Artificial
intelligence (AI) has the unique ability to process mass
amounts of
structured and unstructured data—a task that is unparalleled
by the
human brain. Since over 90% of the world’s information was
created only
recently from the volume of information stemming from
intelligent,
enhanced computer systems, various industries are adopting AI
as a
means to tap into this data source and gain a competitive
advantage.
Other Industries that Use Artificial
Intelligence
Hedge
funds rely on artificially intelligent, advanced machine
learning
algorithms to predict trends, produce client-specific
responses,
suggest future plans, and display reports in visually
appealing
layouts. For example, the leading hedge fund manager,
Bridgewater
Associates, invested in an AI unit to make predictions based
on
historical data and probabilities. The Healthcare
industry also
implements AI to determine treatments and foresee sickness
before a
doctor can. Studies indicate that in terms of diagnosing lung
cancer,
AI’s success rate is an astounding 90% in comparison to the
50% of
human doctors.
After
much
success in the financial and healthcare sectors, the insurance
industry
is considering AI adoption as a means to estimate risk
associated with
individual clients and enhance
claims
processing. Now, insurers are subject to federal and state tax
credits
made available to stimulate efforts in AI research,
development, and
adoption.
The Research & Development Tax
Credit
Enacted
in 1981, the federal Research and Development (R&D) Tax
Credit
allows a credit of up to 13 percent of eligible spending for
new and
improved products and processes. Qualified research must meet
the
following four criteria:
New or improved products,
processes, or
software
Technological in nature
Elimination of uncertainty
Process of experimentation
Eligible
costs include employee wages, cost of supplies, cost of
testing,
contract research expenses, and costs associated with
developing a
patent. On December 18, 2015 President Obama signed the bill
making the
R&D Tax Credit permanent. Beginning in 2016, the
R&D
credit can be used to offset Alternative Minimum Tax and
startup
businesses can utilize the credit against $250,000 per year in
payroll
taxes.
Development of AI in the Insurance
Industry
One of
the first questions is whether AI would be beneficial or
detrimental to
the insurance industry. Insurance companies are convinced that
AI will
aid them in enhancing their data-processing capabilities,
determining
risk, and predicting the future of claims and complaints.
These tasks
can be accomplished more efficiently and arguably more
accurately than
with a human. The time and resources saved from AI integration
would be
unprecedented in the insurance industry. According to the
Financial
Times, “AI was one of the most popular themes in insurance
tech
investments in 2016, capturing more than $500m of
funds.” These
investments are intended to provide numerous benefits for the
industry,
namely in cutting time and costs in claims processing.
Improving Claims Processing
Artificial intelligence would play an integral role in
enhancing claims
processing. Similar to its application in hedge funds,
robo-advisors
can benefit insurers in predicting claim durations. This is
made
possible by tapping into historic data about previous claims.
Strategic
planning is also more effective when implementing predictive
analytics.
The historic data sets a good precedent for determining costs
of
specific claims. An insurer can derive prices for a new claim
based on
a previous claim of similar magnitude. Without AI, this is
still
possible, however it would take the insurer a considerably
longer time
to find a correlative claim. This undoubtedly increases costs
in time,
profit, and resource management.
AI’s
relationship with historic data has positive impacts on other
aspects
of the insurance industry. Its predictive capabilities help
anticipate
repeat claims on an individual client basis. An AI system
would conduct
an in-depth review of a claimant’s personal information and
previous
claims history. In so doing, the insurer can cater to
the
specific needs of a client in order to prevent the repeat
claims from
occurring in the first place. Furthermore, with all these
predictive
capabilities of AI, the insurer can discern the next best
actions to
address a claim.
The
most
intrinsic way that AI will benefit insurers is through the
elimination
of repetitive, manual tasks, most notably in linking claims
with client
profiles. This enhances end-to-end automation, with a
trickledown
effect in synthesizing valuable information instantaneously,
decreasing
a claims cycle duration, improving customer satisfaction, and
diminishing overall operational costs.
Customer
satisfaction, one of the most pivotal aspects for success in
an
organization, is more challenging to maintain in the insurance
industry. As long as insurance companies continue to maintain
manual
and repetitive human-driven processes, closing the gap between
customer
expectations and actual experiences will be a significant
roadblock. AI
algorithms are expected to be 15 times more productive in
accomplishing
repetitive tasks than the average human, which is promising
should
insurance companies pursue revamping their services to better
meet the
needs of their client base.
