AI Needs Thoughtful Regulation to Prevent Catastrophe

It was 2016 when Jeff Turner, cofounder of Tangilla, a REALTOR® association management platform, gave his first presentation on artificial intelligence. At the time, Siri and Alexa were the only real examples of AI, so the urgency he conveyed wasn’t received by the audience.

Today, there are ChatGPT, apps like HeyAnnie, and websites like Humanistic.ai and heypi.com. AI is transforming the world at a much more rapid pace, and real estate is no exception. The foundational models of openAI and ChatGPT provide the infrastructure for what’s to come, much like a computer chip enables software. Society is at an inflection point with technology, almost unlike anything Turner—who has worked in technology since the first personal Macintosh computer came out in 1984—has ever seen. Even the internet wasn’t as transformative as what we’ll see with AI, he said.

But, he warned, the pace at which AI is advancing is unfathomable. “Think about it: When you went to bed on Nov. 29, 2022, none of this was possible. But you woke up on Nov. 30, 2022, all of this was possible,” Turner said.

It’s been a mere six months since ChatGPT launched and it’s already in its third generation. In the next six months, Turner said, “our heads will be spinning.” As an example, Turner says a customer can upload an entire real estate contract and talk it through with AI. A site like heypi.com can answer any question a customer might have about that contract. AI can help the customer understand risk, pricing, logistics—just about any aspect. “It’s pretty accurate. It’s also patient and kind because it doesn’t care how long it takes for me to get this information.”

Though he uses and loves AI, Turner is sounding the alarm once again. The conversations around regulation and control of AI should have happened in 2016. The conversation must happen now, and action must be swift, he said. Data privacy is one of the biggest risk factors for the real estate industry when it comes to AI. Real estate data is highly valuable, unique, and largely private, which is why policy is needed right away.

Turner proposed a four-point framework in which to discuss what policy and regulation might look like for AI:

1. Ethics. This is where transparency, fairness and societal impact come into play. Algorithms play a role in decision-making, so they need to be transparent and easily understood. Algorithms can also lead to bias, which, in turn, can lead to the unfair treatment of certain groups of people, so ethical guidelines need to be clear. There’s fear around AI’s potential to replace humans in certain sectors of the workforce. Policy must take into account the use of AI in real estate and the “broader societal implications, like job displacement or housing impact,” Turner said.
2. Bias. Right now, AI is trained by human beings—and human beings have implicit bias. In order to ensure that algorithms aren’t inherently biased, proposed regulations need to “define strategies for identifying and mitigating bias risk,” Turner said. There is potential, however, for AI systems to promote a more fair and equitable housing market, which should be the goal. Therefore, the use of “diverse and representative data sets” is needed to ensure that AI is developed in a way that minimizes bias.
3. Privacy. Privacy issues range far and wide in the age of the internet, and it’s no different with AI. Data must be protected, but that can be done only if strong and clear “measures to prevent unauthorized access or misuse of sensitive personal information used in AI-driven real estate applications” are implemented, Turner said. This also means clients and professionals need a clear understanding of what data is being collected and why—and they must be able to consent to this collection of data. Additionally, data must be anonymized in order to “protect individuals’ privacy and comply with data collection regulations.”
4. Control. AI tools will soon—if not already—have the capacity to make decisions based on the information received. It is imperative that AI have some form of robust human oversight to provide a checks-and-balances process, Turner suggested. A decision needs to be made on what roles government entities and the real estate industry will play in the regulation of using AI in home sales transactions. Turner said guidelines and policies must be clear and easily understood to “prevent misuse or unintended consequences. This kind of regulation cannot be done in a vacuum, though. The public and stakeholders need to be informed and have a say.”