A Vancouver real estate agent uses artificial intelligence to help sell homes and invents a tool that pulls data from every multi-listing (MLS) property in the Lower Mainland to provide positive and negative analysis for buyers and sellers.
Richard Morrison, a local real estate agent for 20 years, led a team that spent months creating AI PropertyIQ, which merges the MLS data with GPT-4 OpenAI technology, which analyzes information and converses with users in an chat based system to tailor their prospects.
“You should essentially be able to check that you’re making a good purchase,” Morrison said.
“This data already exists on certain websites, but it’s just data — what changes this is that it’s analytics.”
While the tool can be useful, an AI expert warns users about the technology’s limitations.
says Morrison the tool on its website accessible to everyone and that hundreds of people use it every day, with all kinds of questions.
“A user was looking for a particular property and asked, ‘Can I subdivide this property?'” he told Breaking:.
“The engine can’t tell you that specifically at this point, but I can go to the back and train it so that it can look up the zoning laws in the Vancouver area. So it will let the seller or buyer know what they can do with this kind of zoning.”
It also has the capacity to perform competitive analysis. For example, if you’re looking for a two-bedroom apartment in downtown Vancouver, it will soon be able to compile sales data for the past three months to provide ballpark figures on comparable apartments sold in that area, based on market trends. It can also consider amenities such as a swimming pool and a concierge.
On the privacy side, Morrison said the questions aren’t tied to a user’s IP address. So while he knows the questions being asked of the system, he doesn’t know who is asking the questions.
One of the first users of AI PropertyIQ says it has been helpful.
“I’ve played around with ChatGPT before, but seeing that it matched a real-world scenario coupled with a business focus made it seem very obvious,” says David Macdonald, who is also one of Morrison’s clients.
“As if, of course, this is next in the evolution of technology.”
Morrison said he consulted an attorney before launching the tool, then scaled it back after investigating the Real Estate Services Act.
“I trained it not to give specific advice,” he said. “I’m not acting as their agent, so it’s still a gray area, and I certainly don’t want to be responsible for the machine giving advice.”
Morrison posted a disclaimer on the site, saying anyone using the tool does so at their own discretion. He wrote one blog post about the future of AI in real estate, including ethical considerations.
A warning
While these kinds of tools can be useful, an AI expert warns that there are limitations.
“It’s still pretty early days for these kinds of systems,” said Steve DiPaola, a professor in the School of Interactive Arts & Technology at Simon Fraser University.
He said that while it can be useful for things like asking if the price is in line with other home sales in the area, or where the nearest parks or grocery stores are, he cautions against relying on the tool for more complex questions.
“For example, what are the schools like? And not just the level of schools, but suppose my son needs a very special type of school – I hear there’s one here, would it even work for him? That’s really not going to be because of this level of AI at the moment.”
It’s still pretty early days for this kind of system… I certainly wouldn’t use it as the only way you decide to buy a house.”– Steve DiPaola, SFU School of Interactive Art and Technology
He added that GPT4 can make mistakes and it can be risky to put a lifetime investment into it.
“It’s an exciting space, but I’d caution people to do your homework; I certainly wouldn’t use it as the only way you decide to buy a house.”
For Macdonald, there are also intangible assets to consider.
“Buying a house, such an investment, is also an emotional commitment,” he said, “and technology isn’t very good at emotion.”