The Future of Risk Management: Adopting Artificial Intelligence in Financial Services, a Risk Management Perspective

A DISCUSSION OVER DINNER HOSTED BY VALERIE DU PREEZ (DUPRO ADVISORY & ACTUARTECH) AND ISAAC ALFON (CRESCENDO ADVISORS)

It was a beautiful Autumn’s day and we were joined by experts across the insurance and finance industry, including Chief Risk Officers, Fintech Professionals, Risk Management Consultants, Actuaries, and Data Scientists.  We discussed over a roundtable dinner, the future of Risk Management in financial services in a world that is increasingly impacted and characterised by AI technologies.  This note summarises the main outcomes of the roundtable discussion.

We are seeing that the current landscape and the availability of AI technologies as well as the competitive pressures within financial services are supporting the justification for adopting AI technologies.

We discussed the slow pace of progress of adoption of AI technologies based on a paper shared with attendees (discussion paper available on request).  A useful reference point was the historical experience of slow adoption of other general-purpose technologies, for example, electricity.  This suggests that the main hurdle for adoption of AI technologies is being able to adapt the business to make full use of new technologies.

We also explored other practical explanations for the observed low level of adoption.  We identified:

• A limited understanding and appreciation of AI technologies beyond technical circles in senior management, businesses, and risk functions
• A prevailing technology perspective rather than a business perspective
• Concerns about bias and handling difficult ethical challenges, e.g. privacy, vulnerable customers, dampening trust on AI technologies
• In some cases, (still) limited availability and accessibility of relevant data in usable formats and data compatibility issues

Interestingly, concerns about the extent to which AI technologies might challenge risk pooling as the essence of insurance, and thereby undermine existing business models, were also raised.

Overall, this should lead to an openness to rethink business models, the role of risk management and, in particular, the role the actuary can play.  This means understanding the practical use of AI technologies in risk management.  We aimed to vision an insurance company in 2030 that makes efficient use of AI technologies and then tease out from that which aspects of risk management and governance across the business would need to change to move towards that vision.

The main features, of the impact on a future risk management function, that emerged from the discussion are summarised below.

• First line, business delivery will be characterised by automation with streamlined processes and controls. Risks associated with business processes and human error are minimised.
• Dashboards are available and updated on a real-time basis so that risk management goes beyond compliance and regulatory-driven checklists to support decision-making.
• This allows the risk function to go back to enterprise risk management and focus on risk aggregation to drive a strategic approach to decision making within organisations.
• An important outcome of the above would be trust in the strategic direction of the business, which in turn will enhance cooperation and collaboration across the business.

Interestingly, participants shared the view that these features are consistent with today’s best practice in risk management.  This was somewhat surprising but with hindsight perhaps not entirely.  It suggests that there is a transition characterised by the adoption of AI through the application of today’s best practice in risk management.

To request the full discussion paper exploring the pace of adoption of AI, or to share your thoughts on the future of risk management with us please contact info@actuartech.com.

With special thanks to article authors Valerie du Preez valeriedupreez@dupro.co.uk and Isaac Alfon isaac.alfon@crescendo-erm.com and for his valuable contribution during the event.