Artificial Intelligence
Oct 08, 2018

The reality of AI in claims

Recently, a claims director from a large European insurer told me, "I've already looked at AI, and it's not going to work for us – we are too complex." If there is one takeaway from this article it is this: believe the hype. AI will impact your claims department, especially complex processes where conventional automation technologies fail.

AI refers to the ability of computers to exhibit human-like intelligence (we call it augmented human intelligence), for example, solving a problem without the use of software-coded instructions. Unlike human claims adjusters, AI has the ability to quickly comb through large amounts of unstructured data, understand and interpret it, identify trends, and draw accurate conclusions that take into account historic loss data. An AI-enabled claims department means claims adjusters can spend 95% of their time optimizing indemnity and customer service. Meanwhile, AI will be making sense of data and reports, filing relevant data into the claims system, and highlighting trends.

Your customers are adopting AI, too

Industries and sectors insured by P&C carriers are increasingly investing in AI. From loan underwriting that relies on natural language processing to read financial statements, to aerospace manufacturers that use machine learning to analyze huge volumes of flight data to solve maintenance and safety issues, AI is changing how insureds operate. And that, in turn, changes their risk profiles.

As the AI rollout gathers pace, the amount of data generated will be too much for underwriters or claims adjusters to make sense of. This is probably why 30% of CIOs include AI in their top-five technology investments by 2020, according to research from Gartner. You can either put your leakage ratio at risk by not using the data to make better decisions, or you can explore how combining a claims adjuster with AI is more powerful than either of the two alone.

How can AI impact claims handling?

As AI is integrated into claims, it will be able to:

  • Analyze and rank medical providers based on performance metrics from historic claims, including unstructured expense types and the duration of rehab. Unlike conventional automation technology, AI can understand and interpret medical reports, notes, and prescriptions to suggest the right course of action
  • Read contracts and understand liability and obligation clauses. For instance, AI can understand the terms and conditions and fee model in a third-party administrator contracts and match invoices against them
  • Predict the risk of litigation and the outcome of a lawsuit in a way that can be statistically validated by augmenting the research of even specialized and highly context-sensitive case law carried out by paralegals and litigators. AI can highlight and extract content for further analysis, such as win-loss rates, time-to-injunction, the performance of a law firm or the opposing counsel, and more
  • Comb through large amounts of data on drugs and analysis of medical devices, not only to determine the aggregate exposure, but also to highlight volatile lines
  • Make fast-track and straight-through processing more cost effective while improving subrogation identification and identifying leakage trends
  • Make sense of loss documentation in multiple languages and create the necessary scale to drive standardization in segmentation, indexing, and use of templates for smaller European operations, powered by language-neutrality engines and machine learning
  • Structure and process hundreds of bordereaux as part of a treaty, reinsurance, or aggregate process, and feed relevant information into underlying systems
  • Provide building-level damage data that adjusters can inspect from their desks using 3D-enabled photos and videos. AI can determine construction type, detect wear and tear, and estimate damages based on previously approved losses
  • Process the vast amount of data required to support e-trading as it makes its way into more commercial lines

For insurers to take advantage of these innovations, they must tackle back-end legacy issues while also offering customers front-end digital services. Having a digital transformation roadmap will ensure companies adopt technologies in the right sequence, and re-engineer back-end processes to match the desired customer experience.

AI cannot replace claims adjusters, but it can untangle unstructured data from multiple sources – and make sense of it – so that claims adjusters can be more effective. Moreover, it can redefine how insureds see the claims promise fulfilled.

Putting AI to work

This is my advice for claims departments looking to adopt AI:

  1. Begin by focusing on use cases where today's technology can drive efficiencies and smooth the customer journey
  2. Develop hypotheses for each line of business you insure on where and how fast AI implementations are rolled out
  3. Develop a comprehensive digital transformation roadmap that factors in claims complexity, processes, and what you would like to achieve with the additional business intelligence from AI
  4. Experiment with AI technologies to increase organizational comfort and identify enterprise-wide opportunities

About the author

Matthew Madsen

Matthew Madsen

VP and European Lead, Claims Practice

Follow Matthew Madsen on LinkedIn