SW: Artificial intelligence means different things to different people. what do you think?
Johor: The core of artificial intelligence is nothing new. All AI is predictive analytics, even generative AI like chatGPT and chatbots. What it does is take the data and predict the outcome, or in the case of generative artificial intelligence, what you expect to hear in the answer. For those who are truly looking to improve efficiency in their business platforms, workflows, and more, machine learning is probably where artificial intelligence is most commonly used.
SW: Some parts of automation and artificial intelligence have been around for years, especially for a company like UWM that has all this data. If you don’t have the right data and know that the data is good, you’re essentially starting from less than zero.
Johor: You are right, you are starting from scratch. The good news is that we all start from scratch. Whether it’s mortgages and fintech or any industry, no single organization manages data so well year after year.
If we take the mortgage industry as an example, whenever we were busy, no one really cared about their data or developing a good long-term strategy. We were all just trying to make ends meet. No matter how you slice it, the fact is: the garbage comes in and gives you garbage, it’s just prettier and a little better packaged. So the key behind all this is to make sure you have a solid data structure and a solid data plan before you really start moving forward,
SW: How do you see the balance between humans and artificial intelligence?
Johor: In the 36, 37 years we’ve been in business, we’ve never fired anyone. This will definitely continue. No matter how we leverage artificial intelligence, we are so invested in our team members, their growth at UWM, and their career paths that we won’t [replace them with AI] – Absolutely not.
People should be thinking: How do I retain my current employees? How do I train them properly? How do I leverage artificial intelligence within my existing workflows, within my existing products, so that I don’t need to hire more people. I need to be able to work well with the team that I have, make them feel challenged, make them feel satisfied. But as more and more business starts coming in, I shouldn’t be hiring more people, I should be able to have technology delegate a lot of the tasks that I’m paying people for.
SW: How do you see the pace of change in artificial intelligence? Do you take a cautious approach, or do you want to be first and move fast?
Johor: I’m notorious for not using vendors—I keep everything in-house and rely on my 1,600 IT members to manage it. But the truth is, we do have to work with other companies sometimes. But I don’t want to be responsible for anyone who’s not part of UWM, so what we do in all the modeling we build, including generative artificial intelligence, is we build backends that do the following: plug and play.
We built it so that we could pull out one vendor and one partner and replace them with another vendor and partner if needed. We make it a bit of a headless experience, being able to say, hey, we have a product, we built this product, we own this product, we control this product, and it doesn’t need to be with someone. If a partner falls behind technologically, then we can quickly move on to someone else and give them the opportunity to provide the services that we need to provide to our broker team members to be able to continue to level the playing field and then outperform the competition from others. environment.
The question I get asked most is: Will artificial intelligence take over my job? The answer is very simple and very clear. No, artificial intelligence will not replace your job. People using artificial intelligence will take your job.
SW: You have a large team and you have open office hours where people can come in and like sell stuff. When it comes to artificial intelligence, do you see people getting really excited about some of the possibilities here?
Johor: Oh my god, they were so excited. There is an endless stream of people with different ideas and different solutions. But even better is that everyone wants to understand our strategy moving forward and why we choose this path. Our team members are staying with us longer and longer. They understand why we build things the way we do, and they can provide solutions because they’ve been working on building the technology or using it.
So they can come in and say, hey, now that I have a basic understanding of artificial intelligence and the direction and strategy we want to take, I think we should completely change everything and do this. This is where the best ideas come from. I can empower them to actually execute it.
SW: You have so many employees that you save 10 minutes, maybe even 5 minutes, that would be the equivalent of thousands of employees – that’s really where the ROI can add up quickly.
Johor: The true ROI is: Can we provide the same experience that brokers and borrowers expect from UWM? We all have our secrets to making our mortgage companies successful. It’s embracing whatever secret sauce you have and then figuring out how to never lose it, no matter what the business is. That’s where technology and solutions and things like that are really supposed to come into play – not just artificial intelligence, but generally speaking, it’s about ROI.
SW: What part of gen AI do you use?
Johor: Starting with gen AI, we want to provide information to our agents and internal team members in a conversational way that happens very quickly and in a way that’s different to them than the traditional way of getting answers by going to the library and trying to look it up. A book provides the answers. You can take these generated snippets a step further to start automating conversations.
And then where AI really goes to the next level is in providing questions based on how they behave within our systems. As they move around the screen, we can actually provide generative dialogue as they work. Being able to say, Hey, you know what, you might want to do this right now or you’re choosing this product, but by the way, there are four more products, and if you add another document, we can analyze this and secondly, tell you that The product may actually be a better fit for your borrower now.
This is what we call the fusion of traditional artificial intelligence and generative artificial intelligence to create an experience that truly enhances the work that people are doing and makes them smarter, faster and more efficient at work.
SW: What parts of using artificial intelligence worry you?
Johor: In finance, it boils down to the ethics of making financial decisions based on computer models. I think that’s really the biggest problem. You can worry about fraud, but…I think it’s more important not to reject potential borrowers based solely on computer models. This is where the next step and the next evolution of artificial intelligence in financial services begins. The technology is moving so fast and learning so fast that it’s hard to figure out where those guardrails are going to be because no one is really taking full advantage of artificial intelligence yet.
SW: At this point, I think one of the things we’re most hesitant about is decision-making. Lenders are using artificial intelligence to do all sorts of things to make transactions faster, among other things. . What do you think about this?
Johor: I would say that there are definitely people out there who are considering artificial intelligence for decision-making and would definitely be happy to be exposed to this! So I think it’s just a question of, what is your interest in technology, how quickly can you move, what is your understanding of the mortgage industry, and how does the technology fit into the process to be able to do that.
SW: We’ve heard regulators say that when it comes to artificial intelligence, all the same rules still apply. If your AI decides something, we’ll still hold you accountable for anything the technology does because you built the technology. One lawyer I spoke to said, “If a person makes a mistake, I can explain it to the regulators, but if you build a machine that makes a mistake, it’s harder to understand.”
JS: I almost want to say that’s not the case. If you look at decision-making, the most important part is the level of confidence you have in what the data is, what problem you want to solve, and how you solve it. So if you take the time to make sure that you have complete verifiable confidence in what you’re doing, in what the model looks like, and you’ve tested it thoroughly, and you have that comfort level – it’s almost like working with people Conversation is the same because you can show exactly the logic behind that modeling and why it does what it does.
And then you can actually tell it where to find all the information, how to extract it, how to calculate it—all of that. Everything creates a paper record: people and machines create paper records. Just make sure that when it comes to technology, you’ve created a proper paper trail to be able to show why or justify the decision to adjust it.
