Google Tanslate

Select Language

Sign up and be the first to know

About Hugh Terry & The Digital Insurer

Hugh Terry & The Digital Insurer Video

Contact Us

1 Scotts Road
#24-10 Shaw Centre
Singapore 228208

Write an article

Get in touch with the editor Martin Kornacki

email your ideas at [email protected]

Pre Registration Popup

itcasia2020 Registration Popup

Share Popup

Prime Member: Find out more

Access a unique programme!
  • 56 pre recorded lesson of online content from industry experts over 7 courses
  • The best in digital insurance for practitioners and by practtioners
  • Online MCQ after each lesson
  • Join the discussion forum and make new friends
  • Certificate upon completion to show your expertise and comitment
  • 3 months to complete
  • Normal price US$1,400 Your Prime member price is US$999
  • Access to future versions included in your Prime membership!
Become a member

Prime Member: Contact Us

Reach out to us. Please fill up the form below
Let us know how we can help. You can expect a response within 24 hours
Services of interest
Untitled

Arthur D. Little

Arthur D. Little has been at the forefront of innovation since 1886. We are an acknowledged thought leader in linking strategy, innovation and transformation in technology-intensive and converging industries. We enable our clients to build innovation capabilities and transform their organizations. ADL is present in the most important business centers around the world. We are proud to serve most of the Fortune 1000 companies, in addition to other leading firms and public sector organizations. For further information, please visit www.adlittle.com

China In-Depth: AI and Machine Learning in insurance – Challenges, examples, and advice for successful implementation

View Newsletter

Artificial intelligence (AI) and machine learning (ML) technologies offer exciting and far-reaching potential for insurers to improve operational processes within an insurer’s business, and allow for the creation of completely new business models.

This article will examine AI initiatives across the spectrum of product development, customer acquisition, underwriting, policy management, and claims and look at:

  1. the general benefits;
  2. specific uses cases in China; and
  3. what is required to make AI projects successful. 

The potential for AI

A 2017 report by PwC assessed the potential real value of artificial intelligence (AI) and machine learning (ML) for business was in the trillions of US$.

The report said that by harnessing the power of AI and ML, global GDP might be increased by 14% in 2030, the equivalent of an additional $15.7 trillion. That is the largest commercial opportunity in today’s rapidly changing economy.

It said the greatest gains from AI would be in China, boosting GDP in 2030 by 26%. North America would be the next greatest beneficiary, with a 14% boost to the sectors identified as being ripe for development.

While all sectors will benefit, retail, financial services and healthcare were thought to benefit most quickly from productivity increases, while improved product quality will drive consumption.

This is not so much a case of the rise of the machines, as AI and ML will be used to augment human activity in business.

Source: McKinsey data visualisation from The executive’s AI playbook

Three years is a long time, and there will be some ground to be made up due to the impact of COVID-19 on the global economy. However, more recent data from McKinsey agreed that the upper end of the scale of contribution to global GDP would be around the same point as the PwC report.

Working smarter and harder

The McKinsey data projects that AI/ML will contribute between $9.5 and $15.4 trillion. Of course, the impact of COVID-19 remains to be seen, but the crisis may precipitate greater commitment to the digitalisation project among many insurers in the future.

The benefits of AI are accepted. It delivers productivity gains from automating processes, which will extend to include robots and autonomous vehicles in time.

The use of assisted and augmented intelligence within existing labour forces will result in more personalised and higher quality AI enhanced products and services becoming available. Building products that consumers want and need will stimulate consumer demand in the market.

Healthcare, automotive and financial services are among the sectors expected to receive the greatest benefit from disruption and product enhancement due to the use of AI. But there is considerable potential for competitive advantage in other sectors that might deliver on-demand manufacturing, or better content targeting within the entertainment sector.

There will be opportunities that arise from the deployment of this technology. While autonomous vehicles will take jobs away from delivery drivers and warehouse workers, there will be a need for the equivalent of air traffic controllers to oversee the operation of autonomous vehicles on the road.

