Artificial intelligence and automation of the insurance industry

Supposed to be safe from any kind of automation, sectors such as finance, medicine, journalism, research or arts are now more and more attracted to this movement which is only starting.

Artificial intelligence experiments in insurance industry

Digitization is indeed well instilled in the insurance companies of developed countries which are up against a new challenge, that of robotization.

According to a survey published by Mckinsey Global Institute, in early 2017, 43% of the tasks entrusted to the insurance sector may be carried out by machines and 25% of the jobs should be replaced by robotized systems by 2025.

According to a survey conducted a few months earlier by Accenture: “Technology vision for insurance”, the insurance business is anticipating this evolution, preparing for this new order. Approximately 82% of the insurers quizzed said they would be willing to invest in solutions integrating a component of artificial intelligence (AI) in the course of the next five years.

Automation is consequently at the core of this new phase of restructuring. Numerous market players have already taken the lead such as the Japanese life insurer Fukoku Mutual Insurance. By means of IBM’s Watson Explorer system, Fukoku Mutual has integrated AI in its work chain since March 2017. This decision triggered the replacement of 34 wage earners by a robot.

This system, whose cost amounts to 1.7 million USD in installation fees and 128 000 USD in annual maintenance expenses, has the capacity of analyzing thousands of data in a few seconds. It may therefore process 132 000 files in a single year, which, for a year of 220 working days, represents 600 files per day. Watson Explorer is also able to gather customers’ medical data, read certificates and other medical documents, determine payments and bill expenses.

Still in Japan, two other companies, Dai-Ichi Insurance and Nippon Life Insurance are about to imitate Fukoku. Automation and artificial intelligence are being integrated in their work chain.

In Europe, French insurer, Natixis Assurances, has taken the step. He entrusted contract termination and email processing to a robot capable of carrying out the equivalent amount of six-day labor of a staff member in a single night.

British Aviva is also among the insurers who have acquired this system. The company has proposed the redeployment and relocation of its employees whose jobs could be threatened by a robot. It is thus considering the possibility of entrusting the tasks related to call centers and premium calculation to a system of artificial intelligence.

Equally present in research, the insurer is keen on investing 123 million USD by 2020 in start-ups specialized in AI.

The use of the artificial intelligence in insurance

Within an insurance company, artificial intelligence may play out at two levels: the back-office and customer relationship.

Back-office

Most of automation announcements or initiatives pertain to the operations carried out at the level of the back-office: establishment and drafting of contracts, calculation of contributions, claim reimbursement… That is the case of IBM’s Watson Explorer program which is integrating more and more working languages and which continues to evolve toward new uses such as claim amount computation.

Customer relationship

In the field of insurance, artificial intelligence is more and more focused on customer relationship. The system operates at the insured’s convenience regardless of the time or the day when the call is made. Availability and proactivity are the genuine assets for the integration of the automation system. Hence, it is all about sub-contracting low value-added tasks while ensuring reactivity and service quality. The resources released may consequently be redirected toward better targeted and further customized consulting and monitoring tasks.

Customer relationship may also be ensured by means of conversational (talking) robots. That is the case of the “chatbots” system: a virtual assistant which uses AI to lead conversations. Another example is the “Robo-advisors” which replaces the task of insurance advisor and asset management by automated systems.

Several insurers and mutual insurers have already embarked on these kinds of programs. Others are contemplating the issue and would unveil their projects soon.

Benefits of automation and artificial intelligence on the insurance industry

The adoption of the artificial intelligence would characterize some of the insurers’ activities with more rigor and rationality.

Subjected to strict regulation and rigorous requirements with regard to auditing and to data quality and security, the insurance market will only benefit from automation. Indeed, AI will make it possible for the insurance market to meet most of its requirements.

Consequently, the recourse to artificial intelligence will lead to:

• an optimization of management costs: a nearly 80% reduction in expenditure,
• more attractive rates,
• a substantial gain in time of about 40 to 80%,
• a reduction of human error,
• a decline up to the suppression of recurring tasks,
• insurance fraud detection, which represents an economic stake of 2.6 billion USD ⁽¹⁾, for insurers. Shift Technology start-up has, for instance, suggested an algorithm capable of detecting fraudulent conduct even before the occurrence of the fraud,
• forecasting and controlling risks thanks to a visual identification system of the seriousness of the claim,
• reinventing the insurer’s profession while taking advantage of an evolving system,
• improvement of the efficiency of the current processes in the insurance industry,
• identification of customer termination risk.

Disadvantages of robots in the insurance industry

The establishment of a system based on the technologies of artificial intelligence is not risk-free. Some concerns may be raised as follows:

• the insured’s preference to maintain human contact,
• final decision making: Once the customer file has been finalized, it is up to the human staff to take the final decision,
• robots are confined to some administrative, recurrent or accounting tasks taking place in the back-office,
• concerns regarding job cutting,
• risks may stem from the robot itself: malfunction, failures, bad programming, misinterpretation of data.

Impact of artificial intelligence on employment in insurance

In a context characterized by soaring automation, the labor market will be up against two scenarios: a wave of technological unemployment or just the opposite, that is, the emergence of new posts, not available today.

For the first hypothesis, analysts are exhibiting some concerns as regards job shedding and their replacement by robots with artificial intelligence. These concerns rest on the results of several researches conducted with some advanced markets.

In France, for instance, the survey carried out by Roland Berger’(2) firm has found that by 2025, 3 million wage earners will have been laid off from all kinds of sectors of business. In the United States, researchers at the Oxford Martin School found that 47% of the overall jobs may become automatable.

Other analysts, however, have come up with a more optimistic vision. For them, there is no doubt that some activities are doomed to disappear, but again, others will transform. Moreover, we have to expect the genesis of new posts in order to meet recent technological requirements.

AI will, therefore, give birth to new jobs, new ways of utilizing technologies and ultimately new resources.

⁽¹⁾ 2014 data
⁽²⁾ Study conducted in 2014 by Roland Berger’s firm on the effect of robotics on employment