AI Predictive Analytics Assess Risk
Associated to Individual Customers
Several
factors are under consideration when determining how to insure
a
customer who is considered risky. For example, accidental and
unintentional loss is examined, but quantifying it is more
resource-consuming in the absence of predictive analytics.
Another
factor that would be facilitated by AI is determining
mutuality of the
claim.6 Drawing a conclusion that there is a large number of
similar
exposure units, thus implying a different premium rate, is a
timely
process that, with the assistance of artificial intelligence
and
predictive analytics, can be resolved within a matter of
minutes.
Since
AI can
generate detailed client profiles almost instantaneously, the
base of
insured units can be scaled down. Without the assistance of a
computer
system, an insurance company, subject to human error, may
incorrectly
associate clients with exposure units that do not actually fit
their
specific needs, gender, age, or other provisions required by a
state’s
Department of Insurance. Thus, AI can benefit insurers in
creating a
base of exposure units that are thorough, precise, and
accountable to
clients with similar claims.
In
the most
fundamental aspects of AI, forecasting the probability and
magnitude of
potential losses is possible by accessing a repository of
historic
data. According to the Financial Times, AI will “look for
relationships
that traditional techniques would not necessarily pick up,”
and would
greatly impact underwriting, “where the insurers assess—and
price—each
customer’s risk.” Because of this new ability to assess risk
on a more
granular level, insurers may change the scope of insurable
risks by
eliminating some but making others more insurable.
Artificial
intelligence can uncover data about potential clients that may
make
them uninsurable. In other words, they would be too risky to
be fairly
priced and covered by Insurance companies. On the other
hand, AI
may be more beneficial to insure those who were previously
deemed
uninsurable. An insurance company can adjust policies based on
individual profiles. With AI’s predictive capabilities,
insurers can
procure predictions and profiles in a matter of minutes.
Providing
clients with speedy service and personalized policies will
improve the
customer experience and expectations.
Increasing Transparency in Insurance
The
Insurance Industry has the opportunity to greatly improve its
reputation in achieving customer expectations and maintaining
trust in
comparison to other customer service oriented companies.
However, AI
can potentially enhance the industry’s lead time to produce
services
that will undoubtedly increase the industry’s transparency and
improve
customer satisfaction. For example, Lemonade Insurance
Company, based
in NYC, developed algorithms to quickly sign and approve
claims that
insure renters and homeowners. Tasks that previously would
have taken
days to complete can be settled in a matter of minutes. It has
been
reported that the company is “fast and transparent, rather
than slow
and opaque.” Such attributes make the company more
attractive to
potential clients.
Because of
the transparency, Lemonade reported 81% of their customers
being
between 25 to 44 years old. This is a younger demographic than
most
insurance companies attract and maintain. Furthermore, since
2014,
consumers under 44 years old are two times more likely to buy
life
insurance than those over 65 years old7. This is
reflective of
the ease of use in internet services, which younger
generations are
more familiar with.
Another way
that customer satisfaction is enhanced stems emerging
chatbots.
Chatbots are another form of robo-advisors that handle
communications
with clients. They can discern the perceptions and concerns of
clients
via algorithms, such as a sentiment analysis. In a sentiment
analysis,
words are given ratings pertaining to positive, negative, or
neutral
connotations. After all ratings are gathered, the algorithm
adds up the
values and produces an overall rating. As a result, the
virtual
assistant, or chatbots, targets specific needs or concerns per
client.
AI
company
CogniCor Technologies Inc., based in Barcelona, Spain, offers
chatbots
to various industries around the world. CogniCor develops a
personalized assistant to effectively understand and respond
to the
needs of customers within a designated industry. This
translates into
costly savings and increased efficiency. In the insurance
industry, it
can offer a humanlike conversational interface with customer
care that
immediately answers questions, resolves complaints, and deals
with
claims. Incorporating a sentiment analysis into the
technology
achieves individualized services based on analyzing customer
intentions and comments.
Crop Insurance Matures with AI
It
is
expected that the farming insurance sector will greatly
benefit from
artificial intelligence. Traditionally determining coverage
for crops
damaged during adverse weather conditions is a difficult
assessment.
Damages tend to be expensive and inaccurate since they rely on
photographs and eyewitness accounts from farmers and insurance
assessors. However, artificial intelligence offers a new way
to assess
crop damage by incorporating satellite and drone imaging with
in-field
sensors. The process of quantifying damage becomes more
reliable,
quick, and effective.