Most projections will rely on long-term steady economic growth, but whether that is possible over the coming decade remains to be seen, as countries rebuild their economies after the global pandemic.

How will businesses adapt to AI?

Healthcare, retail and financial services may prove to be the largest beneficiaries from the growth of AI, but all sectors including automotive, transportation and logistics, technology and communications, energy and manufacturing will benefit from the innovations.

In healthcare, AI will support diagnosis to determine small variations from baselines, providing for early identification of illnesses but also potential pandemics, which given recent events will pay dividends within short order. Imaging diagnostics for radiology and pathology will also become more efficient and lead to more effective prevention of illnesses and hospitalisation. The benefit to consumers is obvious – it will deliver faster, more accurate diagnoses and the treatments will be more personalised. And potentially more effective as well.

Scheduling is being transformed by AI within medical insurance. In the medium term, diagnostics, and drug development will benefit from it.

Working with machines

Longer term, we may see robot doctors carrying out diagnosis and treatment of patients, though with healthcare, privacy and protection of the individual is paramount.

Human biology is also very complex, but a breakthrough here will deliver the ultimate tailored service with AI diagnostics making comparisons against an individual’s unique history. This will allow it to determine their relative health to their own baseline, as opposed to a more generic approach against a population or even specific demographic that is typical of traditional medicine.

This won’t necessarily mean replacing physicians with robots, but it may improve diagnosis and prognosis for almost every patient.

And while this example uses healthcare, digital touch points will be replicated and replace basic human interactions across all sectors and parts of society.

Source: McKinsey data visualisation from ‘The executive’s AI playbook’

The path to successful AI implementation

The first wave of digitalisation is making the insurance value chain more efficient, but there remains a considerable disconnect between siloed datasets (see boxout below: Building the best technology).

Developments in cognitive technologies will help such insurers integrate learning to adapt value propositions in real-time, thereby providing a holistic and unique customer experience.

In time, these critical processes will be joined, and insurers will have transitioned to be digital insurers.

Once data quality and algorithms have been improved, machine learning will play a bigger role, and will be more capable of identifying patterns and learning from the data it accesses with value chains that ‘learn’ from data generated by consumers, ecosystems and governments.

Challenges to digitalisation

The regulatory framework will be important in shaping the deployment of new technology in insurance. Data protection and privacy have become a major focus for all regulators as society gradually digitalises its daily work.

The tech can receive bad education, with mistakes and biases being hard coded into solutions. This not only poses danger from the point of view of misplacing risk, but from creating an entanglement with a regulator through breaching safety or fairness for consumers.

Data has to be of the same ‘industrial strength’ as the systems insurers must build (see boxout below: Successful implementation).

Data governance must be robust and transparent to satisfy consumers and regulators.

Successful implementation

Hugh Terry, founder, The Digital insurer

There are four areas in this process: cleaning your data; understanding your data; modelling your data; and then getting actionable insights from that data.

That is what people need to optimise. There’s a number of challenges in this area, but what you do not want to do is reinvent the wheel with every project.

To create enterprise level actionable insights using data is the goal, so that being able to do this becomes business as usual.

Legacy companies suffer because their data is inaccurate, out of date, incomplete or doesn’t necessarily have the right validation.

It may sound simple to resolve that, but we’re talking about millions of items of data and personal data. That can be quite challenging. And if your data isn’t clean, then you will be undermining your entire insights process.

People tend not to spend enough time understanding what the data is and what data is available.

Quite often the business will outsource this, but I believe you need data owners inside the business. That means understanding what that data really means from a business perspective, understand where it came from. Just because something’s labelled x doesn’t mean it’s x. That’s a basic level of hygiene that’s needed.

Once that is done, you want to train models that can give insights. There are a couple of approaches that may be used and this is specialist work. You either need your data scientists as a centralised team to build those models for you, or hopefully you may make use of tool sets that are available from your solution provider, whereby AI becomes a service.

Even with the best model in the world, without knowing what you’re trying to solve or what you want to do with that outcome, you’re in danger of becoming stuck in the theoretical.

Before starting any AI project, determine the potential value to the business. Where should you spend time building models and testing?

