Putting AI to Work in Insurance

Learn how to leverage AI to streamline operations

and improve customer experience
Putting AI to Work in Insurance 2

When we hear about the rise of Artificial Intelligence (AI), it’s easy for our minds to make
the leap to images of robots taking over the world. (Thanks, Hollywood!). Certainly, AI
doesn’t need to be that sophisticated­—or that threatening—to benefit the insurance industry.
In this ebook, we’ll look at AI applications that help carriers streamline operations and
improve the customer experience. Then, we’ll discuss how carriers can prepare for the coming
AI revolution in insurance.
Putting AI to Work in Insurance 3

Contents
WHAT IS AI?

Understanding Data Analytics, Machine Learning, and Artificial Intelligence . . . . . . . . . . . 4

3 TYPES OF AI

Examples of Cognitive, Autonomous, and Business AI Applications . . . . . . . . . . . . . . . . . . . 8

IS YOUR ORGANIZATION READY FOR AI?

Answers to Commonly Asked Questions from Insurance Leaders . . . . . . . . . . . . . . . . . . . . . 11

THE RIGHT ENVIRONMENT FOR TRANSFORMATIVE CHANGE

Innovation Framework Designed for Insurance Solutions . . . . . . . . . . . . . . . . . . . . . . . . . . . 14

Putting AI to Work in Insurance 4

What is AI?
AI is the popular new buzzword in tech. Software and hardware developers are
jumping on the bandwagon by claiming AI capabilities in their marketing materials.
But just how accurate are these claims?

As Dr. Michael Wu, a noted data scientist and AI expert, told us in a recent interview,
“[AI] is such a hot trend. Everybody wants to say they do AI. But a lot of people are
confused about what these terms mean, how they work, and what the actual business
implication is.”

“
[AI] is such a hot trend. Everybody wants
to say they do AI. But a lot of people are
confused about what these terms mean,
how they work, and what the actual
business implication is.”
– Dr. Michael Wu
Putting AI to Work in Insurance 5

So before we talk about how the insurance industry can leverage AI, let’s do a
quick level set on what exactly AI is. To do that, we need to look at two related
technologies: data analytics and machine learning.

DATA ANALYTICS
Data analytics is the systematic use of data to improve decision making. Analytics
used to be called business intelligence or BI. More recently, big data analytics has
become the term du jour, with “big” being the operative word.

In healthcare, sensors and other connected devices in the Healthcare IoT (Internet of
Things) are driving a dramatic increase in available data for decision making. At the
same time, increases in processing power and computational capabilities will allow us
to make the right decision faster.

No doubt, big data analytics is a good thing. Mountains of data and better
computational capabilities will allow healthcare providers to diagnose diseases earlier
and prescribe more effective treatments. Carriers will also be able to better assess
risks and create new insurance products targeting humanity’s most pressing needs.

However, while big data analytics will make us seem like smarter humans, the
technology simply combines sophisticated algorithms with huge amounts of data­­—a
powerful combination but not technically AI.

CONTEXT: What’s Big Data?

Big in Growth, Too.

9000
SENSORS & DEVICES

90% 50x
8000
Volume in Exabytes

7000

6000 of the world’s data growth from

was created in the last 2010 to 2020 SOCIAL MEDIA
5000
2 YEARS
4000 VOICE OVER IP

3000
ENTERPRISE DATA

2010 2020

1 EXABYTE EB = 1,000,000,000,000,000,000 BYTES B Source: PromptCloud

Putting AI to Work in Insurance 6

MACHINE LEARNING
Machine learning takes analytics a step further. To leverage analytics in decision
making, we start with a model. For example, we assume that certain lifestyle
variables have a measurable impact on mortality rates. Our model could
conceivably weigh hundreds of factors to determine the level of risk involved with Artificial
issuing an individual policy.
intelligence,
With analytics alone, the model is static. Machine learning uses gathered data to
analyze and adjust the assumptions built into the model. It might be as simple as according
adjusting the weight assigned to different factors. Or it could even discover links
to experts,
between seemingly unrelated variables to develop a new model.

