AI's Role Shifts to Augmenting Underwriters
The conversation around AI in underwriting is maturing from replacement to augmentation. Industry leaders are now focusing on using AI to automate repeatable tasks, preserving expert judgment for complex cases. Meanwhile, large language models are being deployed to help adjusters more accurately parse complex policy language, speeding up outcomes.
The shift to AI augmentation is accelerating, with Accenture noting that up to 40% of an underwriter's time is spent on administrative, non-core tasks. AI platforms can now automate repetitive jobs like data extraction from documents, pre-filling applications from public records, and flagging missing information, freeing up underwriters for complex risk analysis. This operational efficiency is driving significant investment. US insurers are projected to more than double their AI spending, from 8% to 20% of IT budgets in the next three to five years. Globally, the AI in insurance market is forecast to grow from $8.13 billion in 2024 to over $141 billion by 2034. Leading the charge are specialized InsurTech firms like GradientAI, Zesty.ai, and Appian, which offer AI-powered platforms for enhanced risk assessment. These tools leverage machine learning and vast datasets, including telematics and geospatial data, to improve loss ratio predictions by up to 15% compared to traditional methods. In claims processing, Large Language Models (LLMs) are unlocking the value of unstructured text data found in adjusters' notes and reports. This allows for a more complete narrative of a claim, aiding in faster, more accurate settlements and the early detection of potentially high-cost claims. The next wave of innovation involves "agentic AI," which can act autonomously on an underwriter's behalf. These AI agents can monitor news about a potential insured, generate and send emails to brokers for missing information, and even draft declination letters for submissions that don't fit the insurer's risk appetite. However, the adoption of these "black box" systems creates significant regulatory challenges. Insurers face the risk of embedding historical biases into their algorithms, which could lead to discriminatory outcomes against protected classes, creating new compliance and reputational risks.
