.  No artsy stuff.  Its gotta be detailed, sequential, coherent, cogent, and consistent.

AI  never does well with today’s fuzzy definitions of a comparable.  We are told a comp must be: “similar, competitive, and able to be compared.”  These words alone are too generalistic, and themselves need much better clarity.

This clarity is not available to us in traditional appraisal writings, classes, and in “accepted” methods.

These words “similar, competitive, and able to be…” themselves require deeper understanding.  These broad concepts themselves require more parsing and reduction and exactness to smaller pieces and parts.  Parts which can be logically organized and grouped, associated, or sequenced.  Gotta be specific.  Gotta be in small parts.  A balance between science and art.

Human intelligence is needed to work with artificial intelligence.

This is part 2 of a deeper look at the needed intelligence for the future of valuation and risk analyses. Read part 1 here.

In the Valuemetrics.info curriculum, starting with the “Stats, Graphs, and Data Science + AI” course, we build these tools, attitudes, and principles needed for success in a rapidly changing world.