A year has passed since the launch of ChatGPT, and many InsurTech start-ups are still exploring how to effectively leverage the new technology. Artificial intelligence (AI) has become the norm, and differentiation will come from how it is creatively applied to solve practical issues.
The recent Plug and Play Winter Summit in Sunnyvale, CA, fully displayed this phenomenon. Plug and Play connects start-ups with large corporations through industry-focused accelerator programs. The Winter Summit InsurTech line-up unveiled three key AI themes that we can expect to see progress over the coming year.
AI as a co-pilot
At the beginning of the AI frenzy, employees across all industries worried they would be replaced by new technology, and for good reason. Many corporations were openly considering implementing AI solutions that would decrease the need for human workers, thus cutting costs and increasing efficiency.
As InsurTech start-ups explore applying AI to the insurance agent workforce, they are shifting away from viewing AI as a potential replacement for agents, instead viewing AI as a tool to enhance and streamline the work that agents do. For example, using AI solutions to reduce processing time and help tackle highly complex claims.
It is not uncommon to hear InsurTech AI solutions being pitched as a “co-pilot” for agents, likely inspired by the similarly named Microsoft Copilot, an AI chatbot. Microsoft also recently made waves in the InsurTech industry by partnering with SCOR and ReMark to host a collaborative hackathon exploring generative AI and reinsurance.
As Pooja Shah, Senior Associate at Avanta Ventures, put it, “Ultimately, despite all of the hopes that we have for GenAI and AI in general, at the moment there still is a human involved. So, it’s not going to be adjuster vs. AI, it’s going to be adjuster with AI vs. just a normal adjuster.” Shah also noted that the key to reducing employee AI concerns is quality training that shows that AI is not something to fear. Rather, it can be a valuable tool for agents that may give them an advantage over others in their field.
The current limitations of AI capabilities are not the only factor keeping humans involved. There are also emerging regulations such as the AI Act and ethical concerns aiming to prevent the elimination of our jobs. Check out SCOR’s AI and the Future of Insurance publication within the Expert Views series for more information on this topic.
Integrating with legacy systems
Throughout the InsurTech evolution, the community has come to learn the complexities of legacy systems. The overall mindset has shifted from attempting to replace these systems to integrating into them, and AI is no exception. AI start-ups have learnt from the past and are focused on incorporating AI into established tech stacks.
It is imperative that AI tools are easy to integrate into existing systems and do not require a complete overhaul. No matter how valuable an AI solution may be, if it is deemed too complex to integrate with internal tech and use with ease, it will not be considered. For many organisations, the time and resource costs that a system replacement requires outweigh the benefits a new system would offer.
There are also security advantages to integrating into legacy systems instead of trying to replace them. Companies may be hesitant to export all their data to an unfamiliar system, especially since regulations that govern AI data security are in their infancy. Instead, startups can provide value if their AI tools integrate into what already exists, allowing companies to keep their data private and build on their own infrastructure without having to give that data to someone else’s model.
Growing interest in multimodality
Large Language Models (LLMs) have been the foundation for many InsurTech start-ups. However, multimodality is considered by some to be the next leap forward for artificial intelligence solutions. While LLMs process text and perform language-related tasks, multimodal AI can intake many data types, including image, text, and speech, to understand behaviours and patterns in larger contexts.
When applied, LLMs can help underwriters understand long, complexly-written site assessment reports and speed up administrative tasks by identifying what is important and answering specific questions. However, multimodal AI solutions go beyond just describing what happened to also understand why it happened and use that to predict future risks.
The concept of multimodal AI in the insurance space is still young and a bit nebulous, but founders are confident that it can help professionals understand behaviours and patterns of behaviour in larger contexts, providing just as much or even more value than the LLMs we’ve seen.
"The word LLM - Large Language Models - is getting outdated,” said Dr. Ivan Poupyrev, CEO and founder of Archetype AI. “As we move into the physical world, there will be even more separation from the LLM because we will be adding time dimensions, spatial dimensions, and behavioural dimensions." If start-ups manage to perfect the multimodal solutions they are working on, the insurance industry has a lot to gain.
As we kick off 2024, it is helpful to reflect on what we learned about AI in 2023 and consider the trends being led by the start-up community that have the potential to become the industry norm.
At ReMark, we are focused on working with the latest technologies to help insurers enhance the experience of their policyholders, enabling them to attract and retain new business. Our Digital Solutions, developed combining our unique technical expertise with SCOR’s comprehensive data and analytical power, cover the entire consumer journey, from purchase to claim.
Find out how our Digital Solutions can help you to meet your customers’ expectations today and check out SCOR’s AI and the Future of Insurance for more information on the current landscape of AI in insurance.