the silent barrier: how limited AI understanding is crippling system modernisation in specialty insurance

We’ve seen the specialty insurance market coming under enormous pressure to modernise, driven by complex risks, rising costs, increased regulatory scrutiny and client expectations for faster, more accurate service. Yet despite significant investment in transformation programmes, many organisations find themselves stuck. Systems modernisation runs over budget, digital initiatives stall and promised efficiencies never materialise.

One of the least discussed but most damaging blockers?
A limited organisational understanding of artificial intelligence and its role in modernisation.

This isn’t about lacking deep technical knowledge of how AI works - few organisations need that level of expertise. We have seen first-hand that the real challenge is the absence of a shared, practical understanding of what AI is, what it isn’t and how it fits into the broader digital ecosystem across underwriting, claims, operations, actuarial science and distribution.

The result is a widening gap between modernisation ambitions and actual progress. 

where the lack of AI understanding hurts modernisation

Misaligned Expectations About What AI Can Solve

Some leadership teams assume AI will fix foundational issues such as data quality, workflow design and systems integration. But AI applied on top of broken or fragmented systems risks magnifying the dysfunction.

Without a clear understanding of AI’s real functions, teams:
· Set unrealistic milestones
· Approve the wrong projects
· Underestimate dependencies
· Lose confidence when early pilots underperform

Poor Technology Decisions

Executives may invest in tools with “AI” branding while neglecting the unglamorous but essential building blocks of modernisation:

APIs
· Standard data models
· Automation foundations
· Cloud infrastructure
· Data governance

These elements don’t just support AI they enable future value. Without them, AI initiatives become isolated proof-of-concepts that never scale.

Language Barriers Between Business and Technology Teams

We frequently hear terms like “AI model drift,” “technical debt,” “data pipelines” and “governance frameworks”, but these terms mean little to most insurance teams. Likewise, many technologists lack the nuance of delegated authority, exposure modelling or the realities of bordereaux chaos.

The result?Meetings where no one realises they are agreeing to completely different things.

Fear-Based Decision-Making

When people don’t understand AI, they default to one of two extremes:

· Over trust (“AI will replace our manual processes in months”)
· Over fear (“AI is too risky to touch; we could breach regulations”)

Both stances prevent progress and paralyse decision-making.

how to know if your organisation is caught in this trap

You’re likely stuck in this AI-modernisation deadlock if any of the following sounds familiar: 

1. Your modernisation programme feels perpetually ‘in planning.’ Workshops keep happening, but execution never gains traction.
2. AI pilots look impressive in demos but never go live. No one is confident about implementing them.
3. Business stakeholders say: “We’ll deal with data later.” This is a major red flag, data is fundamental to the modernisation journey, not just a task to bolt on at the end.
4. Technology teams repeatedly explain the same terms. If you’ve heard “APIs”, “cloud migration”, or “data model” for a year and still aren’t sure what they practically mean for underwriting or claims, your organisation is stuck.
5. There’s confusion about the difference between automation and AI. If these two words are used interchangeably, strategic misalignment is almost guaranteed.
6. Different teams use different language to describe the same initiative.
   For example:
   · The business says, “We need to streamline submissions.”
   · IT hears “We need an AI triaging model.”
   · Operations thinks it means “We need a new workflow tool.”

This mismatch cripples progress more than any technology challenge.