Testing ChatGPT: Evaluating Its Perspective on Insurance Industry Technology Trends

In this blog, one of our insurance experts asks ChatGPT about industry trends and evaluates the results.

The insurance industry is starting to find more and more use cases for emerging technologies, allowing InsurTech to settle into its own as targeted solutions within the value chain mature.

AI, as an example, has gone from simply a discipline to being heavily integrated into many industry solutions. As AI tools gravitate to the mainstream, the industry is collectively learning how to incorporate them responsibly.

A common theme in the ongoing debate about what AI models can and can’t do is the degree to which it requires human participation and the role of a specialist in the process.

Given this background, I took the latest AI trend, ChatGPT, and asked it for its perspective on insurance industry technology trends. In this blog, we’ll take the output and apply a human specialist’s (that’s me!) perspective on it.

ChatGPT’s Insights About Insurance

First, a little context. ChatGPT is not the first model of its kind, but it is one of the first widely available AI-powered chatbots. It is similar to what search engines did in the mid to late 90s in that it takes available technology and applies it to massive amounts of data. It requires significant infrastructure to pull this off, and the model to commercialize it fully is still in flux.

Combine the significant dataset with the fact AI algorithms are constantly advancing and improving, and you have a very powerful tool. Imagine a well-trained scholar having time to read everything on the internet, analyze it and formulate a thesis accordingly – basically, that’s ChatGPT.

So, here is what ChatGPT had to say about insurance trends.

1. Big data analytics: With the help of big data analytics, insurers can make data-driven decisions, optimize risk selection and pricing, and improve customer experience.
2. On-demand insurance: Insurers offer on-demand insurance policies allowing customers to purchase insurance for a specific period or activity, such as rental cars or vacation rentals.
3. Mobile apps: Insurers are developing mobile apps that allow customers to manage their policies, file claims and communicate with customer service.
4. Chatbots and virtual assistants: Companies use chatbots and virtual assistants to improve customer service with instant responses to customer queries and requests. They can also help customers file claims and manage their policies.
5. Artificial intelligence (AI) and machine learning (ML): Insurers are working with AI and ML to automate insurance processes and improve the accuracy of risk assessments. They are also using AI to enhance customer service through the above-mentioned chatbots and virtual assistants.
6. Internet of Things (IoT): Insurance providers use IoT devices such as smart-home devices, wearables and telematics to collect data that assesses risk and personalizes policies.
7. Blockchain: Blockchain technology creates more secure and transparent insurance contracts and claims processing.
8. Usage-based insurance (UBI): UBI is insurance that bases premiums on how much the policyholder uses the insured item, such as a car or home. This type of insurance is made possible by telematics and IoT devices that track usage.
9. Digital claims processing: Digital claims processing uses AI and automation to speed up the claims process and reduce the time and cost of processing claims.
10. Cybersecurity insurance: As cyber threats become more sophisticated, insurers offer cybersecurity insurance to help businesses and individuals protect against cyber-attacks.
11. Personalized policies: With the help of AI and IoT, insurers can offer personalized policies tailored to each individual’s needs and behaviors.

So, how did it do? Not bad, but these “trends” are a little lagging, which makes sense given the way ChatGPT works. Let’s evaluate the results. First, let’s look at what ChatGPT got right.

What ChatGPT Got Right About Insurance

Most of the trends are on point and certainly at the forefront of the industry. I was happy to see the model understood its own value and put AI right at the top of the list, although it could be biased. Similarly, blockchain use cases, UBI and IoT are also certainly relevant. We’ve seen a steady flow of industry solutions focus and thrive using all these technologies.

ChatGPT also managed to hone in on a couple of areas of the value chain, such as digital claims, personalized policies and cyber.

What Didn’t Belong

The inclusion of mobile apps on the list didn’t fit – it’s not exactly a trend anymore. While we see new ways to leverage mobile apps to interact with consumers and agents, this is not an emerging trend. Today, employees, customers and business partners demand the ability to interact with businesses using mobile apps.

What Did ChatGPT Miss?

One insurance trend noticeably absent from the list is embedded insurance. This is probably the newest trend that would make most industry lists. Interestingly, ChatGPT identified a few specific use cases but not others. For example, it could have just as easily mentioned automated underwriting as digital claims.

Embedded insurance allows insurance carriers to offer insurance tailored to customers’ situations when and where they need it. As the industry progresses toward specialization, we’ll continue to see trends involving sophisticated marketing and outreach.

Conclusion

Overall, I found ChatGPT very useful for quickly getting a breakdown of some insurance trends. The tool only took a few seconds and provided a serviceable list. However, its insights were behind, and it only surfaced established, stagnated or mainstream industry trends.

The use case is interesting because an emerging trend is often hard to extract from massive amounts of data. In fairness, a language-based model isn’t necessarily intended to perform that way.

This exercise shows that, like industry trends, many of the insurance applications of AI still require specialists who can leverage the convenience of an AI model but not fully rely on its conclusion.