As sophisticated as this analysis and the ability to adjust the model is, machine
mimics human
learning remains a necessary component of AI, but many experts don’t behavior. The
consider it to be true AI.
key here is
ARTIFICIAL INTELLIGENCE
that while
According to experts, artificial intelligence mimics human behavior. The key
here is that while data analytics and machine learning inform our behaviors, AI data analytics
takes action. It does what we would do when given the same data and ability to
analyze it (without the emotional component of human decision-making, of course). and machine
By doing what we would do on our behalf, AI can help us accomplish more.
learning inform
our behaviors,
AI takes action.
Putting AI to Work in Insurance 7

Will AI Replace Claims

Managers?
Employees are sometimes resistant to AI because they fear the
technology will replace them. To ensure adoption and a positive
workplace environment, it’s important to help them understand
how the technology will improve their lives and make them
better at what they do.

Claims processing is a great example. According to the U.S. Bureau of Labor Statistics, an
estimated 400,000 US-based insurance employees are expected to retire in 2020. Many of
these workers will come from claims departments, and many insurance carriers are rightly
concerned about how to replace them.

This shortage of talent will add to the burden of existing claims agents. In a recent survey
that RGA and SALT Associates conducted on workplace culture, respondents cited heavy
workloads and inefficiencies as two of the greatest negative influences. As one manager
said, “I think most of my employees are pretty burned out.”

By helping claims managers understand how AI and related technologies will help them
get more done efficiently and effectively, you can alleviate some of their fears of being
replaced by machines.
Putting AI to Work in Insurance 8

3 Types of AI
As futuristic as it sounds, there are numerous examples of AI being used in business
today. To understand how AI can be applied to the insurance business, it’s helpful to
look at the most common types of AI being developed and deployed today. In our
interview, Dr. Wu separated today’s AI applications into three categories:

#1 COGNITIVE AI
This category of AI is probably the one most familiar to us as it includes chatbots and
digital assistants such as Siri, Google Assistant, and Alexa. These applications all mimic
human verbal responses, sometimes so well it can be frustrating when they don’t
understand us.

Common chatbots found on websites are another example of Cognitive AI. While
many of these chat sites are still manned by humans, increasingly more start as
"conversations" with bots. These bots gather initial information, attempt remedial
troubleshooting, and then route us to the appropriate self-help guide or human for
more assistance if necessary.

The Four Laws of Chatbots

Despite Hollywood’s best efforts to make us fear robots, people are increasingly willing to talk to chatbots. In a
recent study, only 18% of US respondents said they always prefer interacting with humans. Four elements appear key
to implementing chatbots successfully:

While chatbots are getting better at mimicking humans,

people still want to know when they’re talking to a machine.

They also want their problem solved as quickly as it can be

solved by a human—if not faster.

They want an accurate response to their question, not just a

fast response that does nothing to solve their problem.

If they can’t get what they want from the bot, they want the
option of talking to a human.
Putting AI to Work in Insurance 9

#2 AUTONOMOUS AI
Not all AI needs to be programmed with a human personality. Sometimes, all we need is
for our intelligent machines to mimic human movement. Self-driving cars are probably
the best examples, but this category includes other autonomous machines as well.

For example, Amazon is leading the way in using robots in its fulfillment centers.
Drone-based deliveries are another type of autonomous AI that is now in the
experimental stages, but they’ll be commonplace before we know it.

Autonomous AI doesn’t seem like it has much to offer the insurance industry, but one
never knows what the future holds for this category.

#3 BUSINESS AI
This category is sometimes called decision AI because its primary purpose is to
automate the decision-making process and response.

Business AI often gets confused with other related terms such as predictive and
prescriptive analytics. In predictive analytics, a program analyzes a set of historical data
and then extrapolates what’s most likely to happen in the future. Prescriptive analytics
takes that a step further and provides guidance on actions that should be taken.

