A version of this article was originally published in Asia Insurance Review.

Generative artificial intelligence (AI) is at the forefront of today’s technology. Insurers and consumers alike are putting bets on it to improve efficiency and productivity.

According to Oliver Wyman Forum’s Global Consumer Sentiment Survey, 62% of respondents have used generative AI in the past three months, with 23% using it for work purposes.

While conventional AI analyses data and uses predetermined algorithms to make predictions or decisions based on the given data, generative AI’s model learns from large amounts of new data to identify the patterns and structures needed to generate new and original content.

Conventional AI has come a long way for the insurance industry. It has upgraded how insurers carry out underwriting, adjudicate and process claims, manage policies, and detect fraud. Chatbots were one of the first use cases, which have since evolved from simple ones based on predefined conversational flows to more advanced versions that use conventional AI technologies to comprehend user questions and respond in real-time. Generative AI with the right controls in place has the potential to bring chatbots to a higher level in mimicking human interactions. 

How insurers are currently utilizing generative AI 

Modernizing and streamlining processes

Insurers in Asia are constantly enhancing their processes, such as through the introduction of assistant chatbots to provide a better customer service experience. Virtual assistants powered by generative AI can further provide financial consultants with real-time data and ideas on customers’ life insurance plans and coverages and allow them to respond to customers’ queries more promptly.

On the non-life insurance front, an example would be the streamlining of auto insurance distributions and claim processes. Typical processes would include managing administrative tasks, such as quote generation, policy renewals, and e-policy notifications. Streamlining these processes help companies save costs, and free up employees’ time for more complex and higher value work.

Hyper-personalization to improve efficiency

A leading Chinese insurer equipped with cutting-edge technologies has recently deepened customer personalization by introducing smart interactive voice responses to identify and segregate customers, and a user-interaction mining platform to identify intentions and emotions through Natural Language Processing algorithms. With these latest technologies, the insurer hopes to improve efficiency and bring user experience to the next level.

Similarly, a Malaysia-based insurer wanted to create more accurate customer segments. As a result, individualized homepage banners were introduced to improve onsite experience by recommending products based on users’ previous browsing history. Generative AI can be used to further analyze data that provides businesses with deeper customer understanding and develop personalized marketing materials tailored to each customer’s interests, preferences, and behaviors.

Claims management and detection of fraud

Claims management and fraud detection is another area where generative AI has taken a more important role and will potentially help insurers automate claim processes that can be tedious, challenging, and time-consuming.

Generative AI helps to speed up the identification of fraud by checking past cases and incidents. This automation removes the need for employees to check claims manually and enables them to focus on more important decision-making tasks. A notable example is the use of Large Language Models to extract data of claims and develop models to identify the specific causes of loss across claims to improve underwriting. This allows companies to save time and resources and, at the same time, manage their loss ratios.

The challenges in using generative AI for the insurance industry

1. Regulations for the responsible use of AI

As early as 2018, Singapore developed a FEAT (fairness, ethics, accountability, and transparency) guidance for financial institutions and more recently released a Veritas Toolkit 2.0 to support them in integrating the principles into their internal risk governance. This initiative will help examine the risks and opportunities of generative AI for the financial sector by bringing together the data resources and domain expertise of the banking sector with the top AI companies’ state-of-the-art technologies and advanced algorithms.

The Cyberspace Administration of China (CAC) wants to ensure that the chatbot technology will be “reliable and controllable” and thus require all generative AI algorithms and products to undergo security testing and review by the CAC before they can be released.

In 2018, the Ministry of Economy, Trade, and Industry of Japan published the contract guidelines on Utilization of AI of Japan. More recently, the National Institute of Advanced Industrial Science and Technology released the third edition of Machine Learning Quality Management Guideline.

2. trusting ai to address complex human sentiments

AI-powered chatbots often help resolve large volumes of data. However, chatbots today lack the ability to address complex human sentiments and emotions. Generative AI is also prone to hallucinations, a process where the models produce an incorrect answer with “confidence” because the model is simply filling in what they think is the most likely next word. Trust remains a challenge with many insurers still testing generative AI tools, and only a few using them in production.

Customer service centers, for example, can be equipped with chatbots to tackle large volumes of requests, while human touch can be used in parallel to bridge the need for personalized interactions. This combination will allow for better customer experience.

3. safeguarding the data privacy of customers

The current biggest concern when using generative AI is data privacy, where unauthorized access and misuse of personal data can have legal implications. Generative AI collects personal data and could potentially generate sensitive information that violates data privacy regulations in many countries. Most recently, a class action lawsuit was filed against OpenAI for violating copyrights and data privacy of millions of internet users without their notice or consent to train its AI tools.

According to InRule Technology’s survey, customers are not excited to meet ChatGPT in their insurance journey, with nearly three in five (59%) saying they tend to distrust or fully distrust generative AI. In addition, most respondents (70%) said they still prefer to interact with a human. The distrust lies with the concern of generative AI possibly making errors in insurance quotes or claims, hence producing a significant risk for businesses that set up automation around ChatGPT.

While generative AI presents an opportunity for insurers to reinvent their value chain, there are still challenges and considerations to be considered.

Read the original article here (paywall).