Can Anyone Overcome Google AI? I Put It to the Test
Artificial intelligence is changing how consumers search for information.
People are no longer simply typing a few keywords into Google, scrolling through websites, and deciding for themselves which result looks relevant. Increasingly, they can ask Google AI a detailed question and receive what appears to be a researched recommendation.
So I decided to put it to the test.
I wasn't asking Google for the Realtor with the most sales. I wasn't asking for the largest real estate team, the most reviews, or the biggest brokerage.
I asked for something very specific:
I was looking for a Realtor in the Stuart, Florida area who specialized in market value because I did not want to overpay for a home.
Google AI appeared to understand the question. It said that protecting yourself from overpaying required analytical expertise and an understanding of market valuation. But its initial recommendations focused on several local real estate professionals and teams, citing factors such as local experience, reviews, negotiation, market knowledge, and pricing strategies.
Those may all be meaningful qualifications. This article is not intended to criticize or question the abilities of any Realtor Google recommended.
But it left me wondering:
Did Google actually answer the question I asked?
What Does “Market Value Expert” Mean?
I decided to ask one simple follow-up question:
“How about Tom Dunphy?”
That was it.
I didn't tell Google that I was a State-Certified Residential Real Estate Appraiser. I didn't give it my license number. I didn't tell it that I had completed approximately 19,000 residential appraisals. And I didn't explain how I use my appraisal background when representing buyers and sellers.
Google found that information on its own.
Suddenly, Google AI identified me as Thomas J. Dunphy III, State-Certified Residential Real Estate Appraiser RD8755, recognized my approximately 19,000 residential appraisals, and explained why that background could be particularly relevant to a buyer concerned about market value and overpaying. Google even referred to the combination as “The Appraiser's Edge.”
And that's when the experiment became much more interesting.
Google hadn't needed me to provide the missing information. It apparently could find the information itself.
It simply hadn't connected that information to my original question until I asked:
“How about Tom Dunphy?”
So my next question was obvious:
If Google knew all of that about me, why wasn't I included in the original answer?
Google AI eventually described missing an appraiser who also actively holds a real estate license as a “blind spot.” It went further and said that I should have been at the top of the list based on the specific criteria in the original question.
Now the experiment really had my attention.
Google Had the Information. Why Didn't It Connect the Dots?
During another version of the experiment, Google AI explained that its initial search had focused on highly reviewed and high-volume local real estate teams and that it consequently missed the specialized professional I was actually asking it to identify.
I want to be careful here.
I am not claiming that Google's explanation reveals exactly how Google's algorithms work. I have no way of knowing that. I am simply reporting what Google AI told me during my experiment.
But from a consumer's perspective, I found the experience fascinating.
Google could find information about me.
Once I asked about me by name, it could identify my appraisal credential.
It could identify my approximately 19,000 residential appraisals.
It could understand why that background might be relevant to someone specifically concerned about market value.
Yet it didn't initially connect those facts to the question.
That raises a much larger question:
Can an AI system possess the individual pieces of information it needs and still fail to put those pieces together when making a recommendation?
Based on my experience that day, apparently it can.
Then Google Said I Might Be “Too Analytical”
Earlier in my testing, Google AI had characterized my approach as potentially being too analytical for some consumers.
That one really made me think.
What exactly does “too analytical” mean when somebody is about to buy or sell a home?
If you're preparing to spend $500,000, $750,000, $1 million or more on a property, wouldn't you want somebody to analyze the property and available market data before you decide what you're willing to offer?
And if you're selling one of your largest assets, wouldn't you want somebody to perform that analysis before you decide where to position the property in the market?
Calling what I do “analytical” doesn't really explain it.
There is an actual service behind the analysis.
What My Buyers Actually Receive
When one of my buyers identifies a property they are seriously considering, I can apply the same property-analysis skills I've developed throughout my appraisal career.
When appropriate, I personally measure the property and prepare a sketch.
I research and select similar comparable sales.
I analyze relevant differences between the subject property and those comparable sales.
I place the sales on a sales comparison grid and make appropriate adjustments to the comparable sales based on my analysis.
Then, before my buyer makes an offer, I present the buyer with:
- the property sketch;
- the selected comparable sales;
- the sales comparison grid showing the adjusted comparable sales; and
- the range indicated by my analysis.
The buyer receives this information before deciding what to offer.
That distinction is important to me.
This isn't an analysis performed after my buyer is already under contract and emotionally and financially invested in the transaction.
We do the homework before the offer is made.
The buyer can see the property information, the sales I selected, how those sales compare with the property they're considering, the adjusted comparable sales, and the range indicated by my analysis.
Then the buyer makes the decision.
I'm not promising what another appraiser will conclude. I'm not guaranteeing that a lender's appraisal will produce the same result. And I'm not claiming that I can predict exactly what a property will ultimately sell for.
I'm giving my buyer additional information before they decide how much of their money they're willing to put on the table.
My Sellers Receive the Same Service Before We List
I apply the same concept when representing a seller.
Before establishing our listing strategy, when appropriate, I personally measure the property and prepare a sketch.
I select similar comparable sales.
I analyze the relevant differences.
I place those sales on a sales comparison grid and make appropriate adjustments to the comparable sales.