Once that is done, these models need to be released by the data scientists. These algorithms need to be made ‘industrial strength’ and delivered as close to the user as possible to make them effective.

If it is an underwriting decision being made, then you want that AI to be sitting on the underwriter’s desktop so it will digest the data and flag complex cases immediately. 

 

Building the best technology

Yannick Even, head of digital and smart analytics, Swiss Re APAC

Data Scientists working in silos never work, says Yannick Even, head of digital and smart analytics at Swiss Re APAC. What insurers need is data scientists working hand in hand with business experts who know how to quickly translate ideas that will solve industry well-known challenges into “tangible analytics”.

In order to succeeded implementing advanced analytics, the insurer must be clear about the digital and data capabilities that need to be established, says Even.

“You need talent in house that can manage data partners, because these days a lot of information comes not only from insurers’ backend, but also from distribution and data partners that can enrich traditional data.

“You also need businesspeople and actuaries that understand tech and data science potential to turn initial ideas and use cases into scalable solutions to further digitalise the value chain and create a more data driven business to better serve customer while managing associated risk.

“These capabilities, together with the right data-driven culture can create value and potential to scale so you may replicate initial AI successes across the whole organisation,” he says.

AI is already impacting the entire value chain, and this started many years ago, mainly with customer and claims analytics, says Even. Today, a number of markets, including China, are leveraging the data they have accumulated or newly created.

“With the acceleration of insurance digitalisation, and as insurer are getting rid of operational silos, the resulting richer data set fuels the machine learning models that start to perform better and better,” says Even. “There is no going back to the old ways and this could very quickly create competitive advantage.”

It also raises loyalty with customers who see their provider offering a modern suite of more personalised products and services that they can relate better to.

There is indeed a virtuous circle at play, the technology and associated data/models also help to reduce fraud, identify the valid claims to be paid more quickly, and support the insurer’s journey to “prevention”. Customers feel rewarded by being truthful, having access to more personalised services and protection, and by being provided with accurate pricing and day to day prevention advice – and the more customer touchpoints, the more the machine learning model improves.

 “This helps you retain the good risk longer in your portfolio,” says Even, “but it can also help you to manage the bad risk better than your competitors.

“We also already see some of the digital/tech savvy (re)insurers tapping into the opportunity to diversify and receive additional revenue from providing their superior tech and digital platforms to the rest of the industry.”

 

Uses cases in China

1. Product development

Macroeconomic trends such as urbanisation are driving the emergence of new customer segments that require new product development.

There are many sub-sets including young professionals and small business owners, many of whom offer support to their parents.

To date, there have been surprisingly few AI initiatives, but there are some exemplars that are offering others a roadmap.

 a) CausaCloud

CausaCloud has developed a medical underwriting engine, intelligent claims platform, data intelligent platform and other systems in order to assist its customers – insurance companies – to operate more effectively – and intelligently.

It works with insurance companies to develop products for single diseases, such as reoccurrence of breast cancer, specific children’s blood disorders and fertility treatment to broaden the market of eligible customers.

It integrates treatment, medication, and protection while covering areas that might normally be excluded from more general policies.

The underwriting engine uses optical character recognition (OCR) and natural language programming (NLP) and uses its clinical judgement model to make clinical decisions.

It has been shown to increase underwriting efficiency by around 50%, with accuracy of up to 90%, which is equivalent to an experienced underwriter.

More than 20 insurers were using the platform and in the month after the COVID-19 outbreak, more than 50,000 pieces of underwriting were complete, covering premiums of 120 million yuan.

CausaCloud’s intelligent claims platform use NLP which helps to minimise data entry errors by customers or agents. Claims processing times have been reduced by more than 80% – down to less than two minutes from 10 – and guaranteed the consistency of payment.

b) Malgo

Malgo operates Semantic Healthcare, a data driven, individual risk oriented, interactive healthcare. Compared with traditional models, Malgo claims it can reduce loss-ratio by more than half.

The core intelligent risk management system uses all relevant health data to manage risk throughout the business process.

