DSNews delivers stories, ideas, links, companies, people, events, and videos impacting the mortgage default servicing industry.
Issue link: http://digital.dsnews.com/i/1511756
MortgagePoint » Your Trusted Source for Mortgage Banking and Servicing News 64 November 2023 T H E P O I N T automated mode. As you already pointed out, from a compliance perspective, the standards are still the standards. Just because things are automated does not mean the compliance standards have changed, and so doing rigorous testing of the rollout of new processes and new methodologies is super critical as we roll out new functionality. Q: AI has been a topic that people have been talking about forever, what have you seen so far, and what are you working with? Is AI still in its early days? Are people getting overexcited about something that is not quite there yet? The way I would frame it is AI is a technological capability. We think of it as, what are the use cases where this technol- ogy can be truly implemented? One thing that we use is called "computer vision," and "computer vision" is the new name for OCR technology. Computer vision is basically AI's ability to read a document. Not to go too far into the weeds, but OCR is Optical Character Recognition. That technology used to work a certain way. Computer vision is a new capability that allows bigger mortgage loan files and more documents. How do you go through a 500- to 700-page loan file to find what you are looking for? Computer vision allows any entity—an originator or servicer—to index documents and extract information from documents. The interesting part is the accuracy of that is now much higher. Five to seven years ago, this technology was still in its infancy. In our industry, due to compliance require- ments, we need things to be at a certain level of accuracy before we can accept them. I would say that computer vision technology is at that level of accuracy. These are the same capabilities that have been created by the big tech com- panies. Amazon has one model, Google has its model, Microsoft has a model, and then people have built models on top of their basic core technology. I see people experimenting with virtual chat boxes, and again, because of compliance standards in our industry, I think that needs a little bit more work. The last thing that we are working on is the next best action capability, whether that's consumer-facing or our internal em- ployee-facing system. Working on the next best action capability and building that capability within our toolsets is the other area, where I think we are in the third in- ning of that game … it's coming along, but there's a lot more that can be done. Q: One of the things that you hear anecdotally in a lot of the coverage of AI, in general, is concern about it being used to try to gauge credit scores, as there is inherent bias sneaking in based upon what data has not been fed and what assumption the AI is making. Where is that part of the conversation unfolding right now? You are right that in AI and the way some of these AI models work … how does one test AI models for unconscious bias or bias that the model may not be looking to produce, but it may produce bias in the end? I would suggest that other aspects of consumer finance, things such as credit cards and other consumer finance prod- ucts, are a little bit further ahead of us in terms of evaluating credit decisioning. Much of the credit decisioning in our industry is done by some of the large inves- tors, the GSEs, etc. They are the ones who are making these broader credit-decision- ing guidelines, and then we as an industry follow through on those guidelines. Q: What are some of the frontier technologies that you are looking forward to that are maybe not ready for prime time? Where do you see the industry heading on a technological front? I feel our industry has not developed a good use for mobile technology as of yet. Mobile technology is not a new tech- nology, and it is not a frontier technology, but we have not found that killer app that has the borrower saying, "Yes, I like it." I also feel like blockchain is a technol- ogy that just has not broken through as of yet. I think if there were a public block- chain that was, if I may even call it almost open source, everybody could write on, that would be interesting. If you think of what merged it to us as an industry, blockchain could fit in that ilk of things where it is owned by the entire industry or is almost a public good that everybody uses. I know a couple of people have tried to implement that and execute on that, but I do not think they have managed to thread the whole needle all the way through. At some point, somebody's going to figure it out, but we haven't figured that out as an industry just as yet. "I feel our industry has not developed a good use for mobile technology as of yet. Mobile technology is not a new technology, and it is not a frontier technology, but we have not found that killer app that has the borrower saying, 'Yes, I like it.'"