Insuring a Brighter Future With AI

By Iain Brown, Senior Data Scientist, SAS UK & Ireland

How ready is your insurance organisation for the advent of AI? In the summer of 2017 I took part in a pan EMEA review of the readiness of insurance companies for implementing AI, the full report can be found here: Pragmatic AI for Insurance. Some of the findings were surprising, highlighting pockets where AI adoption is already moving at pace, contrasted with a lot of hype and noise with limited execution.

As part of the study, we identified that although AI-powered underwriters, claims handlers and customer service agents may sound like a utopian future, in fact it’s already a reality right now in the present day. Some of the leading insurers we work with are already deploying AI capabilities that are transforming the customer experience, improving underwriting accuracy and cutting claims assessment decisions from weeks to mere seconds. A good example of this is the application of machine learning to automate car insurance claims processing. Indeed, the AI advantage is so powerful that there’s a very real risk that these early adopters will open up a head-start that followers will be unable to close. Little wonder the question we get asked the most is “How far are we behind the curve, and how do we catch up?”

We can offer some reassurance however. Although we highlight in the report that some insurance companies are reaping the rewards of early AI adoption already, for now, they are the exception. For the majority, AI is still at the hype stage – a lot of discussion, encouraging amounts of experimentation but very limited operational execution.

Quite simply, many organisations are not yet able to deploy AI. The advanced algorithms that typically empower AI only astound when fed the right data – and too many insurers are struggling to collect and store the right data or their data is held in vertical product siloes, blinding them to a single customer view. With the deadline for the General Data Protection Regulation (GDPR) now looming, there’s more incentive than ever for insurers to get their data housekeeping in order – this will then create a solid foundation on which to build an AI-enabled enterprise.

A word of caution we can offer: AI is a powerful tool for enterprise-wide transformation, arming human staff, be they underwriters or frontline service agents, with the data and tools to delight customers, optimise operations and capitalise on new opportunities. But AI, particularly deep learning black box solutions, needs governance and oversight to ensure the outputs are consistent with the company’s values, ethics and regulatory obligations. We are reassured to see companies now hiring Chief Data Scientists to make sure these issues are debated at the highest level. Any company readying for an AI future should make sure transparency, compliance and ethics are addressed now and not bolted on as an afterthought when it may already be too late.

A final consideration is that insurers need to focus now on developing a clear data strategy and carefully consider what data they would like from customers, and what value the data can provide to both parties. Without the fundamentals in place, the true rewards from AI will be harder to achieve.

By Iain Brown, Senior Data Scientist, SAS UK & Ireland

How ready is your insurance organisation for the advent of AI? In the summer of 2017 I took part in a pan EMEA review of the readiness of insurance companies for implementing AI, the full report can be found here: Pragmatic AI for Insurance. Some of the findings were surprising, highlighting pockets where AI adoption is already moving at pace, contrasted with a lot of hype and noise with limited execution.

As part of the study, we identified that although AI-powered underwriters, claims handlers and customer service agents may sound like a utopian future, in fact it’s already a reality right now in the present day. Some of the leading insurers we work with are already deploying AI capabilities that are transforming the customer experience, improving underwriting accuracy and cutting claims assessment decisions from weeks to mere seconds. A good example of this is the application of machine learning to automate car insurance claims processing. Indeed, the AI advantage is so powerful that there’s a very real risk that these early adopters will open up a head-start that followers will be unable to close. Little wonder the question we get asked the most is “How far are we behind the curve, and how do we catch up?”

We can offer some reassurance however. Although we highlight in the report that some insurance companies are reaping the rewards of early AI adoption already, for now, they are the exception. For the majority, AI is still at the hype stage – a lot of discussion, encouraging amounts of experimentation but very limited operational execution.

Quite simply, many organisations are not yet able to deploy AI. The advanced algorithms that typically empower AI only astound when fed the right data – and too many insurers are struggling to collect and store the right data or their data is held in vertical product siloes, blinding them to a single customer view. With the deadline for the General Data Protection Regulation (GDPR) now looming, there’s more incentive than ever for insurers to get their data housekeeping in order – this will then create a solid foundation on which to build an AI-enabled enterprise.

A word of caution we can offer: AI is a powerful tool for enterprise-wide transformation, arming human staff, be they underwriters or frontline service agents, with the data and tools to delight customers, optimise operations and capitalise on new opportunities. But AI, particularly deep learning black box solutions, needs governance and oversight to ensure the outputs are consistent with the company’s values, ethics and regulatory obligations. We are reassured to see companies now hiring Chief Data Scientists to make sure these issues are debated at the highest level. Any company readying for an AI future should make sure transparency, compliance and ethics are addressed now and not bolted on as an afterthought when it may already be too late.

A final consideration is that insurers need to focus now on developing a clear data strategy and carefully consider what data they would like from customers, and what value the data can provide to both parties. Without the fundamentals in place, the true rewards from AI will be harder to achieve.