Where RPA and artificial intelligence diverge matters for insurers

RPA in insurance has made its way into The New York Times, with an interesting article about how State Auto Insurance Companies has eliminated 25,000 hours of manual human work from their back office using screen-scraping process automation technology. This is an excellent outcome with clear value to the organization. As noted in the article, it’s not about replacing humans (it’s likely to mean fewer humans hired but not humans losing current jobs), but freeing humans from mundane tasks and letting them focus on higher-value work.

This Isn’t Actually AI
In the article, Thomas Davenport (author of Only Humans Need Apply and previous keynote speaker at Novarica’s annual Research Council meeting) says, “This is the least intelligent form of AI” in regards to the State Auto use of RPA. I agree and I’m willing to take that a step further: this is the least intelligent form of AI because it is not AI at all. This is not an issue of semantics. RPA takes technology for emulating human/machine interaction that has been around for a long time and pairs it with much more configurable and adaptable algorithms. While it is also possible to pair RPA technology with machine learning technology (to help train and improve the manual emulation algorithms, for example), it’s important to consider them separately. Vendors often conflate the two because AI brings as much marketing value as it does actual value.

di-stock-server-070618-1.jpg
Servers stand in a computer room at the Yahoo! Inc. Lockport Data Center in Lockport, New York, U.S., on Friday, Sept. 26, 2014. Yahoo Inc., a $40 billion Web portal, is expected to release third quarter earnings on Oct. 21. Photographer: Andrew Harrer/Bloomberg

There Are Real Savings to Be Had
That being said, the State Auto case study shows that there are significant and measurable benefits to be had through RPA, whether or not one considers it artificial intelligence. The insurance industry has decades and decades of manual processes due to the slow modernization of legacy systems, the organic innovations taking place at a different pace within different groups, and the disparate technology stacks from M&A. The result is manual back office work that does not bring value to the organization but, rather, is compensating for how IT has evolved.

The issue isn’t that RPA isn’t true AI, the issue is that the industry should be investing in any technology that helps automate and fix broken processes even if it lacks the hype and excitement of AI.

RPA Paves the Way for Comfort with Real AI
AI is coming and it will have real impact on the insurance industry. Novarica may push back against how RPA is defined within the emerging technology space, but it does fit within the broader category of tech that will help insurers automate and augment human activities. Many AI applications will do the same, though at the level of insight and decision making rather than screen scraping. So RPA’s classification might be a misnomer, but it prepares the industry for more changes to come, and that’s a good thing!

This article has been reprinted with permission from Novarica.

For reprint and licensing requests for this article, click here.