We live in a world where false claims are being tossed around like a basketball. I had a recent and very annoying experience like this from my cell service provider. I’d purchased a new cell phone and mobile hotspot. Because I was continuously running over my data limit, I purchased my provider’s unlimited data plan too. We needed to download a tremendous amount of data to do last-minute updates for a demo of our software at a conference. Smack in the middle of the download; the data speed became labored and very sluggish. I was stunned to find out during my service call to the cell provider that my “Go unlimited” plan was not unlimited. They offered other choices – “Beyond Unlimited” (for better for music and video) and “Above Unlimited” (for better performance and global access). Who knew that unlimited didn’t actually mean unlimited!
It has been my experience that we see this false advertising everywhere. Tech companies are self-branding their services as innovative or disruptive when they’re not. Real estate agents call themselves “area experts” when they just started working in a geographic area for the first time. So, when Brad Inman asked his audience on Facebook “Which company is leading the way with real estate AI?”, I got a little fired up. Although it’s an interesting question, there is a lot of marketing hype around AI and no obvious way to substantiate claims. In my opinion, a better question might be “How do we know which real estate tech company has real AI?” Much of the information available on the internet is confusing at best, so I turned to my CTO, Tavi Truman for some answers. Here’s my spin on what I learned from him.
As most of us know, machine learning is a type of AI. There are many types of AI and many kinds of machine learning. There are a few factors that are common amongst all AI and machine learning that differentiate the “real deal” from the imposters.
First, there must be a veridic source of data. Where this can go wrong would be in claims like the following… A company says they can target just the right buyer for a seller’s home by analyzing what buyers click on which houses and which home features. Since many individuals that search homes are doing this fun or for friends and family, there’s no way to say this could result valid data. Still, many companies derive their data from equally ambiguous sources.
Next, the data should be structured data. There must be a way to capture and organize the data so that it is effectively used for the type of ML or AI being used. Depending on how the data is stored, there can be limits to the usability of that data.
Third, there must be an algorithm that has been blessed by AAAI, which is an international non-profit standards organization for everything related to understanding AI. AAAI publishes standard open source algorithms, and these are also taught in University programs. These standards are not big secrets, so I’m curious to learn why companies claiming to have artificial intelligence are not shouting at the rooftops which of these methods are used by their company.
Once a company has valid data, structure and proper algorithms, they have what’s known as “weak AI.” For the AI to be elevated to “strong AI,” a fourth component of being able to associate real-world knowledge [representation and reasoning] to that data is required. This association means that the software understands what the data is. In other words, it knows what a house, or a cat, or love is. With this understanding, it can draw conclusions [reasoning] that resemble human thought but are exponentially more expressive and agile.
I’d love to learn what companies in real estate tech are using a correct standards for development of AI versus the seemingly common practice of AF (Alternative Facts). If you’re the real deal, I’d love to connect and see how we can help each other.