One
of the
most popular AI plans for crop insurance is Aerobotics. This
plan
instigates “applying AI to aerial photos of farms.” This is
part of an
effort to decrease costs for farmers but also improve the
accuracy of
crop insurance policies. Aerobotics were first introduced to
South
African farms in recent years to determine where crops suffer
in the
fields. These completely autonomous drones have the capability
to
“detect leaks and determine whether crops are receiving too
little or
too much water…[they] also count each individual plant as well
as track
crop maturity and test drainage.” It is speculated that
such
services can be extended to incorporate assessing risk and
damage in
crop insurance.
Determining Auto Insurance in a Time
of
Driverless Cars
The
automobile insurance sector is beginning to invest in
artificial
intelligence portals in order to keep up-to-date with changes
made in
the automobile industry. Solaria Labs, which works with
Liberty Mutual
Insurance, began developing their own portal that collates
public data.
Users access this portal to determine the safest route when
driving
somewhere.
Solaria
Labs,
based in Boston, Massachusetts, intends to diminish the
possibility of accidents and claims from arising in the first
place.
The portal also incorporates an AI Auto Damage Estimator, so
that in
the event of a collision, the user can immediately determine
the damage
costs. The damage estimator is effective because it
sifts through
a depot of anonymous claims photos to identify ones similar to
the
current claim. As a result, it can estimate a cost based on
historic
trends in damage claims.
Services
such as this one offer enough “insurance expertise and
consumer testing
to help guide the decision of what services to make available
and how
to organize the data.” This will especially be the case
in the
future when cars become more automatic and rely on software
instead of
people to determine actions.
Berkshire
Hathaway Inc., based in Omaha, Nebraska, implements AI
algorithms to
offer policies to small and medium businesses over the
Internet. With
such algorithms, they tap into a plethora of data to create
more
personalized and specific policies. At the annual 2017
Berkshire
Hathaway meeting, CEO Warren Buffet remarked on the impact
driverless
cars would have on the insurance industry. He explained that a
future
with driverless cars would relate to increased safety on the
road,
hopefully leading to fewer accidents. Although beneficial for
society,
a decrease in accidents can potentially hinder profit
maximization in
the insurance industry, because “the overall economic cost of
auto-related losses [would] go down, and that would drive down
the
premium income of GEICO,” the auto insurance business for
Berkshire.
Despite
Buffet’s postulations, the insurance industry has not
experienced such
negative repercussions as a result of more widespread use of
driverless
vehicles. In this future scenario, the insurance industry will
likely
shift from insuring drivers to insuring the software in the
self-driving cars. Thus, automotive insurance companies will
merely
broaden their scope of defined risks . Cyber security might be
the
biggest threat that driverless cars will be faced with in the
future,
thus creating new types of risk.
Insurers That Use AI
Several
insurance companies implement some form of artificial
intelligence to
enhance their services and better facilitate the needs of
their
clients. If they aren’t already, then they are finding ways to
incorporate AI into their future business processes. For
example, Swiss
Reinsurance Company, the second largest reinsurer in the
world, is
currently working with IBM Watson. Together, they are
developing an
underwriting solution that will price risks more accurately on
a
case-by-case analysis. This will decrease speed process
times
without compromising on efficiency, effectiveness, or accurate
cost
assessment.
An
Austin,
Texas-based company named Buyonic Insurance implements a web
interface
contingent on artificial intelligence. The company’s
robo-advisor,
Siber, can “rate, bind, and issue policies on the spot, while
simultaneously answering the phones and making robocalls to
prospects.” Siber solves many of the issues previously
indicated
in the insurance industry. This robo-advisor offers enhanced
services
to the clients, which enhances satisfaction and transparency
within the
industry.
How AI May Disrupt the Insurance
Industry
but Benefit the Insured
Although speculations about job replacement by AI and
robo-advisors are
provocative, research indicates that AI acts as an enabler.
This is
primarily the case because machine learning and AI algorithms
have a
great degree of complexity that require oversight and input
from
humans.
Furthermore,
AI will provide more benefits for the insured. Clients, both
individual
and corporate, can discern whether or not they require
insurance in the
first place. AI would help assess their levels of risk in
varying
scenarios. In this regard, businesses can model risk to
“decide what to
retain on their balance sheets and what to transfer to
insurers.”
