This talk was recorded in London on October 30th, 2018 and can be viewed here: https://youtu.be/LFVIGMMlfhI
A view on what is driving AI and ML developments in insurance and why.
• What is driving the change in insurance and why is AI/ML so important?
• What does the future look like?
• Which AI/ML use cases are being worked on in the industry?
• Which ones are needed?
Chris Madsen is Chairman and CEO of Blue Square Re N.V., Aegon’s internal reinsurer and a company he co-founded in 2010.
Mr. Madsen holds a Masters in Engineering from Princeton University in Princeton, USA. His undergraduate degree is in Mathematics and Economics. He is an Associate of the Society of Actuaries, a Member of the American Academy of Actuaries and a Chartered Financial Analyst.
He started his professional career in New York in 1990, working as Consulting Actuary and later Principal. Mr. Madsen has published numerous articles on innovative underwriting risk solutions and is a frequent speaker on the topic and related developments.
Mr. Madsen is an avid proponent and driver of integrating start-up and insurtech expertise into insurance solutions - including internet-of-things applications as well as blockchain initiatives such as “B3i”. He is also responsible for the ground-breaking longevity solutions that Aegon brought to the capital markets totalling over EUR 20bn of reserves.
A View on AI in Insurance - Chris Madsen - H2O AI World London 2018
1. Chris Madsen
Co-Founder and CEO
Aegon Blue Square Re
https://www.aegon.com/about/Aegon-Blue-Square-Re-NV/
A view on AI in
Insurance
2. Cost of sensors are decreasing and data availability increasing
Driving an inflection point for change in insurance
Time
Cost
SensorsInflection zone
3. Insurtech has emerged
Level of Interest in Insurtech over time
Compared with Biotech and Fintech
Source: Google Trends Source: Venture Scanner
International Trends
Though still relatively nascent
4. Insurance is increasingly personalized and data driven
CONNECTED CAR
driving style, speeding,
Braking, fuel savings,
Maintenance, e-call
CONNECTED LIFE
daily activity, diet,
sleep, stress
CONNECTED
FINANCE
pension, investments,
savings, payments
CONNECTED HOME
Smoke alarm, water leaks,
burglary
6. * Tends to be significantly higher for chronic conditions such as diabetes
As a result, the role of insurance changes
Helping you when bad
things happen
Helping you prevent bad things from
happening, but when they do, we
will help you manage
Price =
Expected claims +
Loading for Risk* +
Loading for Expense
Expected claims: can be lowered through active engagement
Loading for Risk: can be lowered by more frequent touch-
points as long as pricing can vary
Loading for Expense: The greater the automation, the lower
the expense
7. And the insurance value chain along with it
Claims
Product
Development
Operations &
Servicing
Sales/
Marketing
Under-
writing
Pricing/
Reinsu-
rance
Business &
Market
Intelligence
Automatic
Claims
Product
Development
Interaction
and Advice
Distribution Scoring
Calibra-
tion
R&D
Data Analytics / AI / ML
Distributed Ledger / Blockchain / Smart Contracts
Robo Advice & Tools
CurrentFuture
Process is labor intensive driving high fixed costs
Process fully automated significantly reducing fixed costsEnvironment creation
Pre-market In-market and Interaction
Pre-market Post-marketUnderwriting and Sales
AI and ML touch every component and are value enablers
8. Calibration
Calibrate pricing to
better match
commercial conditions
Examples
• ”Pricing” as
opposed to
“costing”
• Focus on value
added
Subject to local rules and regulations
Leading to many potential use cases
Distribution
Use alternative data
sources to reach
customers in a cost
efficient manner
Examples
• Targeted
advertising
• Peer networks
Scoring
Determine risk score for
given data and product
elements
Examples
• Mortality by postal
code
• Driver analytics
• Health by wearable
data (fitness, ecg, etc.)
• Health by new data
(images, microbiome,
diet, epigenetics,
genetics)
Interaction and
advise
Determine optimal
insurance structure for
customer
Examples
• Finding proper
“bot” responses
• Leveraging data to
help match
customer and
product
Automatic claims
Develop ingredients to
automate claims
process
Examples
• Fraud detection
• Coverage analytics
and trends
9. Through Driverless AI
Financial Market Data
Financial Market Data
• Simple time series file with 12
month S&P 500 returns
• Relatively simple short and long-
term indicators – momentum and
fundamental
• Fascinating results in about 1 hour
15 minutes
10. Through Driverless AI
All Cause Mortality Risk Scores
All Cause Mortality Risk Scores
• A bit of reverse engineering
• Tested results of model already
designed to see what Driverless
AI would do
• Again, fascinating results in a
little over an hour essentially
picking the design of the model
and consistent with our research
11. * Dutch life expectancy
Based on Lifestyle
Assessing life expectancy
Chance of age
45 living to
age 65: 98%
Chance of age
45 living to
age 65: 92.5%
Chance of age
45 living to
age 65: 73%
*
12. Chris Madsen
Co-Founder and CEO
Aegon Blue Square Re
https://www.aegon.com/about/Aegon-Blue-Square-Re-NV/
Questions?