Where AI Stops in Commercial Real Estate, Judgment Begins

In this column on July 14, I wrote about something I believe is going to become increasingly important in business , particularly in commercial real estate: The value of information is declining while the value of judgment is increasing.

The premise was relatively simple. For most of my career, information was a great differentiator and had tremendous value because it was difficult to obtain. Knowing who owned a building, what it sold for, what the zoning allowed, what comparable properties traded for, who the active buyers were, what financing was available and what was happening in a particular submarket gave you a meaningful competitive advantage. 

Today, much of that information is available instantly, and artificial intelligence is accelerating that trend at an extraordinary pace. 

After that July 14 column was published, someone on social media asked me a great question: Can you provide practical examples of situations where judgment took you further than where AI stopped? Absolutely. In fact, investment sales brokerage provides examples almost every day because AI can increasingly replicate the information layer of what we do. The judgment layer is something entirely different.

Take something as fundamental as determining the value of a building. AI can digest comparable sales, rents, expenses, cap rates, interest rates, zoning, property taxes and dozens of other variables, and probably produce a very credible valuation range. But suppose the answer is $40 million. The more important question may be: Should this particular owner sell the building for $40 million today? That requires understanding the owner’s basis, debt, taxes, partnership structure, family situation, alternative uses for the proceeds, appetite for risk and view of the market. 

There have been many times during my career when someone asked me what their property was worth and my ultimate advice was: Don’t sell it. AI can help determine the value. Judgment determines what you should do with that information.

Consider another common situation. We market a property and receive 10 offers ranging from $47 million to $52 million. Technology can organize those bids beautifully and AI can rank them instantaneously. But is the $52 million offer really the best offer? Maybe. Maybe not. Who has the equity? Who requires financing? Who has investment committee approval? Who has retraded sellers in previous transactions? Who understands the problems with the building? Who has a history of closing difficult transactions? Who is stretching simply to win the bidding process and planning to renegotiate later? 

I might recommend that a seller take $50 million from Buyer B rather than $52 million from Buyer A. A spreadsheet might conclude that we left $2 million on the table. Judgment might tell us that we dramatically increased the probability of actually closing the transaction. The highest offer and the best offer are not always the same thing.

Comparable sales provide another example. AI can identify 20 properties that appear statistically comparable to the building being valued. After 42 years specializing in this market, I might look at those 20 transactions and immediately eliminate 15 of them. 

One buyer desperately needed the property next door. Another seller was under unusual financial pressure. One transaction included favorable financing. Another had development potential that wasn’t obvious from the public record. Another involved a buyer with a strategic motivation that caused them to pay more than an ordinary investor would. The computer sees 20 data points. Experience tells me that only five of them actually matter. 

Information identifies the comps. Judgment determines which comps deserve weight.

This becomes even more interesting with vacant buildings, which are a major focus of our business today. AI can calculate what a traditional investor should pay based upon the income a property can generate. But what if the best buyer isn’t an investor? What if it is a school, hospital, university, nonprofit, corporation, religious organization or wealthy individual that needs the property for its own use? That buyer may look at value completely differently. 

After analyzing more than a thousand user transactions over my career, we have observed that users frequently pay significant premiums over investor pricing. Why? Because investors generally ask, “What return can this building generate?” A user may ask, “How is owning this building going to impact my business and the perception of my business?” Those are completely different questions. AI may accurately value the real estate as an investment. Judgment helps identify the buyer for whom the real estate has strategic value.

Development sites provide another great example. AI can tell me the zoning designation, allowable floor area, permitted uses, available air rights and theoretical development potential of an individual parcel. But some of the best development transactions I have worked on began with a question that wasn’t contained in the data: What if we combine this building with the one next door? What if we purchase air rights from another property? What if a retail tenant can be bought out? What if an easement solves a light and air problem? What if three ordinary properties can be assembled into one extraordinary development site? 

AI is extremely good at analyzing the pieces sitting on the table. Judgment sometimes means imagining a table that doesn’t exist yet.

Prospecting may illustrate the changing value of information better than anything else. When I started in 1984, simply figuring out who owned a building could require meaningful work. Today, databases can produce thousands of owners, addresses, phone numbers, transactions and property characteristics almost instantaneously. AI will make that process even easier. 

But give a young broker a list of 2,000 owners and another question immediately appears: Who should I call this morning? Which 20 deserve my attention? Which ownership situations may be changing? Which conversation from six months ago deserves another call today? Which owner is likely approaching an inflection point? 

Information gives you the 2,000 names. Judgment tells you which 20 matter today.

Perhaps the most important example, however, has nothing to do with buildings. AI is extraordinarily good at answering questions. Judgment helps determine whether we are asking the right question in the first place. 

An owner may ask me, “Bob, what’s my building worth?” Before answering, I may ask: “Why are you asking? What are you trying to accomplish? What happens if you don’t sell? Who else is involved in the decision? What would make you regret selling?” 

Sometimes the greatest value an adviser provides isn’t answering the client’s question. It is discovering that the client’s original question wasn’t the one that actually needed answering.

After more than four decades and thousands of building sales, my advantage is not that I possess more raw information than artificial intelligence. Increasingly, I won’t. AI may ultimately have access to more information than any human being could possibly absorb. 

My advantage is that I have watched thousands of decisions play out. I have seen buyers overpay and prosper, and buyers overpay and get crushed. I have watched sellers reject offers they later desperately wished they had accepted. I have seen seemingly certain deals collapse and seemingly impossible deals close. I have watched interest rates soar and fall, financing markets freeze, neighborhoods transform, zoning change, partnerships fracture and conventional wisdom prove spectacularly wrong. 

Those experiences create pattern recognition, and pattern recognition accumulated over decades becomes judgment. That is why I don’t believe AI makes expertise less important. I believe it makes genuine expertise more valuable. When everyone has access to essentially the same information, competitive advantage migrates to what you do with it. 

AI can shorten the distance between a question and an answer. Judgment shortens the distance between an answer and the right decision.

Robert Knakal is founder, chairman and CEO of BK Real Estate Advisors.