How AI is reshaping the P&C (re)insurance market
Data is the lifeblood of insurance, and AI can radically improve the effectiveness of data analysis. Given which, the insurance sector has been surprisingly slow to adopt AI.
Part of the explanation may be that insurance remains a culturally conservative industry with a long history of technological late adoption. But things are starting to change.
Arguably, the moment when the world finally woke up to the power of AI occurred late last year when OpenAI’s user-friendly natural-language AI tool ChatGPT exploded across the media, alerting insurance C-suites more fully to the broader strategic opportunities presented by AI.
AI can process data at a scale and speed no human could match, rapidly picking out potentially significant patterns that even the most observant of analysts could easily miss. Harnessing the analytical power of AI enables (re)insurers to save both time and money and to operate more profitably. It also helps carriers provide more tailored and responsive customer service.
A recent survey carried out for the Central Bank of Ireland found that 10% of insurance and reinsurance businesses are already using AI within their underwriting and claims processes — a figure that is likely to rise significantly within the next three to five years.
Risk and Reward
The main limitations on the rising use of AI were identified in the same survey as the concerns about data management, cyber risk, privacy and security, along with a skill shortage in safely and effectively using AI in an insurance environment.
Conversely, the most compelling opportunities cited by survey respondents clustered around the ability to operate more efficiently, reduce costs and to deliver an improved level of customer experience and engagement.
With its ability to identify and interpret potentially significant patterns in the data, AI can also play a valuable role in detecting and predicting fraud. Combining AI and machine learning to flag up suspicious or concerning patterns — in near-real-time — maximises claims professionals’ ability to make a timely intervention.
With the right programming skills, AI can be trained to ingest data — and extract valuable insights — from an almost limitless array of sources, encompassing everything from incoming emails, scanned documents to satellite data, aggregated online data (either free-to-access or paid-for), smart devices, vehicle ‘black boxes’, wearables and bespoke remote sensors set up to record temperatures, windspeed, water levels, or pretty much any other metric, insurance underwriters or claims people might want to know about.
Garbage In, Garbage Out
The key is setting AI models up, to factor in the extent to which decision-makers, human or automated, can afford to rely on each data source selected for inclusion. AI-based models are no exception to the time-honoured rule that if you put garbage in, garbage is what you can expect to get out. But insurers’ increasingly sophisticated tech strategies are massively extending (re)insurers’ ability to make intelligent allowance when drawing on uneven data inputs. AI can play a part in helping (re)insurers save time and effort by identifying, early on, whether or not a particular data source is likely to repay the costs entailed in constructing a viable feed from it.
There is a great deal of fascinating work being done with AI in the underwriting space right now, but in this article I want to focus on what AI and machine learning can do for claims teams. Getting claims right - and doing so efficiently and cost-effectively - matters for many reasons, not least from the perspective of bottom-line protection, but also, crucially, in terms of customer satisfaction.
Whether we’re talking about private motor insurance or a major loss in the oil and gas sector, a claim remains the moment of truth for any (re)insurance product. Handling claims professionally, punctually and fairly can make the difference between consolidating or acrimoniously terminating even the best-established customer relationship. Contentious discussions about how a claim came about and whether - and to what extent - it is covered can derail a previously harmonious relationship. But AI can help dispel the grey areas that give rise to such differences of opinion or interpretation.
For example, insurtech businesses have begun working with technology for capturing imagery of buildings and construction sites over time and using AI-powered modelling to evaluate the extent of damage following an incident that could result in insurance claims.
With climate change running out of control, extreme weather events are becoming increasingly common. Predicting where and when they are likely to occur is a significant challenge for insurers, but so too is responding appropriately when they do. Leading (re)insurance organisations are now using AI-powered damage assessment tools that use machine learning algorithms to evaluate aerial imagery in the wake of an event to deliver an unprecedentedly rapid analysis of where insured properties will have sustained damage. This enables triage to commence before claims have even been reported.
These kinds of AI-driven technologies can help insurers assess post-event losses faster and more accurately, thus reducing the scope for contention. The sooner liability can be accepted (or rejected) and any appropriate interim or final payments find their way to insured parties, the easier it will be to provide clarity and keep costs down for all parties concerned.
Since ChatGPT broke through late last year, burgeoning interest among P&C (re)insurance executives has dramatically accelerated. But there remains a degree of cultural resistance to the prospect of an increased role for artificial intelligence in the insurance sector. In addition to the cultural conservatism noted above, there are real concerns among some insurance professionals that technology could threaten their future employment prospects. In reality, increased use of AI is more likely to threaten businesses that are slow to adopt, than individual employees within successful businesses. As with previous technological innovations, the main effect of AI will be to enable skilled claims people to perform more effectively and efficiently - and to get more done in less time.
More material impediments to wider AI adoption include, those around the risk of outsourcing decisions requiring human sensitivity or the fulfilment of a legal or regulatory duty to a non-human entity. Clearly defining, let alone resolving all such issues, remains a work in progress. But solutions will be found, and ways found to ensure that ultimate responsibility remains with a responsible human decision-maker. (Re)insurers seeking to move to an AI-powered future under their own steam can also encounter significant challenges in sourcing the human expertise they need. But, again, there are ways of sourcing all the necessary skills without the burden of attempting to hire in-house against the backdrop of a highly competitive jobs market.
Right now, our industry stands on the brink of a make-or-break evolutionary leap. The (re)insurers who are best able to access the benefits of AI without incurring excessive costs or experiencing excessive disruption in the process stand to gain significant competitive advantage. Those who dither, or misjudge their AI investment strategy, will pay a corresponding penalty. With deep experience of both the P&C (re)insurance market and AI, DOCOsoft is ideally placed to help carriers make cost-effective use of this transformative new technology.