AI in Mortgagetech

December 13, 2020
Antonio Flores

The mathematics of what we know as artificial intelligence have been around for several decades, and have not radically changed. However, AI in recent years has developed revolutionary potential in many areas due to Moore’s Law. The continual increase of processing power has put once computationally prohibitive sorting and generation algorithms in reach for even the average user. (For reference, the average smartphone of today contains more processing power than the space shuttle.) Many of these algorithms and networks have potential uses in the mortgagetech sector.

One major use of AI in the mortgage industry is in the replacement of optical character registration (OCR) for transcription. OCR, which was developed in the 1920s and 30s, only has a transcription accuracy rate of 85% or so per character, and does not allow computers to interact with the transcription afterwards. AI improves upon both of these: the company AI Foundry has used computer vision in their new product, Agile Mortgage, to reach accuracy rates of 90% and allow users to interact with the text on the computer afterwards. The improvement of transcription practices allows for greater automation of various practices and makes it far easier for these tasks to be done remotely. 

Another major area in which artificial intelligence is widely used is in the field of data processing. Thanks to various sorting algorithms, it is easier than ever to compare and generate data, which includes data on mortgages. This greatly decreases the cost and time needed to process mortgage documents at many steps of the process. One major problem with this right now, however, is the fact that in many places in the mortgage industry, data format is not standardized. The algorithms cannot sort through them because they are not properly compiled. It is for this reason that the US’s Mortgage Industry’s Standards Organization received several million dollars last year from the Mortgage Banker’s Association in order to promote across-the-board data format standardization in the mortgage industry. Different algorithms allow us to prepare data more efficiently and precisely than ever, but there are still quite a few obstacles in the way of this particular application of AI that will need to be cleared.

To conclude, there are quite a few opportunities to improve the mortage industry using artificial intelligence. Various repetitive tasks that would once demand manual sorting can now be done quicker and more accurately, leading to greater value produced per employee.

Leave a comment