If you like what you read below, we invite you to subscribe to the Insurance Research Letter here. Each issue has topical articles written by industry experts that you won’t read anywhere else. In addition, you will stay current with major developments around the world with our Global Briefs; a monthly Synopsis of insurer and broker news; and our acclaimed Back Page with all sorts of non-insurance fun and interesting subjects.

Robotic Automation – How Many Humans Can We Realistically Replace?
By Marci De Vries-Todtz

I am writing this article from InsureTech in Las Vegas, where the newest and brightest technology companies launch their solutions. The top question posed this year is: “Are humans replaceable in insurance.”

To answer that question, the industry needs to examine the scope and course of most of their employees’ responsibilities. To make it easier, I’ve divided work into two categories, “Monkey work” and “Thought work” Monkey work is defined as any part of a job where, given enough training, a monkey could perform the task. (No offense to monkeys intended, of course). Thought work involves analyzation of inputs, and decision making. Thought jobs benefit from experience, education and expertise, while Monkey jobs do not.

Monkey jobs include data entry, opening software, printing and emailing reports, combining data from one database with data from another database, sending vendor assignments, and so on. These jobs have been and continue to be automated using technology with huge success, allowing the jobs to be performed faster, more consistently, and 24/7. Moreover, Monkey jobs are often cited as a major factor in employee turnover and dissatisfaction.

The biggest examples of successful Monkey work replacement include Facebook and Google, where the draconian tasks of looking up information at the library (Google) and calling/writing letters to your friends (Facebook) was replaced with technology with well-documented benefits to users.

Technologists are now trying to tackle Thought jobs with technology, wherein artificial intelligence calculates inputs, “learns” how to do jobs and then performs decision making with only minimal human input, with the goal of replacing the human input. While I don’t doubt that this is possible eventually, I do have my concerns today. The current methods by which data is recorded and processed in insurance is inconsistent and could lead to wrong conclusions when automated.

The path to full integration is first to automate all of the Monkey work, thereby guaranteeing a consistent data set. When data is consistent, a machine learning tool can be much more effective and perhaps someday achieve the goal of replacing thinking humans.

So how close are we to automating enough for machine learning? Based on my observations, it looks like there is too much “Monkey Business” (Get it? It’s funny) still happening, where lower level employees are still entering and manipulating data the same way over and over, like a typing pool but with computer terminals. In order to reach the goal, more intentional automation efforts need to be explored. The good news is that tools exist today to fully automate ALL of these repetitive processes without creating novel technology. It’s just a matter of who putting all of these technologies together in the right way.

Data is power, and the ability to provide insights across millions of files is a direct path to greater profitability for the Insurance industry. To reach this goal, we need to take steps today to make sure the insights are based on correct, consistent data today.