How AI for Commercial Insurance Differs From Rules-Based Approaches

Artificial Intelligence - July 2 2019

As the global insurance industry continues to experience disruption and digital transformation, it is imperative that commercial insurance brokers and carriers select the right technologies that can best meet their needs now and in the future. Digital transformation can mean different things to different organizations, and the technologies to support digital transformation can have different applications and use cases. Choosing the right technology that fits seamlessly into existing infrastructure and meets the needs of the business is essential.

Two technologies that are often compared with each other are Artificial Intelligence (AI) and Optical Character Recognition, commonly known as OCR. Use cases for both solutions exist in today’s commercial insurance market, however, it’s important to understand their relative capabilities and shortcomings.

The AI Advantage for Data Extraction

There are many different types of Artificial Intelligence. In this post, we will examine the advantages of Natural Language Processing (NLP) and Machine Learning. AI allows insurance companies to perform functions such as data parsing, similarity, categorization, topic analysis, keyword/key-phrase extraction, and summarization. In addition to advanced features, the following micro-level document intelligence features are also possible in NLP: extracting facts, extracting named entities, identifying entity relationships and metadata fields. For instance, in commercial insurance, it is common for each insurance company to use their own words, phrases, and descriptions which results in a diverse variation in data capture. And to compound the problem, there is no standardization of forms. For example, some forms may use the term “company name” while others may use variations such as “Name of Company”, “Corporate Entity Name”, “Company”, “Corporate Name” etc. You get the point.

Natural Language Processing can automatically recognize and identify the data points through semantic analysis. In linguistics, semantic analysis is the process of relating syntactic structures, from the levels of phrases, clauses, sentences, and paragraphs. Semantic analysis begins with the relationship between individual words.

For instance, let’s take a commercial insurance submission form that includes the address as street, city, state, country and zip code all in one line. OCR would give you this text as one sentence whereas Natural Language Processing is able to extract the data and automatically associate the data with Zip Code, Street Address, City and State.

You might be asking yourself, “Why is this important?” Well, if you’re a commercial insurance carrier that wants to triage your submissions by leveraging semantic analysis and setting up rules for auto-declining, auto-routing or prioritizing submissions, NLP can identify insurance-specific data points that enable carriers to identify the best business to write in seconds. By automatically extracting the data points based on SIC code, if the carrier is not interested in writing business for forestry, they can set up a rule “if SIC code is 800” auto-decline and send an automatic email including the reason they are not interested in writing the business.

Insurance companies looking to power up their operational efficiencies by extracting meaning and insights from their data beyond simply using OCR to create templates of documents should be considering an investment in AI.

What is Named Entity Recognition?

Named Entity Recognition (NER) is an information extraction technique that refers to the process of identifying and classifying key elements from text into pre-defined categories and revealing direct relationships. NER helps transform unstructured data like insurance policies, binders, submissions, quotes, binders, loss run reports, statements of value, etc., to data that is structured and can be used in downstream systems like rating engines, policy management, agency and broker management systems.

NER is the ability to identify entities or data points like people, places, organizations and numerical expressions like dates, times, currency amounts, telephone numbers, postal codes, etc. Purpose-built AI solutions for commercial insurance can recognize insurance-specific named entities such as premiums, endorsements, loss run statements, coverages, etc. Being able to take unstructured data and extract the data to provide a useful view of the data in a structured format, provides underwriters with greater access to data for better pricing and risk management.

To Rule or Not to Rule?

OCR is a rule-based or template-based solution. It requires rules and templates in order to be able to capture the necessary data. Unlike AI that can be trained and automatically learns over time, OCR requires rules to be set up continuously. This means a long and expensive setup process as each individual data extract requires a new rule. Unfortunately, OCR cannot be totally automated as there will always be more rules to be set up. For example, for a policy, for every field, there needs to be an individual rule set up. OCR is slow and time-consuming compared to AI.

From an insurance document standpoint, OCR is typically used for the following documents:

  • Certificate of Liability Insurance
  • Certificate of Property Insurance
  • Policy Declaration
  • Endorsement
  • Notice of Third Party’s Coverage
  • Inland Marine Declaration
  • Loss Payable Clause
  • Additional Interest Schedule
  • Notice of Cancellation
  • Lienholder Notification

AI can be used across all the insurance documents outlined above plus the following:

  • Policies
  • Applications or submissions
  • Quotes
  • Binders
  • Statements of Value
  • Loss Run Reports

AI is OCR on Steroids

Purpose-built AI solutions for commercial insurance can read documents hundreds of times faster than a human, and with significantly greater accuracy. The ability to instantly recognize and understand more than 500 insurance-specific data points has significant advantages over both OCR and human knowledge workers. More intelligent, flexible and powerful than OCR, AI also integrates seamlessly with your existing systems to streamline workflows and unlock previously inaccessible business insights in your unstructured data.

Get the AI Technology Guide

For more information on the capabilities and benefits of deploying an AI solution, download your complimentary copy of the technology guide: AI in Commercial Insurance.

Get the Tech Guide: AI in Commercial Insurance

Browse different topics

Recent Posts