Quarter of insurers using AI for storm risk assessments

For example, advances in AI for catastrophic weather modelling may not have much bearing on general or professional liability insurance. As such, regulatory compliance must be tailored to the specific areas in which AI is applied. “Wielded by a qualified data engineer or data scientist, AI tools offer deeper insights into risk than ever before,” Queen explained. He emphasised that the use of AI in root cause analysis and risk forecasting opens the door to a “golden age” for captive insurance professionals, providing them with better tools to enhance decision-making. “While AI does automate certain tasks, it is more likely to augment human capabilities, allowing employees to focus on higher-value activities rather than replacing jobs entirely,” he said. The rise of AI-as-a-service platforms has made AI more accessible and affordable for firms which, Schmalbach argues, will help demystify the technology and dispel fears surrounding its adoption.

The platform aims to improve the overall underwriting process, helping insurers capture more business and accelerate quote turnaround times. Kanchetti emphasizes the importance of developing regulatory frameworks that promote transparency and fairness in AI-driven processes. The insurance industry, along with regulators, must work together to establish clear guidelines that ensure AI technologies are used ethically and responsibly.

The National Institute of Standards and Technology (NIST) and the proposed Algorithmic Accountability Act in the US are developing frameworks to improve AI system management and governance, focusing on transparency and accuracy. This inclusive approach enhances the acceptance and adoption of AI technologies, promoting equitable outcomes. In an additive model, new weak learners (typically decision trees) are added sequentially, each one improving upon the performance of the previous models by correcting their mistakes (residuals).

Who’s using what in P&C insurance: November 4, 2024

The past year has brought key developments in the use of artificial intelligence in captive insurance. 27% of respondents believed traditional actuarial models to be the most accurate, while 26% favoured stochastic models. While Alan is better known as a health insurance company, the French startup has always tried to offer more than insurance coverage. It now wants to build a super app for all things related to healthcare and announced three new product updates on Tuesday morning, including an AI chatbot that’s vetted by doctors.

The insurance workforce is already accustomed to using low or no code apps, so it’s not a massive leap to see them using AI to augment tasks through AI colleagues and co-pilots. For instance, AI-driven chatbots and virtual assistants are streamlining ChatGPT App customer queries and claims processing, providing quick and CX-friendly responses 24/7. The insurance industry is poised to harness the latest technologies, including artificial intelligence (AI), to innovate and shape the future.

This level of accuracy not only improves profitability for insurers but also makes premiums fairer for customers. Using the data, insurers can better assess risks and increase operational efficiencies. While the advantages of AI in claims settlement are undeniable, the paper also explores the ethical and regulatory considerations that insurers must address as they adopt these technologies. The use of AI raises important questions about data privacy, transparency, and fairness.

Member firms of the KPMG network of independent firms are affiliated with KPMG International. No member firm has any authority to obligate or bind KPMG International or any other member firm vis-à-vis third parties, nor does KPMG International have any such authority to obligate ChatGPT or bind any member firm. Financial services firms are performing better because of technology investments but now they need to fine-tune their digital transformation journeys. Market insights and forward-looking perspectives for financial services leaders and professionals.

Insurers Rapidly Adopt Generative AI Despite Potential Risks

There are also concerns about data security and privacy, as AI systems require vast amounts of sensitive information to function effectively. Moreover, the regulatory landscape around AI in insurance is still evolving, creating uncertainty about how the technology can be implemented within existing legal frameworks. These challenges contribute to doubts about whether AI will ever truly revolutionise the insurance industry in the way that many predict. The respondents who believe AI has already met expectations may represent those who have seen early successes in specific areas. For example, insurtech Lemonade has effectively used AI in customer service, using chatbots to handle routine inquiries and free up human agents for more complex tasks.

Doing so could enable businesses to work faster, more flexibly and develop more sophisticated models in response to the evolving market. As AI technology advances, insurers have the opportunity to refine their customer interactions, making them more intuitive and value-driven. Customer service is evolving with real-time updates, omnichannel communication, paperless and automated documentation, and virtual assistants. This transformation extends beyond the insurance industry, as companies embrace innovation to enhance customer experiences. As AI systems take over repetitive and analytical tasks, the human workforce can shift towards roles that require empathy, ethical judgment, and complex problem-solving.

• Insurers could use AI to accelerate the claims process, simultaneously improving productivity, resolving a longstanding customer pain point and improving access to care.
• Similar to the ‘happy path’ concept, more routine claims can be partially or even fully automated.
• This strategy minimises the need to “reinvent the wheel” for each new application, saving time and resources.
• Insurers also face lengthy implementation timelines, with 58% reporting over five months needed to make rule changes—a timeframe that puts them at a disadvantage in the face of market demands.

