The Future of Insurance. Analysis of artificial intelligence perspectives and how AI will change the insurance business

Artificial Intelligence promises to revolutionize a number of industries, but with its tremendous promise comes new risks. In its study, Allianz looked at new loss and liability scenarios.

In a new report, the insurer identifies key benefits and challenges brought about by the increasing use of artificial intelligence in society and industry, including the insurance industry. Artificial intelligence, also called machine learning, is essentially a software tool that can think and learn like a human.

"Artificial intelligence brings both potential benefits and challenges in many areas: economics, politics, mobility, health, defense and the environment. New risk management strategies will be needed to maximize the benefits of adopting advanced AI-based technologies in society," says Vadim Pinskiy, who heads the Pheno, a startup aiming to transform insurance underwriting.

The insurance industry was one of the first to use artificial intelligence-based technology to automate repetitive processes, drive risk analysis and improve customer interactions.

AI-based technologies will increase companies' vulnerability to cyber-attacks and technical failures, leading to larger scale losses and disruptions.

Companies will face the challenge of shifting responsibility from humans to machines and its manufacturer. A new concept of risk management is needed that takes into account the growth of losses caused by the use of artificial intelligence.

Risk management is a key tool that can maximize the net benefit from the use of artificial intelligence. In doing so, software availability, security, accounting, accountability and ethical issues become critical aspects for identifying threats.

Scientists are already confronted with the fact that AI is capable of self-learning to almost the same extent as humans. The work of the artificial brain, like the human brain, is based on a multitude of neural connections that are built up as a result of the experience accumulated by the machine. Given that AI performance is much higher than that of the human brain, the learning process of the robot will go much faster. For example, in 2014 the supercomputer was able to solve one of Erdős's mathematical problems, whose solution is so complex that humans are not able to check it - one file with the equation takes 13 gigabytes. Obviously, in terms of intellectual work, AI has already surpassed humans, and after 2030, according to experts, the artificial intelligence will surpass us in terms of overall level of consciousness.

Given the speed of AI development, in the near future, intelligent machines will be able to do without humans at all.

Chat bots, autonomous robotic cars and connected machines in digital production facilities are harbingers of the near future: the widespread use of artificial intelligence technologies gives companies many advantages, such as increased efficiency, reduced repetitive tasks and improved customer service. At the same time, if these technologies fall into the wrong hands, potential threats can easily counterbalance such significant benefits. As society and the economy become increasingly interconnected, so does vulnerability to cyber-attacks or technical disruptions, as well as the potential for large-scale disruptions with enormous financial losses. Companies will also have to face new liability scenarios as responsibility for decision-making shifts from the human to the machine and its manufacturer.

Today, "weak" or basic forms of artificial intelligence are capable of performing specific tasks. Future generations of technology, however, based on so-called "strong" artificial intelligence, will be able to solve more complex problems and perform complex operations. Artificial intelligence is beginning to be used in virtually every industry, from chatbots that offer financial advice to technology that helps doctors diagnose cancer. Artificial intelligence is being used to drive autonomous cars, predict the weather, perform financial transactions, or monitor and control industrial machines. According to Accenture, artificial intelligence could double the annual economic growth of 12 developed countries by 2035.

But along with this potential benefit come risks. Two different faces of new technologies such as artificial intelligence illustrate cyber risks, which are among the biggest risks for companies according to the 2018 Allianz Risk Barometer: Artificial intelligence-based software could reduce cyber risks for companies by improving attack detection, but could also increase them if hackers gain control of systems, machines or vehicles. Artificial intelligence can pave the way for more serious and better-targeted cybersecurity incidents, reducing the cost of organizing attacks. The same hacker attack or programming error can be repeated on multiple machines. It has already been estimated that a major global cyber attack can cause losses in excess of $50 billion,[2] but even a 12-hour outage of a cloud service provider can cause about $850 million in losses.

Emerging Artificial Intelligence Risks

To identify the emerging risks associated with artificial intelligence, Allianz has focused on five key aspects - software accessibility, security, accounting, liability and ethical issues.

If appropriate measures are taken in each of these areas, the development and deployment of artificial intelligence causes the least harm to society. Preventive measures that reduce the risks of unintended consequences are extremely important."

For example, in security, the accelerated introduction of AI-based systems into the marketplace can lead to inadequate or negligent compliance testing, which is necessary to ensure that systems are installed that are safe, functional and cyber-attack-proof. This, in turn, can lead to an increase in defective products and product recalls.

As for liability, artificial intelligence-based systems can then take over many of the decisions that were previously human prerogative, but they cannot legally be held responsible for those decisions. In general, liability for defects that are detrimental to users is borne by the manufacturer of intelligent systems or their software developer. However, artificial intelligence decisions that are not directly related to design or production, but are made by the system because of its interpretation of reality, will not have an explicit culprit under current law.

New Emerging Risks of Artificial Intelligence

Leaving decision-making to the courts could prove expensive and inefficient if the number of losses brought by artificial intelligence begins to grow. A solution to the lack of legal liability could be the creation of expert agencies or authorities that would develop a liability concept in which developers, manufacturers or sellers of products based on artificial intelligence would be subject to limited civil liability.

