How Intelligent Automation Is Changing the Nature of Work in Insurance, Collision Repair, and Beyond

When mobile phones were first made available to consumers more than thirty years ago, it was difficult to imagine that they would become the hubs of Internet access, information and activity—even health tracking—that we know them to be today. Similarly, it may be difficult to imagine that smart phones, and indeed the world around us, will soon be embedded with artificial intelligence. But that appears to be the direction things are headed. This has significant implications for companies of all types, including those in the Property and Casualty (P&C) and collision repair industries, especially when AI is part of overall automation of business processes. As artificial intelligence and other advanced automation technologies become deeply embedded in business processes, they appear to be coalescing around two key vectors: automation is freeing knowledge workers from routine tasks, enabling them to focus on higher value ones, and it is augmenting human intelligence, enabling better-informed decision making.

Robotic Process Automation: Increasing Productivity

Robotic process automation (RPA) is simply the automation of repetitive tasks that humans typically do, including things that happen all the time in the Property and Casualty and collision repair industries: filling out forms, taking notes, and transferring information from one system to another. The general idea is that if a person does a two-minute task thirty times a day, automating that task saves an hour of work a day, adding up to weeks over the course of a year.

The potential savings companies will experience from RPA will be between $5 and $7 trillion worldwide

Business-information portal Statista predicts the global RPA market will reach almost $5 billion by 2020, but the savings from RPA will likely be much higher. Researchers at open-source, distributed data processing framework Hadoop estimate that by 2025 the potential savings companies will experience from RPA will be between $5 and $7 trillion worldwide. If the P&C and collision repair industries see even a fraction of this cost reduction, they will have much to gain—and save—from implementing RPA. Of potentially greater benefit to our industries: when RPA handles repetitive tasks such as data entry, people will have time to address more complex tasks that require uniquely human strategic thinking and emotional intelligence.

Intelligent Automation: Changing the Way We Interact

Things get even more interesting when artificial intelligence is added to the automation mix. Take Google Duplex, a feature of Google Assistant that calls restaurants and make reservations and is coming to Pixel smartphones in a few test markets toward the end of this year. The artificial intelligence-powered phone bot is virtually impossible to distinguish from a human, if not for the fact that it self-identifies as an automated service. You really must hear it for yourself, if you have not done so already.

Imagine a world in which everyone has an artificially intelligent personal assistant.

The implications for businesses are profound and will affect both front-end customer experience as well as back-end office processes. Imagine a world in which everyone has an artificially intelligent personal assistant. The already complex conversation around where to use chatbots in claims process would be infinitely more difficult. Not only would companies have to determine whether a customer engaging with its business is willing to interact with a chatbot, but it might also have to determine if that customer’s chatbot wants to interact with another chatbot or a human. In the future, businesses may not market to customers, but to its customers’ chatbots.

Intelligent Automation: Personal Assistants in the Workplace

Now imagine the productivity gains your company would see if every person in your workplace could have their own similar, artificially intelligent personal assistant. At least one company, NICE, is bringing that vision to life. It has developed an artificially intelligent personal assistant that can observe and interpret desktop activities, open applications, and run relevant processes in the background while the human user interacts with a customer or handles other high-value tasks. One of the ways we are thinking about this here at Mitchell is through the lens of workflows that benefit from deep human interaction. Take nurse case management, for example. Our nurse case managers tell us that one of the most important things they can do to positively influence workers’ compensation claims outcomes is to build strong relationships with injured workers. Yet, they spend a good deal of their time on repetitive tasks that do not contribute to these essential relationships.

If knowledge workers in the claims ecosystem could automate repetitive tasks, they would enjoy a significant increase in productivity.

If nurse case managers and other knowledge workers in the claims ecosystem could automate repetitive tasks by using a combination of simple and intelligent RPA, they would enjoy a significant increase in productivity. Better yet, they could use that time on tasks that require a human touch: making personal connections, providing consultations, being empathetic—things that drive different behaviors and actions and can truly help improve outcomes and restore injured people’s lives.

Artificial Intelligence: Augmenting Human Intelligence

Duplex not only underscores the power and growing accessibility of intelligent automation, it also signals a step toward embedded intelligence as a part of everyday life, both for consumers and in workplaces. Indeed, there is technology on the horizon that will help us do more than automate repetitive tasks—it may help us make complex decisions. IBM has developed an artificial intelligence project called Project Debater that serves as an example of how artificial intelligence can augment human intelligence. Project Debater, as its name suggests, is trained in the art of debate. When assigned a topic, Project Debater scans a database of millions of articles from established publications and identifies possible arguments to make. It listens to its opponents, identifies key points in their arguments, and delivers a rebuttal. Unlike Watson, IBM’s more established AI project, Project Debater goes beyond parsing facts. It has a point of view and uses sophisticated natural language to persuade.

Project Debater goes beyond parsing facts. It has a point of view and uses sophisticated natural language to persuade.

While Project Debater appears to be a long way from commercialization, if that is indeed IBM’s intent, I see two significant implications for technology like this in the workplace. First, it can research vastly more information than a person can, and much faster, so it can reach informed conclusions faster. Second, it could serve as the ultimate fact-based sounding board without the bias that’s inherent in people—that is, of course, if we can rid ourselves of algorithmic bias in programming.

Algorithmic Bias: The Caveat

Algorithmic bias is the notion that because artificial intelligence is typically trained using datasets developed by humans, it is susceptible to biases based on those datasets. Amazon, by all accounts a leader in automation on all fronts, recently made the news for an internal test project that used artificial intelligence to evaluate job candidates. Trained using historical data that reflected the predominately male tech industry, the program began to penalize female applicants. Even after making some refinements to address the bias, Amazon scrapped the project on the basis that there could other harmful biases lurking.

Understanding algorithmic bias is going to play an important role in the work of any company using algorithms in decision-making.

As the world becomes increasingly automated and dependent on algorithms, understanding bias, managing it, and preventing it from contributing to undesirable or even dangerous decisions is going to play an important role in the work of any company using algorithms in decision-making.

Artificial Intelligence Today: Smarter Solutions

Today there are already many practical ways technology in general, and artificial intelligence in particular, can support humans in decision-making. At Mitchell, embedding “smarts”—building access to decision-making information and resources directly in our solutions—has long been a focus across the company. This belief is foundational to our commitment to develop solutions for the auto physical damage ecosystem that support the safe and proper repair of increasingly complex vehicles. One example, Mitchell WorkCenter Assisted Review uses computer vision and machine learning to validate repair-versus-replace decisions.

Intelligent Automation: Driving Better Outcomes

Similar to the adoption of mobile technology, as intelligent automation becomes more deeply embedded in consumer experiences, it will also become an integral part of how businesses operate. As artificial intelligence becomes increasingly ubiquitous, it will allow knowledge workers to spend less time on repetitive tasks and more on the higher value, human ones that drive better outcomes and help restore lives.