Underwriters are at the heart of any insurance business. Their ability to assess risk accurately based on a wide variety of information is critical to product innovation, effective pricing, and ultimately whether a business can make a profit in an increasingly competitive market. Consequently, insurance companies are always striving to optimise their underwriting processes, in a bid to get ahead. In today’s market, that increasingly means leveraging artificial intelligence (AI).
Underwriting has already advanced enormously in recent years; as technology and modelling have developed, the process has become increasingly sophisticated and digitised, with underwriters now sifting through rafts of proprietary and third-party information. But as data sources have increased, underwriters face the challenge of how to process and analyse this ‘big data’ as accurately and quickly as possible, to keep up with customer expectations regarding service and convenience. This is where AI comes in.
AI technologies including natural language processing (NLP), large language models (LLMs), generative AI, and machine learning, promise to solve this problem, with the ability to ingest, process, and analyse huge volumes of unstructured data significantly faster and more accurately than humans. With Hiscox recently claiming AI has enabled it to cut underwriting times from three days to three minutes, it is no surprise 77% of insurance decision-makers are at some stage of adopting AI in their value chain.
How is artificial intelligence being used in insurance underwriting?
It’s now more than two years since ChatGPT burst onto the scene, and since then, it has been joined by numerous competitors, including Claude, Google Gemini, and Microsoft Copilot. In the insurance sector, these and other foundation models are being harnessed to create AI-powered underwriting systems, solving challenges such as manual data entry, fragmented data sources, and incomplete or unstandardised risk assessments. AI tools can even help evaluate a person’s risk profile by scanning huge amounts of structured and unstructured data sources, such as social media, financial, and public records, with the ability to make decisions in a matter of minutes in some cases.
London-based broker, Hiscox, recently announced it has collaborated with Google Cloud’s Gemini large language model to assess risks, generate pricing, and draft emails to brokers, accelerating the underwriting process significantly. AIG is also harnessing generative AI to improve data quality in underwriting, with the Chairman and CEO Peter Zaffino claiming it has improved data collection and accuracy rates from around 75% to more than 90%. And QBE recently launched a Cyber Underwriting AI Assistant, which has driven a 65% reduction in review times, by performing initial submission reviews.
What are the benefits of artificial intelligence in insurance underwriting?
It’s no exaggeration to say AI is revolutionising underwriting efficiency. According to a McKinsey report of more than 50 insurance leaders, more than half of the respondents say gen AI could lead to productivity gains of 10 to 20 percent, premium growth of 1.5 to 3.0 percent, and improvement in technical results by 1.5 to 3.0. Here are a few reasons why:
- Increased efficiency and accuracy: By automating key underwriting tasks, such as collecting, processing, and analysing structured and unstructured data, AI can significantly speed up the underwriting process. NLP can be used to fill out application forms and submission documents, reducing manual data entry. Meanwhile image recognition can be used to assess photographs, speeding up risk assessments. These efficiency savings in turn help to boost customer satisfaction, increase sales, and reduce administration costs.
- Risk assessment and management: AI algorithms can analyse masses of unstructured data to identify patterns that human underwriters may overlook. It also helps the feedback loop between claims and underwriting by enabling claims data to be factored into underwriting decisions. As a result,companies can develop more accurate and sophisticated risk assessment models, to price policies more accurately, and reduce loss ratios, improving overall risk management strategies.
- Fraud detection and prevention: Detecting and preventing fraud is a significant challenge for underwriting, with research showing that incidences of insurance fraud rose again in 2023. By analysing vast amounts of data, AI can identify suspicious patterns or activities indicative of fraudulent behaviour. This helps insurance companies combat fraud more effectively, ultimately reducing losses and improving the industry’s overall integrity.
- Free up underwriters: Finally, by automating many of the most time-consuming and mundane tasks, AI can also free up human underwriters to focus on higher value, strategic work, such as more complex risk assessments, strategic decision-making, and developing innovative new products.
Is artificial intelligence a panacea for insurance underwriting?
Introducing AI into insurance underwriting isn’t without its challenges and businesses must be careful not to place too much trust in AI tools, which are only as good as the data they digest. If algorithms are trained inadequately or fed inaccurate or incomplete data then there is the danger of ending up with a biased or discriminatory system, which is doing the opposite of what it is designed for.
As AI is still a relatively new concept, companies must also be careful not to lose customer trust, particularly in terms of data privacy and security and complying with their regulatory responsibilities. Many consumers still like to deal with human beings and know human oversight is involved, so firms must be wary of moving too quickly. Human underwriters will still be crucial as a ‘human-in-the-loop’ to oversee both the inputs and outputs of an AI system, ensure that decisions are fair and transparent, and retain the human touch in understanding the needs and behaviours of customers. As Generali’s Farjzaneh Traheritabar said at the MoneyNext Conference 2024: “AI doesn’t replace jobs, it replaces tasks. Underwriters who adopt AI are poised (to replace) those who don’t.”
The future of artificial intelligence in insurance underwriting
AI is changing the face of insurance underwriting and innovation is happening constantly. But, while the potential benefits are enormous, implementing AI involves a significant transition for insurance companies, who need to rethink tried and tested processes and adopt new ways of working. It can’t happen overnight, and firms must first put in place the right digital foundations for collecting and organising data, upskilling the workforce, and managing the cultural change involved in adopting new technologies. For this reason, the sooner insurance companies can start experimenting with AI the better, so they don’t get left behind as the technology evolves.
How to implement AI in insurance underwriting
What should insurance companies think about when implementing AI in underwriting?
- Start small: Rather than trying to run before you can walk, start by applying AI to a small, clearly defined underwriting problem so you can begin to understand the technology and how it can help improve how you work.
- Set your objectives: What are you trying to achieve from using AI on this specific problem? For example, do you want to drive faster decisions, or reduce your loss ratio for a particular product? This will help focus your efforts and investment.
- Define clear KPIs: Implement clear targets and benchmarks for what you want to achieve, to avoid confusion and ensure stakeholders are realistic about what good looks like.
- Risk management: Also consider the risks involved, and how you can start mitigating those early on in the process, including communicating effectively with staff about the proposed changes and how they will be impacted. .
- Digitise the relevant data and processes: AI relies on having the right data to work with, so assess what data and infrastructure are required to deliver that. This could involve implementing software like Insly to fully digitise your underwriting function and other areas of your workflow and support a more holistic data strategy.
- Invest in specialist skills, or external expertise: If you plan to make AI a core part of your underwriting strategy going forward, then you will benefit enormously from hiring individuals with the right expertise, upskilling your workforce, or bringing in external specialists.
- Maximise off-the-shelf tools: Don’t feel you need to start from scratch, as there are now a host of ready-built insurance AI tools out there, such as Insly’s FormFlow, to solve a who variety of challenges. It could save you a lot of costly development time and ensure you see results much faster.
- Test, and test again: As with any new technology, implementing AI isn’t an exact science, so trial and error will be key to working out how it can best help you reach your objectives.
If you’re taking the first steps in your digital journey, then Insly makes it simple, with no-code modular insurance software covering the whole insurance lifecycle, from distribution and underwriting, right through to claims. Laying the foundations for a digital, automated future.