For example, AI could help detect and prevent fraudulent claims or offer predictive insights. Seeking partnerships with AI solution providers that integrate with internal apps is a strong insurance bots approach as well. AI is advancing quickly, with breakthroughs now spanning beyond language models to areas like weather forecasting, including hurricane landfall predictions[6].

The ability of AI to process and analyze large datasets will enable insurers to better understand customer behavior, predict future trends, and offer more personalized services. Embracing ecosystems and platforms can help insurers adapt to market changes and even reduce the risk market disruption. The interplay between traditional insurers and InsurTech firms is vital for fostering sector-wide innovation and expanding coverage to underserved segments. Collaboration could also help steer insurance toward a more inclusive, customer-centric, data-driven and tech-enabled future. From a business perspective, there are promising use cases applying LLMs to efficiently analyse and process large documents and datasets powered by advanced natural language processing (NLP) applications. Engineering high-quality data foundations is key to reaping the many future benefits LLMs may offer to drive efficiency across the insurance value chain.

GlobalData’s poll run on Verdict Media sites in Q found that the majority of insurance insiders (60.2%) believe AI has not yet met expectations but think it will eventually. However, 29.6% remain sceptical, doubting that AI will ever live up to the hype, while only 10.2% feel AI has already met the industry’s expectations. GlobalData suggests the slump after 2021 was likely due to the challenging macroeconomic environment around the world, with the cost-of-living crisis, high inflation rates and high energy creating a difficult investing environment. It also notes that insurers “may have also been getting better at creating their own in-house data teams”, which could partly explain the drop-off. According to GlobalData’s Data Analytics in Insurance report, sector M&A investment in the first nine months of this year stood at $5.7bn compared to $1.8bn for all of last year. “Adaptive Insurance’s GridProtect helps businesses quickly get back on their feet with immediate capital that they can decide how to use, helping them navigate the short and long-term impacts of a power outage,” the firm stated.

With this approach, Munich Re is able to determine the predictive robustness of the AI, quantifying, for example, the probability and severity of model underperformance. Herman Kahn, an American futurist, is often credited as one of the pioneers of modern scenario planning. During the 1950s and 1960s, Kahn used scenarios at RAND Corporation and the Hudson Institute to model post-World War II nuclear strategies. Additionally, industry standards from organisations like the National Association of Insurance Commissioners (NAIC) provide oversight and best practices for ethical AI use in insurance.

While traditional AI has already demonstrated its prowess in insurance, the industry is yet to explore generative AI’s full potential, while also keeping track of its emerging risks. At Swiss Re, we have been testing the capabilities of large language models (LLMs) for more than three years. Selected use cases have been deployed to pilot user groups and we expect to deploy them to a broader user base this year. You can foun additiona information about ai customer service and artificial intelligence and NLP. Artificial intelligence (AI), in its present form, has proven invaluable in insurance, providing more accurate data insights, enhancing operational efficiency and fostering innovation.

The rapid advancements in AI, notably generative AI, outpace existing legal structures, prompting a need for updated regulatory measures. Recent initiatives, such as the US President’s executive order, underscore the commitment to safe and secure AI deployment. This order, along with emerging global initiatives, aims to establish accountability and address the challenges posed by AI innovations in the insurance sector.

“Quarterly and annual earnings calls provide a platform to discuss financial results and respond to investor questions. Investor presentations offer a more comprehensive overview of the company’s strategy, performance and outlook,” Guild explained. Effective communication goes a long way in clearly understanding an insured’s business and future potential. This allows for sustainable partners to develop coverage that fits and to work closely with their Claims team to understand the partnership in context. On the policyholder side, transparency empowers individuals to take proactive steps in managing their property risks. By understanding the factors contributing to their risk assessment, policyholders can prioritize mitigation actions effectively, potentially reducing their overall risk profile and minimizing potential losses.

Since risk management is in the very DNA of the insurance business, it is perhaps not a surprise that many insurers feel due diligence will be necessary before embracing a transformative technology like generative AI in insurance. Given these caveats, many applications will necessitate an AI-assisted approach to scenario development. This process includes sense-checking and adjusting scenarios for specific business use cases, as well as translating narratives into measurable business impacts. LLMs should therefore be viewed as tools to assist with the heavy lifting of generating scenario narratives, rather than a turnkey solution.

