This report is available exclusively to subscribers of Inman Intel, a data and research arm of Inman offering deep insights and market intelligence on the business of residential real estate and proptech. Subscribe today.
With the advent of the artificial intelligence tool ChatGPT at the end of 2022, every day comes with yet another news story (or six) about AI and how it will change our lives, for better or for worse.
Given the varying views of AI in society at large and the tech’s seemingly limitless potential to generate content, Intel caught up with Jonathan Klein, founder of PropTech Consulting and self-described AI advocate, to discuss the role AI could play in residential real estate as well as its challenges in an industry built on data.
This interview has been edited for length and clarity. Responses are gleaned from a phone interview and follow-up comments provided over email.
Intel: Would it be accurate to say that you’re a champion of AI?
Klein: I wouldn’t consider myself to be a champion of AI. I’d say an advocate for sure. But yeah, it’s more than the flavor of the month right now. It’s probably the flavor of the year, to say the least.
What makes you say that?
Whether it’s generative AI or just AI as a whole, all industries are seeing the power and realizing that if they’re not investing into it, that they’re going to be left behind.
What do you think of the open letter circulating in the tech industry asking for a pause in the development of more powerful AI systems than the current version of ChatGPT?
I’m not intimately familiar with what you’re referring to other than like the concept itself, but I would say that the only way to learn and protect ourselves is by actually testing it and training it. So to pause something — you’re deferring the inevitable. I think it’s best we sort of proceed with caution versus pause anything. Because this technology and developers and people all around the world, they’re going to do what they want to do regardless. It’s just a matter of what technology tools they use, so it’s only to our advantage to actually press forward, focus on the positive and learn from the negative.
What would you tell a broker who’s skeptical about adopting an AI tool?
I would say try it first and then tell me what your hesitance is. Because at the end of the day, that’s half the battle with most technologies and most things that are new, is just getting people to just go against the status quo. You have two different buckets. There are the top performers and then there are those that are maybe a little bit more casual. The way that you can even the playing field is by incorporating some of this technology. That would allow for there to be more of an even distribution or just fuel teams to be more efficient and maybe allow for new agents to focus more on their strengths, such as selling, versus things like marketing that can be somewhat automated.
What impact do you think AI will have on residential transactions, how they’re handled?
The data quality should get better. There should be some biases that are removed and added to the system. The bias factor is definitely one where it goes both ways. Privacy is super important. That premise there is just privacy versus potential. Which side are you on? If you want to protect your data, or do you want to share it and maybe get more use out of it? Then probably the adoption factor. Just getting people to adopt things is always the hardest part. But I think that if it’s fully embraced properly, the transaction should be faster. [Transactions] should be better. There should be more transparency just across the board.
What do you think will make them faster, better and more transparent?
The human element is somewhat removed from the process, but instead, it’s adding in a layer of expertise on top of information that is maybe considered to be generic. Ninety percent of the listing could be done by the algorithms in the software and then that expertise can be sprinkled in on top of it to make it go the extra mile, have that local orientation, whether it’s language or insights or perspectives that simply are too advanced for the systems to understand to date. You’re going to see a lot of people using the technology just to basically get a head start.
Are you referring to things like writing listing descriptions? Are you referring to handling some of the very copious amounts of paperwork that come with a residential transaction? Are you referring to marketing? Can you be a little more specific?
Yes, there are so many. The marketing factor is probably the most well-documented in the real estate industry, at least, for agents — writing listings, promoting listings. The more you can cross-reference datasets with your data or third-party vendors that you may be working with, the more intuitive you can be on following up with people that actually have an intent to buy or sell.
Just having more accurate lead generation and better intuition on what you’re looking to accomplish — that’s the main marketing side of things. On the B2B side, it’s a little bit more on the data analytics or predictive analytics side where the valuations of properties and things of that nature are tapping into some of these rich datasets.
Generative AI is a subset of artificial intelligence that involves using machine learning algorithms to generate new data based on existing data. In the context of real estate, generative AI can be used to create realistic 3D models of buildings or to generate new floor plans based on existing designs.
What do you think are the challenges of AI when it comes to residential real estate right now?
Here are some of the key challenges:
- Data quality: One of the biggest challenges in AI for real estate is the quality of data. Real estate data can be messy, incomplete, and inconsistent, which can make it difficult for machine learning algorithms to generate accurate predictions.
- Bias: Another challenge is the potential for bias in AI models. If the training data used to build the AI model is biased, the resulting predictions will also be biased. This can lead to unfair or discriminatory outcomes.
- Interpretability: AI models can be difficult to interpret, which can be a problem in real estate where decisions can have significant financial and legal implications. Stakeholders need to understand how AI models work and be able to interpret their results.
- Privacy: Real estate data can contain sensitive information about individuals and properties. It is important to ensure that AI models are developed and used in a way that protects privacy.
- Adoption: Finally, there is the challenge of adoption. Many stakeholders in the real estate industry may be hesitant to adopt new technologies like AI, especially if they do not fully understand how they work or if they perceive them as a threat to their jobs.
Overall, while AI has the potential to revolutionize the real estate industry, several challenges need to be addressed before they can be fully realized. Addressing these challenges will require collaboration and innovation from stakeholders across the industry.
AI is very much inspirational. You don’t want to give the generic inspiration, but more so use it as a starting point. Another popular tool is Midjourney, where basically you can ask it to, for example, draw a futuristic cityscape of New York City and a skyrise building in a Tesla architecture style. Within seconds, it’ll have multiple versions for you of that. It’s really just an inspiration, a thought starter. Obviously, it’s not going to be perfectly to scale in terms of actual building practices, but in terms of just a quick look and an idea, it’s pretty impressive.
If you ask it to write a listing description for you, it’s not like you can just paste that listing description in and be done. You actually have to check it to make sure it’s not making things up, right?
Of course. The other part of the problem is making sure that we’re not training these algorithms off of incorrect information, because then it puts us down a slippery slope where we’re teaching people incorrect information.
Do you think there’s a danger that agents will begin to over-rely on AI to do things for them instead of picking up skills on their own?
It’s just another way to learn. If you don’t want to learn and if you want to cheat, you can obviously cheat and not learn, but if you actually want to learn, it just basically enables that for those that maybe are better listeners than readers, for example, or just simply need to be trained in a certain type of way. I think it really just opens up a lot of opportunities for a whole other way of learning.
Also, I think it kind of evens the playing field from traditionally where a lot of software development jobs were getting these crazy pay scales and then there were a lot of blue-collar jobs that were not really getting the same pay that had been the case in years past. With this type of technology, developers can write code 50 percent automated and 50 percent based on their actual expertise. Hopefully, there’s a pendulum shifting in job opportunities and pay structures.