AI and big data play a major role through NLP information extraction, a deep learning platform, a medical knowledge graph, data analytics, data mining and decision-making management.

This allows insurers to design and price products based on precise individual identified risks.

By identifying and treating with preventative medicine, medical expenses are reduced.

The claims process provides effective administration of drug and medical analysis, and assists in the identification of fraud.

This provides insurers with a healthcare product that produces precise risk identification with granularity of data.

2. Marketing/customer acquisition

China may be a nursery for the establishment of the AI/ML revolution, but it currently relies heavily on human agents engaged in face to face consultations with customers.

Marketing is becoming more granular and ‘smarter’. As the number of interactions with customers increase, so does satisfaction and loyalty. Customers become used to the interactions and welcome them – provided the insurer offers effective solutions to the problems they face.

Chatbots have become a common part of this process, and have overcome the initial reluctance of consumers to deal with ‘a machine’.

NLP has greatly improved the ability to ‘speak the customer’s language’, making them feel appreciated and in the end, informed or satisfied.

a) Ping An Life

As part of Ping An’s transformation from traditional insurer to technology company, the company is undertaking an overhaul of inefficient processes and developing new and better products that exploit data analytics.

Despite having one million agents, this channel is not working for the company. Recruitment is slow with high failure rates, there are too few agents operating full time, processes are paper based and there are too many managers blocking development and reward of good agents.

A four phase plan has already drastically cut the amount of paper processes being used by agents, reducing underwriting time from 5 days to 15 minutes.

Most (91%) agent training is delivered online, allowing a 70% reduction in training service staff.

This has delivered a 32% increase in productivity.

The third phase will see AI used to improve agent recruitment and screening . It will also allocate financial resources more effectively and assist in career planning for the agency force.

This process will save 1.43 million working hours – or 60,000 days/163 years – every year. This boils down to an annual cash saving of 630 million yuan. AI will also assist in ensuring agents remain fully compliant, monitor real time communications, agent behaviour, and assess agent risk.

AI lead generation has already achieved a 10.8% conversion rate, over 220 million customers through 1.3 billion interactions.

Ping An expects to see a 2.8 billion yuan increase in sales revenues while reducing the loss from lapsed premiums by 2 billion yuan, and reducing fraud by 350 million yuan a year.

b) WeSure

TenCent’s insurance platform WeSure works with Chinese insurance companies to co-design high quality insurance services. WeSure’s users are then offered insurance purchases, queries and claims via the WeChat app.

The company’s 60 strong AI team designs products with providers, and the system manages underwriting risk.

c) Riskeys

Competition is fierce in China, forcing many smaller insurers and intermediaries to think of alternative ways to generate traffic and acquire users.

Riskeys is a Shanghai-based aggregator that is targeting the health and life segment with a new approach to customer acquisition using a combination of WeChat incentives and an aggressive focus on new product development to differentiate itself.

Riskeys develops tailored products for market segments within auto insurance, investment linked life insurance, high-end medical insurance, and personal accident insurance.

The solution’s functions include risk assessment, disease assessment, open question and answer, intelligent recommendation and insurance analysis. From assessing risk, it takes customers through a health management consultation to an open insurance knowledge consultation and recommends insurance products according to user characteristics, and similar products on the market.

It is combining local market knowledge with the opportunities provided by the WeChat platform to become a one-stop intelligent insurance consultant function for comparing insurance products.

3. Underwriting 

Underwriting is a core competency of insurance, but it is changing. Data is driving forward reforms that are identifying new risks, or if not new, extending understanding of existing risks.

This is particularly true at times of crisis such as the COVID-19 pandemic. Data was made openly available in a effort to combat the virus. This provided far greater insight not only into the crisis, but also existing pandemic models.

The future of successful underwriting is predicated on the integration of big data and analytics in the process.

Not only will it deliver better priced cover, but identify new – and insurable – risks, opening up new marketing opportunities.

a) Zhong An

Zhong An’s products use smart underwriting processes, which can complete underwriting assessments in between one and two minutes.

More than 20% of customers previously denied insurance have now successfully obtained one million yuan medical insurance, covering more than 200 diseases.

