Across Asia’s insurance industry, artificial intelligence has moved rapidly from buzzword to business plan. Major carriers tout AI‑driven underwriting, pilot chatbots and run “AI 101” training for staff. On the surface, it looks like the sector is modernising fast.
A new AI proficiency report by Section AI, based on a survey of 5,000 knowledge workers at large firms in the US, UK and Canada, suggests much of this progress is cosmetic. Employees are using AI, but mainly for simple tasks that barely touch core insurance work.
The bar for proficiency has quietly shifted. In 2025, it meant knowing what AI is, avoiding data leaks and writing a decent prompt. In 2026, it means something tougher: weaving AI into real, value‑adding tasks every week – in underwriting, claims, servicing, operations and compliance.
Most workers are not there. The study finds:
Overall, 97% of workers are using AI poorly or not at all. A quarter say they save no time with it. Forty% say they would be fine never using AI again.
The core problem is not prompting skill but a “use case desert”. Workers broadly know how to use a large language model; they do not know what to use it for in their role.
The numbers are telling:
As a result, less than a third of workers say they save four or more hours a week with AI, even though most organisations would need savings above ten hours per employee to move the dial on costs.
The top work uses are revealing: search replacement (14.1%), draft generation (9.6%), grammar and tone fixes (5.7%) and basic data analysis (3.8%). Task and process automation sits at just 1.6%. In all, 59% of use cases are basic task help; over a quarter have no meaningful role in broader workflows; only 2% are considered advanced.
Meanwhile, insurers are investing. Across respondents, 63% say their company has an AI policy, 50% have access to an AI tool and 44% receive AI training. Those with a company AI strategy are 1.6 times more proficient; those with tool access or training are 1.5 times more proficient; those whose managers expect AI use are 2.6 times more proficient.
Yet employees who have undergone AI training still score only 40 out of 100 on proficiency. Most remain “experimenters”, not practitioners.
A striking perception gap has opened between leaders and staff. Among C‑suite respondents, 81% believe there is a clear, actionable AI policy; just 28% of individual contributors agree. Eighty% of executives say tools have a clear access process, versus 39% of front‑line staff. Nearly half of leaders see “widespread adoption with open sharing of best practices”; only 8% of individual contributors say the same.
Individual contributors – claims handlers, service agents, back‑office staff – are also the least supported: only 32% have clear access to AI tools, 27% have received company training and 7% are reimbursed for AI tools, well below managers and executives. Manager support for their AI use is down 11% since May 2025.
For the Asian insurance industry, the message is clear.
If AI success is measured only by policies, tools and training completion, leaders will keep over‑estimating their progress. To move beyond hype, employers will need to:
The technology is already on the desks of many insurance workers. The harder task, now, is to bring AI into the centre of how underwriting, claims and service actually run – and to be honest about how far there is still to go.
