AI will leave claims handlers with only the hard decisions

The most important figure in Oxbow Partners’ new report on AI in claims has nothing to do with cost. I think it is the share of a handler’s working day spent actually deciding claims, and it points to a training problem the industry has barely started on.

According to Insurance Times, the consultancy’s Future of AI in Claims report, released on 5 November, predicts handlers could eventually spend 60 to 70 per cent of their time on adjudication and decision-making, up from 40 to 60 per cent today. First notification of loss and triage would shrink sharply.

If AI leaves handlers with nothing but the hard decisions, insurers need to start hiring and training for judgement now, before the easy work disappears.

In my view, good claims people learn their trade on routine files: the clear liability call, or the claim that turns out to be exactly what it looked like, which is how you learn to spot the one that is not.

Strip that work out and the apprenticeship goes with it.

Oxbow’s own pitch, on its report page, is that the bigger prize is claims intelligence: feeding what every claim teaches back into underwriting and pricing. Its partner Miqdaad Versi told Insurance Times that claims teams “sit on some of the richest data in insurance, yet most of it has never been properly used.” He also warned that none of it works without strong data foundations. I agree on both counts.

The report says human judgement stays essential for complex or novel claims with little history behind them. My worry sits exactly there. Those are the claims a less experienced handler, moved up early because the triage work vanished, is most likely to get wrong, and they are also the expensive ones.

Oxbow writes with specialty insurers in mind. My reading is that the same logic applies to any motor or home claims team planning its next few years of automation.

Where will your next generation of senior handlers learn their judgement once AI has taken the easy claims?