Artificial intelligence is transforming insurance, but a quieter disruption is unfolding.
According to EY, 50% of the current insurance workforce will retire within the next 15 years. In the UK, 64% of brokers identify recruiting young talent as a strategic issue. At the same time, junior underwriters now require stronger data & digital skills to work with AI assistants despite the foundational tasks that build deep underwriting judgement being increasingly automated.
As experienced underwriters retire, the market risks losing not just talent, but the judgement behind decades of underwriting decisions and wisdom. Half of today’s underwriting expertise, the ability to weigh incomplete submissions, interpret ambiguous data, and read a broker’s tone, could vanish unless it’s captured and shared.
Underwriting has always been an art as much as a science. Data informs the process, but human pattern-recognition defines it. Yet the insights that matter most, what to trust, what to challenge, when to walk away, rarely live in structured systems. They sit in emails, notes, and memories. When those memories retire, so does the competitive edge.
At the same time, opportunity data is going to waste. In marine and specialty lines, 20–30% of broker submissions never receive a full review. Each one contains learning potential about new risks, geographies, and client behaviours, yet these declined or unquoted cases are almost never analysed. The industry’s collective intelligence is eroding twice over: experience walking out, insight left unused.
AI can change that. For the first time, the industry has technology capable of preserving underwriting judgement at scale — and making it reusable. Modern underwriting platforms can capture reasoning, flag recurring patterns, and turn every decision, quote, decline or defer, into structured knowledge. Each interaction becomes feedback for the next generation of underwriters, turning experience into institutional memory. Utilising AI-first platforms integrating data extraction, enrichment and explainable decisioning in one place, the industry gains the tools to preserve judgement at scale rather than lose it to fragmentation.
AI models today are trained on trillions of words giving them a scale of exposure that spans centuries of human knowledge. But without the judgment of experienced underwriters to shape that learning, scale becomes noise, not wisdom.
Those who act now by embedding knowledge capture, explainable AI and human-in-the-loop design will future-proof their underwriting advantage. The shift will not come from incremental tooling upgrades, but from adopting AI-first platforms that treat underwriting knowledge as a strategic asset. The greatest risk isn’t the one we misprice. It’s the wisdom we forget to keep.
