All of us remember the introduction of computers in the title industry which was heralded as a time saving method of generating the necessary HUD1 settlement sheets and ancillary documents necessary for us to perform our daily profession. Since then, computers have infiltrated many aspects of our personal lives from banking to Netflix predicting for us what movies we would like to watch next. Few of us could have predicted the astonishing claim of States Title that computers using algorithms can now replace the “…. the time and labor-intensive title search process with a predictive analytics algorithm that underwrites title insurance policies instantaneously, with curative-free coverage for nearly all title defects.” Putting aside for the moment that I do not understand what that even means, who is the company making this claim?
States Title, established in 2016, recently acquired North American Title and predicts that the future of title underwriting and issuance of title policies will involve the artificial intelligence of computers in predicting the probability of liens on individual titles by statistical analysis rather than imperial data subject to human review. States Title, whose website at www.statestitle.com, lists the leadership and management personnel for their company and not unsurprisingly not a single person in the leadership matrix has a background in real estate or real estate law and base their expertise on technology and information management.
Title underwriting, as a pre-condition to the issuance of title insurance is, after all, grounded in risk management. There is an old saying in real estate that no title is perfect, and we know from experience it is true. The title insurance industry and specifically national title insurance companies accept the known risk of claims inherent in an imperfect land recordation system to protect their insureds against loss and partly to justify rate filings. So, if issuing title policies is founded upon assumption of title loss risks, what is wrong with a computer algorithm predicting the likelihood for such a loss rather than human review? The answer I think lay in how States Title perceives the process would work.
This diagram taken from the website of States Title illustrates the concept:
Raw Data Computer algorithm 5-10 second analysis Title Product
After “…pulling data from public, paid and partner sources and intelligently sorting through them….” The algorithm can statistically predict to a 99% probability of a “clean title” and a 1% probability of liens as a result of a 5-10 second computer generated analysis. Essentially what the proponents of this system advocate is that the title insurance issuer is simply playing Russian Roulette on each title and assuming a 1% percent risk on all issued policies that the issuing company has assumed a risk of unknown liability. For the internal personnel of all major underwriters who are responsible for setting loss reserves, the task of setting reserves on that basis not only is impossible but foolhardy.
So where does that leave us? Suffice to say that StatesTitle has given us little in the way of detailed information enough to make a truly informed decision as to whether such a system is the prudent next step for the title industry. As for me, my analysis is simple at this point and I can illustrate it in this way. If I were given the choice of visiting a doctor for a checkup or the choice of a computer generated algorithym stating that I was in all probability 99% per cent healthy….I will still probably go to the doctor.