• Predictive analytics analyzes a set of historical data and then extrapolates what’s
most likely to happen in the future.
• Prescriptive analytics provides guidance on actions that should be taken.
Predictive and prescriptive analytics are widely used in claims and policy application
analyses. From a purely analytical perspective, these programs undoubtedly are better
at making decisions than a human alone. However, to be considered true AI, the entire
process, including the final decision, needs to be handed over to the program.

Even though understanding the three categories of AI is helpful when assessing

insurtech capabilities, AI may not be what you need. For example, if you’re just looking
to streamline operations and help employees make better decisions, a sophisticated
analytics application may be just the ticket. Still, the future of digital transformation
in the insurance industry will undoubtedly include a heavy dose of AI, so organizations
must be prepared.
Putting AI to Work in Insurance 10

Business AI in Action
Researchers have discovered a link between mortality rates, motor vehicle records, and
credit scores. Currently, some carriers are looking at how this data can be used to streamline
the policy application process, in some cases even allowing to bypass certain doctor’s visits.

In one way, this discovery is the perfect example of data analytics. Researchers used available
data to look at the relationship between what would normally be seen as unrelated variables.
And it goes further than that. By replacing the prerequisite exam (performed by a human
doctor) with a decision made by the model, a carrier would effectively be leveraging AI.

MOTOR VEHICLE RECORDS CREDITBASED SCORING

250% 12YEAR STUDY OF U.S. POPULATION
MORTALITY SEGMENTED BY A CREDITBASED SCORE
Relative Mortality

300%
Actual-to-Expected Mortality

200%
250%
150%
200%
100%
150%
50%
100%
0%
1 2 3 4 5+ 50%
Clean Minor Major Violations
0%
5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 100

Number of Violations TrueRisk® Life Credit-Based Insurance Score

The future of digital transformation in the insurance

industry will include a heavy dose of AI, so
organizations must be prepared.
Putting AI to Work in Insurance 11

Is Your Organization
Ready for AI?
As we help carriers innovate, we often get asked questions related to skill sets and
organizational competencies. Insurance leaders often doubt if their traditional
businesses are prepared to take advantage of all the powerful benefits of AI. Here are
answers to some of the most common questions we hear:

DO WE NEED TO HIRE DATA SCIENTISTS?

Data science is behind all AI, as well as other related technologies such as data
analytics and machine learning. Undoubtedly, an understanding of data science is
helpful. However, that doesn’t mean you need to spring for a team of in-house data
scientists to take advantage of the technology.

Think about it this way. In the early days of computing, companies had to hire computer
scientists to operate these machines. Only the largest, most well-heeled companies
could afford it. But as computing power became more affordable, packaged software
became more profitable to develop. Eventually, these systems became easier to use. 93% of survey
Today, most white-collar workers (and a large percentage of blue collar workers) use
computer systems, and only a small minority are programmers. respondents
We’re past the early days of AI and on the cusp of a similar revolution to software’s. claim their
Some organizations may find they need to hire data scientists because they can’t find workplace will
the capabilities they’re looking for in a packaged application. Or they may outsource
the custom programming to a third-party developer. Many organizations are already be improved
finding at least the foundation of what they need in ready-made commercial
applications that can be tailored to their business.
through
greater use of
DO WE NEED TO HIRE TECHNICAL USERS?
For AI to provide the sought-after benefits, as with any technology, it must be used.
technology.
Carriers then have to ensure employees are at least open to using new technology
and that they have what they need to be effective.
Only 18% are
One of the most important elements is adequate training. A 2019 study found that 73%
of US workers believe AI will eliminate more jobs than it creates. Yet only 18% are “extremely
"extremely confident" they can secure the training they need for digitalization.i
confident”
Carriers also need to be careful when making assumptions about their
current workforce. In a survey of 7,000 employees across 15 countries, 93%
they can
of respondents claim their workplace will be improved through greater use of secure the
technology. Many workers want to use technology.ii What’s seen as resistance
may just stem from a lack of training, not understanding the technology’s purpose, or training they
even a poorly designed application.
need for
digitalization.
Putting AI to Work in Insurance 12

WHAT’S THE ONE THING WE MUST DO TO ENSURE SUCCESS?