However, in pursuing these AI-driven innovations, insurers cannot lose sight of the importance of building and maintaining customer trust. In fact, 77% of insurance CEOs said establishing customer trust will have a greater impact on their organization’s success than any specific product or service. This is especially critical given that consumer trust in the insurance industry is already shaky, with trust scores declining 25% since pre-COVID-19. Leading digital product organizations are already leveraging AI to research consumer and user needs, understand product usage, and synthesize customer feedback. For insurers, this translates into delivering not just personalization, but an actual match between customers, their risks, and the insurer’s products. Executives anticipate this AI-powered approach will accelerate product creation in 2025, reducing time to market by 3.6 months and increasing the number of new products launched by 50%.

Legacy data frameworks—originally not aligned with sophisticated AI algorithms—often necessitate major enhancements or complete overhauls to support current AI technologies. Automated systems can quickly assess damage using computer vision, reducing the time it takes to settle claims. The more data insurers can gather and process, the better they can assess risk, calculate premiums, and manage claims.

According to FEMA, businesses that close for just five days due to such events have a 90% chance of never reopening. Noting that these savings can be redirected towards business growth, employee support, and community engagement. By identifying common elements across different use cases, insurers can develop reusable components that expedite AI deployment in new areas. This strategy minimises the need to “reinvent the wheel” for each new application, saving time and resources. If you aren’t yet a client, you can download our complimentary Predictions guides, which cover more of our top predictions for 2025.

As a Data Analytics Lead in the insurance industry, he continues to pioneer new solutions that blend technical prowess with practical business impact. Beyond his work in insurance, Kanchetti is dedicated to mentoring the next generation of data professionals, sharing his knowledge and passion for making data-driven decisions that matter. AI-driven data analytics offers a groundbreaking solution to these long-standing problems. By integrating AI technologies such as machine learning (ML), natural language processing (NLP), and predictive analytics, insurers are now able to automate tasks that were once labor-intensive and repetitive. Similar to the ‘happy path’ concept, more routine claims can be partially or even fully automated. This frees handler resource to deal with more complex cases, improving operational efficiency, as well as  enhancing the customer experience by reducing waiting times and improving the accuracy of decisions.

Regulators are increasingly advocating for enhanced transparency and accuracy in how insurers assess risk,” the survey noted. Traditional actuarial models are considered most accurate by 27% of industry professionals, while 26% favor stochastic models. Their cloud-based software enables insurers to modernise their operations and deliver customer-centric experiences. The offering allows seamless integration of AI models from various industry partners directly into Majesco’s workflows. Today, we are exploring solutions to cover the various ways GenAI could potentially and randomly go wrong.

Yet, many insurers find themselves trapped in “pilot purgatory,” where AI initiatives linger in experimental phases without scaling up or delivering significant returns. AI is playing a pivotal role in enabling safer driving environments, which directly contributes to community wellbeing. For instance, AI systems equipped with telematics can provide drivers with detailed feedback on their driving habits, encouraging safer behavior on the road and potentially reducing accident rates. KPMG combines our multi-disciplinary approach with deep, practical industry knowledge to help clients meet challenges and respond to opportunities.

An overwhelming 90% of insurance executives agree that predictive risk models should be transparent. Adoption of these models varies depending on the specific peril being assessed, ZestyAI reported. For wildfire risk, traditional actuarial models remain the most common tool, used by 54% of insurers. Stochastic models follow at 30%, while AI and machine learning-based models are used by 18% of companies for wildfire risk assessment.

They’re aware that data quality before cloud migration is key to effective AI applications, and that clean, well-organized data is essential for AI to ensure accurate, transparent and fair decision-making. This also links back to regulation as insurers with unstructured or fragmented data will face significant challenges in meeting new legislation and building trust in the market. The swift development of AI has resulted in the increasing integration of AI technology in insurance claims management and insurance underwriting. In certain cases, AI has been used by insurers to streamline administrative work to improve efficiency, especially for day-to-day claims handling.

The data collection and processing required for AI-driven decision-making can lead to potential breaches and misuse of sensitive information. Therefore, insurers must implement comprehensive data security practices to minimise these risks and maintain customer trust. To mitigate these risks, insurers need to ensure full transparency and traceability in their pricing decisions and processes. A new parametric insurance platform, Adaptive Insurance, powered by artificial intelligence (AI) has launched with a mission to change how businesses safeguard against climate risks.

This gap underscores the slow progress in transitioning from traditional systems to advanced technology. Alan recently raised a $193 million funding round at an impressive $4.5 billion valuation. After France, Belgium, and Spain, the company last month announced plans to expand to Canada, where it will be the first new health insurance company in almost 70 years. But given that AI chatbots tend to hallucinate, healthcare professionals may not want to rely on inaccurate information or risk misdiagnosing a patient. This issue has come up in the news lately with AI-based medical transcriptions — eight out of ten audio transcriptions exhibited some level of hallucinated information, according to a study by a University of Michigan researcher.