“In the midst of chaos, there is also opportunity.” — Sun Tzu

We are living in an epoch of profound transformation and paradox, reminiscent of Charles Dickens’ “best of times” and “worst of times.” This dichotomy deeply resonates in the insurance sector as we stand on the brink of a brave new world, one shaped by the transformative potential of artificial intelligence (AI) and large language models (LLMs).

In this era of digital disruption, the danger of inertia for property/casualty insurers is real and imminent. The urgency for businesses to innovate and reinvent is palpable. The challenge is not about whether to act but how best to navigate the uncharted territory of AI and LLMs, such as ChatGPT.

This piece endeavors to articulate the “What to Do” segment of Celent’s flagship report, “ChatGPT and Other Large Language Models: P/C Insurance Edition,” setting forth a pragmatic road map for C-suite executives and operational leaders.

These innovative technologies are rapidly reshaping the insurance landscape, presenting an era of unprecedented opportunity. They promise to redefine various aspects of the insurance ecosystem, spanning from underwriting to product development, claims management, marketing, actuarial tasks, analytics and beyond.

Navigating the Pace of Change

The accelerated adoption of AI-driven technologies in the insurance industry highlights a profound shift. Echoing Jack Welch’s famous quote: “If the rate of change on the outside exceeds the rate of change on the inside, the end is near.” One example of this rapid change is ChatGPT, an AI-based application developed by OpenAI. According to UBS, ChatGPT is the fastest-growing app of all time, reaching over 100 million users in just two months, compared to TikTok which took nine months and Instagram which took 2.5 years to achieve the same user base. This illustrates the rapid pace of AI adoption and the urgent need for insurance companies to adapt quickly and innovate.

Where should they begin? How should they move forward?

Starting at the top, the chief executive spot, this article extracts some basic action steps for leaders from Celent’s recent report in the sections that follow.

Fostering an AI-Inclusive Corporate Culture and Vision (CEOs, CSOs)

The CEOs and chief strategy officers have pivotal roles in forming steering committees to decipher the broader implications of augmented intelligence on business dynamics, operational models and competitive standing. Engaging with integral stakeholders on AI governance frameworks and regulatory safeguards is an equally crucial role for CEOs and CSOs.

These leaders should also evaluate the need for cultural and work environment transformations. What drastic shifts are necessary to empower employees with generative AI? How can initiatives prioritize upskilling? One approach is to reflect on the comprehensive skills employees may require to harness LLMs effectively, including data literacy and the ability to formulate incisive questions.

Restructuring the Business Model With AI at the Helm (Business, Channel Leaders)

Heads of business and channel leaders should actively champion AI tools. One recommended step is to crowdsource use cases, especially those from the younger workforce who are likely early adopters of tools like ChatGPT. It’s important to evaluate these use cases based on several factors such as potential for revenue growth, cost-cutting opportunities, ease of implementation and the expected return on investment.

Streamlining the Operating Model (COOs)

For chief operating officers, envisioning the overarching impact of integrating LLMs into the middle and back office is paramount. For key use cases: chart an implementation road map encompassing integration with existing systems, personnel training and rigorous testing. It’s also important to reassess any existing LLMs and scrutinize the current AI governance structure to ensure fairness, privacy, security, explainability and transparency.

Establishing a Robust Technological Infrastructure (CIOs, Data Leads)

For CIOs and heads of Data Analytics, scrutinizing the technology underpinning LLMs, including their performance, accuracy and reliability, is critical. Anticipating and addressing potential obstacles during the implementation of tools like ChatGPT, such as data privacy, security issues and seamless integration with existing systems, is an imperative. Moreover, enhancing the organization’s technical expertise and computational resources is necessary to effectively access the ChatGPT API and train it using proprietary data. This could be a key factor in providing a potential competitive edge.

Proceed With Caution

As we navigate the exhilarating yet challenging terrain of technological innovation, deploying LLMs like ChatGPT warrants careful consideration. Despite their potential, these tools are nascent and undergoing rapid evolution. Companies eyeing LLMs are possibly still deciphering optimal deployment and regulatory strategies. As a result, policies and practices around the use of LLMs may vary considerably within the insurance industry and across different sectors.

OpenAI is ushering LLMs into the global arena, marking new territory for many insurers. Some firms may opt to prohibit the usage of ChatGPT and other LLMs due to potential bias, ethical considerations or other factors until they gain a more profound understanding. Additionally, the propensity of LLMs to produce erroneous outputs with a deceptive air of confidence, a phenomenon known ashallucination, further underscores the need for rigorous testing and validation protocols before any deployment.

Furthermore, the introduction of novel tools like ChatGPT undoubtedly opens the door for additional cyber risks for companies. Insurers should maintain a heightened vigilance in monitoring a host of issues that could potentially affect cyber coverage. Given the novelty and rapid development of LLMs, questions around regulatory implications remain largely nebulous at present but are likely to emerge as a significant factor for the insurance industry in the near future.

The regulatory landscape for these technologies is still taking shape and presents its own set of complexities. Different jurisdictions adopt varying stances on AI regulation. For instance, while the EU leans toward a more precautionary approach encompassing both high-risk and lower-risk AI systems, the U.S. fosters a more innovation-friendly environment, primarily focusing on regulating high-risk AI applications. This dichotomy creates a challenging situation for insurers, especially those operating across different regulatory regimes, as they try to harness the benefits of AI while staying compliant with diverse and evolving regulatory guidelines.

As stewards in this domain, the responsibility falls on us to not only embrace these technologies but also proactively engage in shaping their regulatory landscape. Collaborative efforts with regulatory bodies, other insurance firms and technology providers are vital to ensure a comprehensive and adaptive regulatory framework. This is instrumental in mitigating potential risks, ensuring ethical use, and fully leveraging the transformative potential of AI and LLMs for superior business outcomes.

Currently there are possible risk mitigation strategies. For instance, the performance of LLMs may be enhanced by integrating them with carriers’ internal models, which have been trained on their proprietary data. The objective of this is to broaden the range of domain-specific topics by leveraging a more extensive language comprehension, thereby enhancing accuracy levels.

The vast potential of AI and LLMs for P/C insurers is unquestionable. As we navigate this transformative digital era, the call for definitive action resonates with growing intensity. The onus is on us not only to embrace these technologies but also to guide their trajectory, leveraging their benefits for superior business outcomes.