The biggest concern carriers have is often their employees’ technology-readiness,
especially customer-facing employees. But this concern places the focus on the wrong
end of the hierarchy. No matter how well thought out the technology and roll-out plan
are, and no matter how willing the employees are to use the technology, all technology
implementations encounter obstacles. At these times, senior leadership must step in
and restate their commitment to the project’s success despite these hurdles.

Getting buy-in from executive leadership early in the process is essential. One way to
accomplish this is to involve them in the problem definition phase. For example, RGAX
conducts problem-framing workshops with clients by bringing in participants from
across the organization to identify the challenges the organization is facing, define the
problems to solve, and discuss potential approaches — including using AI.

What makes a problem-framing workshop different than the in-house brainstorming

experience is that we train our facilitators who lead these sessions. They know how to
keep everyone engaged, help participants move on from unproductive discussions, and
enforce the “rules of the game,” all leading to a useful outcome. When executives are
involved in these workshops, they understand the reasons behind the initiative and its
positive benefits to the organization.
Putting AI to Work in Insurance 13

6 Ways to
Promote Insurtech Adoption
# 1
Let everyone in on the plan. Helping employees

2
understand the reasons behind the technology’s
implementation, project’s goal, and the roll-out process #
will help promote buy-in to new ways of doing things. Don’t rush it.
Implementing
insurtech can

# 3 dramatically
upend long-held
Show them how. People processes and

# 4 are tech-savvy to varying

degrees, so you may need
perceptions about
the right way to do
Don’t make assumptions. to offer different levels of things. You need
Yes, you need to give training to bring people up to go at a pace
employees time and provide to speed. Refresher courses that gives people
adequate training, but are also helpful in ensuring time to adjust.
don’t assume people don’t the adoption of new
technologies doesn’t slip.
want to or can’t adjust
to new technology. In a
survey of 7,000 employees
across 15 countries, 93%
of respondents think their # 5
workplace will improve with Make sure the insurtech fits the business. Insurtech
greater use of technology.iii has to support the defined business processes and
help employees be more productive. If they see using
technology as an extra step in the process, they’re
likely to revert to manual ways of doing things.

# 6
Show what’s in it for them. Technology has the potential to create new career
paths. It can also make people more valuable in their current roles. Moreover, it
can provide day-to-day opportunities for recognition, e.g., insurtech “champions.”
All these create a more positive culture and promote technology adoption.
Putting AI to Work in Insurance 14

The Right Environment for

Transformative Change
AI and related technology offer exciting possibilities most of that time, they’ve worked well. Even automated
for insurance carriers looking to improve the customer processes are often electronic reflections of firmly
experience and streamline operations. Unfortunately, ingrained ways of thinking.
many insurance carriers just aren’t set up for innovation.
If innovation isn’t in your corporation’s DNA and culture,
Two primary obstacles get in the way:
starting an innovation team or even an initiative can be
An ingrained aversion to risk. The insurance industry is an uphill battle. RGAX offers innovation-as-a-service
all about risk assessment and avoidance, so many of the solutions designed specifically for insurance carriers. From
top companies and leaders in our industry are risk averse. collaboration, to co-development, to accelerated product
development, we can dramatically decrease the time
Established thinking. Our industry has been around for
and minimize the risk many carriers face from problem
centuries. Some of our current processes and patterns
statement to product commercialization.
of thinking go back decades at the very least. And for

i Gallup/Northeastern University, Optimism and Anxiety: Views on the Impact of Artificial Intelligence and Higher Education's Response
ii Arube HPE, The Right Technologies Unlock the Potential of the Digital Workplace
iii Arube HPE, The Right Technologies Unlock the Potential of the Digital Workplace

IF YOU’D LIKE TO LEARN MORE ABOUT HARNESSING THE POWER

OF DATA, CONNECT WITH US.

Let’s Connect