Artificial intelligence would be an asset to not only
insurance
companies but also those seeking to be insured.
Challenges that Insurance Companies
will
face in Adopting AI
Despite
the benefits of incorporating AI in the insurance industry,
there are
several limitations in implementation that ought to be
addressed. These
limitations are expected in any industry that pursues AI
adoption.
Regardless
of industry, there are several challenges an organization is
faced with
when implementing AI into its business processes. The most
fundamental
aspect of AI adoption is building the foundation that will
expose AI to
an abundance of domain-specific data. For AI-based solutions
to be
effective and accurate, a system would need access to
constantly
updated information that pertains to all possible business
scenarios. A
popular route to resolve this issue is by investing in cloud
services
that utilize built-in, iterative algorithms to remain
up-to-date. Then,
users are no longer concerned with the data collection aspect
of risk
and claim analyses.
As
previously mentioned, implementing AI into an organization
does not
necessarily mean human employees will lose their jobs. In
fact, they
will be required, with a new skill set, to ensure the AI
system is
glitch-free. It is noted that “technologies such as speech
recognition
and machine learning require human oversight for their work to
equate
with human capabilities.” AI systems can potentially return
unpredictable or inaccurate results. Early implementation of
AI may
demand significant oversight from a human team. The team would
have to
condition the AI system to think before acting. The last thing
any
industry wants is for an AI system to make a rash decision
derived from
literal commands without first understanding and then
incorporating the
intentions of those the system directly interacts with.
The
next
challenge in accepting AI integration is gaining enough
support from
the employees and clients. A reallocation of tasks,
management, and
overall performance objectives is required because the
organization
should yearn for a smooth transition with artificial
intelligence. It
is required that employees develop new skills to work with,
but also
monitor, an AI system. To gain AI acceptance from clients,
insurers
should put themselves in their shoes and address key
questions. These
questions include determining if AI can be trusted to make
long-term
decisions, if it can effectively discern emotions and relate
them in
the context of a complaint or claim, and if it is as reliable
as
talking to an actual person to solve issues all while
maintaining or
improving customer satisfaction.
Perhaps one
of the biggest concerns in integrating artificial intelligence
into any
industry regards privacy and regulations. There is no doubt
that “the
insurance industry will have to battle data privacy concerns
since most
AI solutions are likely to reside on the cloud of a
third-party
technology provider.” Data must be protected from all levels
of
breach and unintended use. This requires significant system
oversight
and reconfiguration of security systems within the company.
The
bottom
line is that insurance companies will have to leave their
historic ways
to support new technological development. Incorporating new
artificial
intelligence functionality into old, outdated systems might
prove more
resource intensive than leapfrogging to a new system that
integrates AI
from the get go.
There
is a
market for hyperconverged packages that incorporate storage,
servers,
and network into one platform. One virtualized storage company
called
Nutanix Inc., headquartered in San Jose, California, created
Xtreme
Computing Platform to merge computing, virtualization, and
storage.
Because of its extensive automation and system-wide monitoring
for
data-driven efficiency, it “is designed to tolerate component
failures
through fault isolation and automatic recovery without
bringing down
the overall system.”
A
service
such as Nutanix’s Xtreme Computing Platform would be
beneficial to the
insurance industry in determining what type of package to
purchase. For
an industry that is more reluctant to make the move towards
technology,
a platform that offers all-in-one would be more beneficial and
inexpensive to implement. This system might require only some
additional adjustments to cater to insurance domain-specific
needs.
Insurance
companies are not dissuaded by the challenges inherent in
implementing
AI. A survey conducted by Accenture, a consulting firm,
indicated that
204 of 550 insurance company executives, approximately 37%,
anticipate
investing in extensive machine learning efforts over the
course of the
next 3 years. An additional 44% plan to pursue moderate
investments in
machine learning integration.
Conclusion
Artificial intelligence has seen recent growth and adoption by
a
variety of industries. Now, the insurance industry realizes
the
benefits of using AI, which demonstrates an interest in making
the move
from their traditional ways to a more modern, AI-enhanced
approach.
More effective risk analysis, predictive capabilities, and
claims
processing are byproducts of AI adoption in insurance. Such
integration
would positively impact customer expectations and
satisfaction, a
crucial aspect for insurance industries to gain and maintain a
young
client base. Insurance companies that incorporate Artificial
Intelligence into their businesses are now eligible for
federal and
state tax credits.
