When the market needs insight, not just input

In my last column, I wrote about how artificial intelligence was going to do lots of great things and make buying and selling real-estate a breeze.

For this column, I asked AI to do a market trend analysis of the Kelowna B.C. real estate market.

First up was Grok. It came up with about 1,200 words, which I then asked it to summarize to 150 words. This is what it said:

“The Kelowna, B.C. real estate market in early 2025 shows signs of recovery. Sales rose significantly, with January 2025 single-family home sales up 41% year over year, reaching 110 units, and October 2024 saw a 60% increase. Benchmark prices increased modestly, with single-family homes at $1,030,600 (up 5.4%) and condos at $517,700 (up 4.3%), though still below 2022 peaks. Inventory levels are high, with nine months of listings in September 2024, creating a buyer’s market, especially in the luxury segment. Interest rate cuts to 3% by January 2025 have boosted affordability, while new short-term rental regulations increased condo supply. Economic growth in the region was slow in 2024, but realty insiders predict a 4% sales increase and a 3% price rise in 2025. Affordability remains a challenge, and while move-up buyers drive mid-to-high range activity, first-time buyers focus on entry-level properties, signalling a cautiously optimistic, balanced market.”

Next, I tried ChatGPT. It did a 260-word analysis. Which I also asked it to condense to 150 words. This is what I got:

“The Kelowna real estate market is showing signs of renewed strength in early 2025. January saw a surge in sales activity, with single-family home sales up 41% year-over-year, townhouses up 35.7% and condos rising 10.5%. Benchmark prices have also increased: $1.03 million for single-family homes, $742,000 for townhomes, and $517,000 for condos. Milder weather, stabilized interest rates and improved buyer confidence have contributed to the market’s momentum. As of Feb. 1, 146 homes were listed for sale—6.8% more than the previous month—giving buyers more choice and indicating a more balanced market. While activity is rising, the market remains sensitive to affordability concerns and potential future rate changes. Overall, Kelowna’s real estate market appears to be recovering steadily with cautious optimism.”

Finally I asked Meta, which was the most concise weighing in at 215 words. Again I asked it to trim it down to 150 and here is the result:

“The Kelowna, B.C. real estate market is experiencing a mix of trends. Home prices have been relatively flat, with a median list price of $977,122 and an average price per square-foot of $486. The market is leaning in favour of buyers, with active listings at their highest level in three years and a sales-to-active listings ratio below 0.25. New listings are rising but home sales have been slower due to economic uncertainty. Mortgage rates remain relatively high, limiting buyer budgets and the rental market is experiencing upward pressure. Overall, the market is characterized by flat home prices, a buyer's market and rising new listings. For expert analysis, consider consulting a local real estate agent or economist.”

The first two are similar, with Meta straying and being more pessimistic. They all cited a few sources in their initial responses so I could see where they were getting some of the information.

In my opinion, a problem with the AI analysis is it depends on published open-source data that is often written by organizations that are cheerleading the results, hoping to assuage fears or massage the data to give their perspective. After all, if no one is buying real-estate, some businesses aren’t making any money, right? (I see what I believe to be just that in lots of real-estate articles).

I do not believe current raw MLS data is available, so the AI programs need to get the information from somewhere and if that information is inaccurate, well GIGO (garbage in, garbage out).

If you wonder why I don’t think AI has access to raw MLS data, I asked other questions and often I got the response, “I do not have access to that data.”

Real estate agents can pull data from the MLS system and pick it apart using their knowledge of local areas, as well as trends in pricing. Let’s say there is a large exclusive condo development that has several expensive units that sold recently. If you don’t take that into account, you may conclude pricing is going up for condos. But actually it is an outlier skewing the average.

I am sure many situations can cloud the waters when it comes to what is happening in a local real-estate market. I think having a living breathing person with local experience is still the best bet if you want advice on what is happening in your area.

There are still some headwinds for AI when it comes to being able to give advice about local market trends. When it can look at, and query, different databases, things may look different. Until then, I would stick with a local flesh and blood alternative.

Score: AI 0, humans 1

If you have suggestions for other real estate-related articles, please email me at: [email protected]

This article is written by or on behalf of an outsourced columnist and does not necessarily reflect the views of Castanet.