Meanwhile, insurers will play a key role in minimizing, managing and transferring the new emerging risks associated with artificial intelligence technologies. To equally protect both consumers and companies, traditional coverages will need to be adjusted. Insurance will need to better neutralize certain risks to companies, such as cyber-attacks, business interruption, product recalls and reputational damage. New liability insurance models are likely to be adopted, for example, in areas such as autonomous cars. This will increase pressure on software manufacturers and vendors and reduce direct consumer liability.

In 2018, the development of artificial intelligence will wipe out 9 percent of jobs. The insurance industry is no exception.

Insurers were among the first to use artificial intelligence

The insurance industry started using machine learning early because it analyzes large amounts of data and many repetitive processes.

Artificial intelligence has enormous potential to improve the insurance value chain. It will help automate insurance processes to give us a better customer experience. It will make policy issuance and claims handling faster and more efficient.

By giving a boost to data analytics, artificial intelligence will also provide insurers and their customers with a much better understanding of their risks so that they can reduce them more effectively and develop new insurance solutions. For example, artificial intelligence-based analytics can help companies better understand cyber risks and improve protection against them. At the same time, technology can help insurers identify scalable cyber risks. Last, but not least, artificial intelligence will change the way insurers interact with their customers, ensuring that they serve them around the clock.

The Impact of Artificial Intelligence on Value Creation in Insurance

Artificial intelligence will also affect the value insurance brings to customers. In other words, today's AI-powered technologies are improving the value chain in insurance, making it more effective in targeting customer needs and delivering services in a timely manner and at a lower cost. The most common technologies of direct relevance to the insurance value chain are described below. Such technologies improve insurance productivity by delivering more value to customers for less cost and in less time.

Intelligent agents.

In the future, artificial intelligence will support underwriters in data analysis and risk assessment. There are many areas, such as reputational risks, cyber risks, supply chain and economic and climate risks, where machine learning can help companies better understand their risks. Artificial intelligence can also be used with other new technologies, such as blockchain, to enable new, faster and more customer-centered services. For example, sensors on shipping containers already provide data on the location and condition of cargo, which, once analyzed, can activate insurance coverage or damage mitigation measures if goods are damaged. The results of data analysis and analytics using artificial intelligence will help push the boundaries of the insurable, complement existing products and give rise to new risk transfer solutions in areas such as business interruption without physical damage and reputational damage.

How artificial intelligence affects insurance operations

Market management and customer relations

Unlike other industries, the insurance sector is characterized by a relatively low frequency of interaction with customers (customers usually only contact the insurer when purchasing a product or claiming a loss). Therefore, the ability to use customer data to understand customer preferences is extremely important. Artificial intelligence can be useful in this by sorting and analyzing customer information and providing accurate customer profiles when developing successful individual marketing campaigns. In addition, artificial intelligence helps make sense of enormous amounts of market data, including economic demand, competitor offerings and market volatility, to set product rates according to current and projected demand.

Financial Advice.

Artificial intelligence-based technologies also help recommend new products to potential customers. In particular, when customers' individual profiles and goals are automatically matched with available products, the product advising process is faster and more efficient. Given that insurance products are inherently complex, artificial intelligence can provide customized illustrations of how products work and help customers identify their insurance needs, especially when it comes to products with multiple components, such as endowment life insurance.

Buying an insurance policy

The insurance policy needs to be processed and analyzed at the time of purchase and before being submitted to underwriting. This process can take hours or even days if done manually. Artificial intelligence allows you to automate order processing by taking over various types of data and fact checking, such as fraud detection and credit analysis. As a result, the customer gets a better experience with the company by speeding up the buying process.

Contract Maintenance

While an insurance policy is in effect, chatbots can provide customer support around the clock. Moreover, policy adjustments, such as portfolio diversification and risk profiling, can be done automatically using algorithms based on artificial intelligence.

Underwriting and loss adjustment

Given the large volumes and relevance of data on economic, demographic, natural and market conditions, there is great potential for more accurate risk determination. Artificial intelligence supports the process of interpreting risk data so that actuaries have state-of-the-art models for effective risk management.

Insurers' investment in artificial intelligence increased eightfold in 2017. Asian countries, primarily Japan, were again leading the trend. In 2018, Japanese insurance company Mitsui Sumitomo will assign algorithms 90% of the administrative work of the sales department. Mitsui Sumitomo's sales department has already begun using algorithms for routine tasks, including responding to customer and counterparty inquiries, processing insurance applications, sorting letters and other administrative tasks.

Japan's largest financial group, Mitsubishi UFJ Financial Group, automates 9,500 workplaces, including 30 percent of the staff of the group's main subsidiary, Tokyo-Mitsubishi UFJ Bank.

As a consequence, highly qualified professionals will pay more attention to important tasks, and routine work will be delegated to software. Such professionals will have more time, and with it their value on the labor market will grow.