Utilizing Artificial Intelligence for Anticipatory Risk Control within Workers' Compensation
In the rapidly evolving insurance industry, a significant shift is underway, driven by the advent of generative artificial intelligence (AI) technologies. This transformation, as seen in the work of W. R. Berkley Corporation, is revolutionising claims processing and risk management.
A recent study has shed light on the potential benefits of this technology. It was found that a staggering 15% of all causes of injury in workers' comp claims are missing or coded as miscellaneous. This highlights the need for a more efficient and accurate method of processing claims data.
Generative AI models, excelling at processing short passages of text and following simple instructions, present a promising solution. They can rapidly discover insights from previously difficult-to-use sources of data, such as scanned documents and notes taken by claims adjusters.
Unstructured data, such as claim notes, accident descriptions, scanned documents, images, audio, and video, is an often overlooked and valuable source of insurance data. By processing this unstructured data quickly and cost-effectively, generative AI models can lead to key insights much faster than traditional methods.
The study, conducted on nearly 4 million claims, revealed that more than 30% of claims did not have valid job class coding, and 25% lacked a free-text occupational description. To address this, generative AI was used to derive the occupation of the injured employee and further classify their occupation, allowing for the analysis of trends and derivation of insights specific to the particular job functions.
This approach led to significant improvements. A large university experiment saw a 75% reduction in the number of poorly coded injuries and identified corrections or further clarifications to the injury coding for 30% of the injuries. The study also identified that 7% of the descriptions of the injury captured by claims adjusters do not contain an adequate description of the event. Improving the accuracy of coded injury information across a population of claims enabled the discovery of additional insights that were not evident in the unenhanced data.
However, it's important to note that a balanced approach that combines AI efficiency with expert judgment is essential for optimising claims handling and risk management strategies. Insurers must exercise caution and ensure that AI-generated insights are carefully reviewed and validated to avoid misguided conclusions and poor decisions.
The utilization of generative AI in claims processing represents a fundamental shift in how insurers can approach risk management. By automating complex tasks in underwriting, claims processes, and risk assessment, insurers can anticipate risks more effectively and improve internal operations. As generative AI continues to evolve, its role in transforming the insurance industry will undoubtedly expand.
Care must be taken when working with unstructured insurance data to ensure compliance with regulatory, data privacy, and information security policies. This is a crucial aspect that insurers must consider as they embrace this technology.
In conclusion, the integration of generative AI into the insurance industry is a promising development. It offers the potential to streamline processes, improve accuracy, and uncover valuable insights that were previously hidden in unstructured data. As the technology continues to evolve, its impact on the industry is expected to be significant.