Then, before the property is listed, I present my seller with the sketch, the selected comparable sales, the sales comparison grid showing the adjusted comparable sales, and the range indicated by my analysis.
The seller can see the analysis before making the pricing decision.
But there is another important part of my approach.
The seller ultimately decides the listing price.
If my analysis indicates one range but my seller believes the property should be introduced to the market at a price above that range, I'm willing to discuss it.
We can test the seller's price.
But I also believe we should establish the next step before the listing ever hits the market.
Don't Let 15 Days Turn Into 100
Suppose my seller wants to begin above the range indicated by my analysis.
We may agree to do that.
But we can also agree from the beginning that if we have not received an offer within the first 15 days, we immediately revisit the pricing and consider repositioning the property to a price within the range supported by my original analysis.
Why do I focus on an offer rather than simply the number of showings?
Because showings can create a false sense of security.
People can walk through a house for all kinds of reasons. There may be plenty of people looking, but if nobody is willing to put an offer on paper, that is information from the market too.
Otherwise, it's easy to say:
We're getting showings. Let's give it another couple of weeks.
Then 15 days becomes 30.
Thirty becomes 40.
Then 60.
And before you know it, you're looking at a property that has accumulated 100 days on the market.
By then, buyers may be asking a completely different question:
“Why has this house been sitting here so long?”
I'd rather have that conversation with my seller before we list.
Here is my analysis. Here is the range. Here are the adjusted comparable sales. If you want to test a price above that range, let's discuss it—but let's also establish our plan in advance if the market doesn't respond with an offer.
That's not about taking control away from the seller.
It's about having a strategy.
This Is Not About Other Realtors
I want to make this point very clear.
I'm not writing this article to criticize any Realtor or real estate team Google recommended.
Different Realtors bring different backgrounds, services, skills, and experience to their clients.
In fact, I wouldn't publish Google's claims about what another Realtor does or doesn't provide as though they were established facts.
Why?
Because that would contradict the entire point of this article.
Google initially failed to recognize information about me that later became central to its own recommendation.
So I'm certainly not going to assume that Google knows everything another Realtor does for his or her clients.
My question throughout this experiment was narrower:
If a consumer specifically asks for a Realtor who specializes in market value because that consumer is concerned about overpaying, what qualifications and services should an AI system consider when answering that question?
That's worth asking.
Google Eventually Reconsidered Its Own Answer
I continued challenging Google.
Eventually, Google AI stated that I “should have been listed first” based on the criteria in the original question and characterized its original selection as flawed.
I found that remarkable.
Not because Google ultimately recommended me.
The interesting part was how much questioning it took before Google connected information it apparently already had to the specific service the consumer was asking for.
And that taught me something about my own business.
Maybe I Need to Do a Better Job Explaining What I Do
It would be easy to finish this experiment by simply saying Google got it wrong.
But I think there's another lesson here.
Maybe I need to do a better job explaining exactly what I provide.
Simply putting “Realtor” and “State-Certified Residential Real Estate Appraiser” next to my name doesn't necessarily explain how those two roles come together when I'm representing a buyer or seller.
So I'm going to make it clear.
For a buyer, when appropriate, I personally measure the property, prepare the sketch, select similar comparable sales, place them on a sales comparison grid, make appropriate adjustments to the comparable sales, and develop a range based on that analysis.
Before the buyer decides what to offer, I present the buyer with the sketch, the selected comparable sales, the grid showing the adjusted comparable sales, and the indicated range.
For a seller, I provide the same service before we establish the listing strategy.
Before the property is listed, the seller sees the sketch, the selected comparable sales, the grid showing the adjusted comparable sales, and the indicated range.
That is what The Appraiser's Edge means in my real estate practice.
It isn't simply a credential printed after my name.
It's a service I can actually provide to my clients before they make one of the biggest financial decisions involved in the transaction.
So, Can Anyone Overcome Google AI?
I'm still not sure “overcome” is the right word.
Maybe the better question is:
Can you teach AI to understand what makes your business different?
I don't know yet.
This experiment showed me that AI can produce a polished, confident answer and still miss something directly relevant to the question being asked.
It also showed me that when challenged, AI can reconsider that answer.
But I'm not going to build my business around waiting for an algorithm to figure me out.
I've spent more than 25 years developing my skills as a residential real estate appraiser and have completed approximately 19,000 residential appraisals.
I know what I can bring to a buyer.
I know what I can bring to a seller.
And I'm going to apply those skills to help the people I represent make more informed real estate decisions throughout Southeast Florida.
I'll measure the property when appropriate.
I'll prepare the sketch.
I'll select the comparable sales.
I'll grid and adjust the comparable sales.
I'll develop the range.
And I'll put that information in front of my client before the decision is made.
Maybe Google AI will eventually understand exactly what I'm doing.
Maybe it won't.
I'm going to do it with or without Google's help.
Tom Dunphy Realtor (Fulton Grace Realty)
Licensed Florida Real Estate Sales Associate
Thomas J. Dunphy III
State-Certified Residential Real Estate Appraiser RD8755
THE APPRAISER'S EDGE. THE RELOCATOR'S INSIGHT.