The rapid embedding of AI across superannuation and insurance has raced well ahead of bespoke regulation, and the industry’s best defence against a “heavy hand” from Canberra is to police itself, according to Colonial First State director of insurance Craig Harrison. 

“I think it’s going to take time for the regulation to catch up with the pace of all that is happening,” Harrison told the Investment Magazine Insurance in Super Summit. “I do hope that we can do this in a way that we don’t need that heavy hand of regulation to come in and try and steer things in a different direction. [We can achieve that] by keeping the member at the centre and making sure that we’re getting the right outcome for the member in the circumstances. 

“Particularly in something like the claims process, if we allowed the AI agents to take over the decisioning on those really complex claims – having sat across so many complex claims committees at life insurers and now on the fund side, those claims that are truly in the middle of the grey [zone], they need that human intervention and protocol. 

“[It’s about] keeping those guardrails in place and making sure we’re not letting it go into that territory where regulators are feeling the need to step in and do something differently.” 
 
To Zurich Financial Services head of group insurance Darren Wickham, a lot of the thinking about guardrails on AI is already being done – and not just by financial regulators. Prime Minister Anthony Albanese recently gave a speech on the appropriate and ethical use of AI, while Pope Leo XIV devoted his first encyclical, Magnifica Humanitas, to safeguarding the human person in the age of artificial intelligence. 

“We talk about human in the loop, and that’s important for trust. This is a new technology, so we have to have trust in the process,” Wickham said. “Although it was interesting – [CoreData founder] Andrew Inwood said that sometimes people prefer not to have a human in the loop; sometimes they would prefer a process which was fast. So it’s balancing that and balancing the quality of the outcomes. 

“I think with [claim] declines it’s much more important to have a human in the loop than an accept. The other thing is empathy – at claim time, human connection is very, very important.” 

For Aware Super head of product and insurance David Evans, there’s been “enough communication and expectation setting” from the regulators already for funds and insurers to have confidence about the standards their use of AI needs to meet, pointing to ASIC’s 2024 report on AI governance and more recent guidance from APRA. 

“I think the key message there was the importance of governance keeping up with – or staying ahead of – any innovation in this space,” Evans said. 
 
“We’re a profit-to-member fund, so I think in terms of the use of AI from a member perspective, it’s around delivering better member outcomes and being clear about the problem you’re trying to solve. There’s a lot of existing regulatory obligations on financial institutions and super funds that, regardless of the tool you’re using, continue to apply. 
 
“The other key stakeholder group around the use of AI is also our people – the staff within all of our organisations, and our ability to make sure that people within your organisation are going along on that journey as well.” 

And while the productivity gains from AI have so far mostly accrued inside the organisations deploying it, UniSuper chief marketing and growth officer Dani Murrie said the dividend is already flowing through to members – starting with claims. 

“We have certain agents that do certain tasks that enable the claims assessors to operate more efficiently. The decision can be made more quickly, and that helps members. The dividend is flowing to members because of efficiency initiatives,” Murrie said. 

“In terms of our claims handling, where we are using the agents – for instance, medical reports – we’re able to synthesise those far more quickly, and that then informs and helps the claims assessor with the decision, and our end-to-end process there is faster.” 

But Murrie warned that whatever efficiencies AI delivers, responsibility for the decisions it touches can’t be handed over with it. 

“Technology clearly can support processes, make them more efficient and productive. But accountability just cannot be outsourced – and so when I follow that bouncing ball, a human needs to be accountable for the final decision, and therefore in the loop, today.”