Customers are able to intuitively understand their cover based on their existing medical conditions.

b) InsuredX

InsuredX is a cloud-based underwriting tool designed around the concept of the product pipeline.

Deployed through the cloud, it quickly – and cost effectively – adapts to changes in the market.

Real-time, immediate changes can be made to specific products without any system downtime.

It copes well with typical internet application scenarios with high concurrency, high load, and high fault tolerance.

It is a lightweight, extensible and configurable system based on the concept of product pipeline management – including products, formulas, decision tables, verification rules, underwriting rules, workflow, interface data format binding, etc.

It offers a progressive insurance industry solution for microservices architecture, simplifying processes and allowing changes of direction as demand dictates.

4. Policy management

A large proportion of the world’s population is happily adopting digital technology, but there will always be some consumers who are less comfortable in an automated world.

Insurers must service both populations and win over those avoiding digitalisation. To achieve this, insurers must demonstrate that AI and ML will improve the consumer experience and avoid mistakes, as this is what tends to lead to frustrations.

Smart contracts are already winning the hearts and minds of consumers, even though very few are probably aware that they have one.

These are self-executing programs that trigger payments when a risk event occurs, such as a flood or other natural disaster, or the loss of luggage.

It automatically transfers payouts to the consumer, possibly without the need for that customer to lodge a formal claim.

But the power of AI is present from the first contact the customer has with the insurer. Chatbots and other AI driven logic processes lead customers through a process without any need for human intervention. Some of these bots have become so sophisticated that many would believe they have been interacting with a real life employee, rather than a machine.

a) Xiao Hui, China Taiping’s AI virtual agent

China Taiping’s virtual agent aims to to make customer service representatives more cost-effective and assist with customer calls.

Fielding calls during times of peak demand allows human representatives to focus on more complex tasks, while maintaining high levels of consistency and customer service.

More than a chatbot, Xiao Hui uses a realistic human-like voice to enable a comfortable and familiar user experience, much like the customer would expect at a branch.

b) Ping An

Biometric authentication helps Ping An form a relationship with pending policy holders (those who have applied for life/health insurance but are waiting for the policy to be activated). Since introducing this feature, the rate of customers that withdraw their application during this window between application and activation has dropped to 1.4% against an industry average of 4%.

5. Customer service and claims

The traditional insurance model separates customer service, claims and marketing, but digitalisation is drawing these functions together – and with good reason.

Marketing in the digital world relies on an increased number of digital touchpoints with customers. This creates a new form of relationship between customer and provider and creates greater loyalty.

Customer service is the function of helping clients when they’re not being sold to, but is essential to maintain good relations.

The crisis point in any relationship is often the greatest moment of stress, when a policyholder needs to make a claim.

The single most frustrating aspect of having protection in place, has traditionally been the process of making a claim.

AI and ML are increasingly being deployed wherever customers contact the insurer to provide a more natural, ‘human’ response.

Data analytics underpins these processes and can identify the status of a claim – even pre-empting a claim in some cases – managing expectations and even weeding out potential fraud.

a) Ping An – One Connect

Ping An’s biometric authentication feature offers policyholders an option to enhance the security of their account through face and voice recognition.

The feature, delivered through Ping An’s ecosystem of apps, scans the structure of the face – in particular the nose and eyes – with a higher degree of accuracy than can be done with the human eye alone. It is also being used to verify the identity of insurance applicants or agents.

b) AIA – Fusion app

AIA’s Fusion app claims to combine the speed and reliability from the processing power of a machine, with the human qualities of creative power, experience and fairness.

This AI/ML driven system can learn and adapt to human guidance and supports customer service, claims and underwriters to work faster and smarter, by facilitating thousands of applications a second.

Automation is rule-based but integrates superior cognitive automation, allowing Fusion to learn patterns from the analysis of 100 years of AIA customer data.

Livefest 2019 Register Popup Event

Livefest 2019 Already Registered Popup Event

Livefest 2019 Join Live Logged-in Not Registered

Livefest 2019 Join Live Not Logged-in