As AI upends financial services, could governance become an advantage?

Each month at WP, we offer a slate of articles and content pieces that go deep on a particular topic. This month, we're focusing on AI in wealth management.

Information scarcity has been one of the key determinants of success in financial services. Firms gain alpha through information and pricing opportunities are found when current prices don’t match the realities presented by new information. A recently published paper from the CFA Institute argues that this dynamic could be about to change. Mona Naqvi, Managing Director of the Research & Policy Center of the CFA Institute and the author of that paper, sees the rise of artificial intelligence as potentially transformative to the way financial services operates as an industry, because it will make analysis “faster, cheaper, and more widely available.”

Naqvi broke down the issues highlighted in her paper, laying out some of where they might apply to the day to day work of Canadian financial advisors. She outlined the layers of risk that the rise of AI might pose to financial services as a whole, explaining where firms may be forced to focus on governance policies to address them. She highlighted, as well, the role of advisors and consumers in the wider social conversation about the role AI could and should play in the way people invest.

“When I look at most of the AI narrative and commentary, a lot of it is discussing the productivity gains. They're talking about it as a technology rollout rather than a governance challenge. And I do believe that this is primarily a governance challenge,” Naqvi says.

Naqvi’s three areas of AI risk

For Naqvi, the rise of AI in financial services opens up three areas of risk that she believes appropriate governance can help manage. The first is model risk, the risk that the AI model’s integrity has not been fully tested, challenged, or understood. Without those tests, the model could make mistakes that might result in serious financial losses for investors.

In her paper Naqvi also notes the risk of what she calls “cognitive convergence.” The idea is that with more widely available and almost commodified vendors, data, and models, market participants might think and act similarly. That could create a herding effect which would introduce a kind of concentration risk.

Finally, Naqvi highlights accountability risk, where AI-supported decision making obfuscates human responsibility for investment outcomes. She stresses that investors still need to know who is responsible for the decisions being made with their money.

The governance solutions for these problems are not yet fully outlined, as the path forward on these risks is not yet fully known. Naqvi says that institutions and firms can begin to address these areas of risk, but their policy will depend on how AI adoption continues. She argues, though, that the Canadian market may have something of a ‘second-mover advantage.’ Because other markets, namely the US, are so far ahead of Canada in terms of AI adoption, some of the hard lessons that will be learned Stateside can be baked into Canadian governance frameworks.

How to govern for AI risk

While Naqvi emphasizes that there are a host of potential pathways for AI adoption in financial services, she believes that certain core principals need to be held to for appropriate governance to work. That begins with an acknowledgement that the same human and firm responsibilities apply even when decision making or information gathering is being outsourced to third-party providers and non-human AI agents.

Human oversight will remain essential, she says. Investment managers and fund issuers will have to understand the limits of the AI models they’re using and ensure that they remain responsible for any final analysis. Documentation and traceability of decision making will also be crucial, as will communication between advisors and clients. AI may make investment management feel more like a black box to clients, which Naqvi believes will make advisors’ jobs all the more critical. Softer skills like communication will be more important than ever, she says. Ethical and transparent communication will be key.

Naqvi notes that while much of the AI conversation has been dominated by a narrative of inevitability, there is a great deal that is not yet decided about how AI will be used. Consumers and citizens will play a role in those decisions and advisors will see how consumer preferences shape AI adoption in financial services.

While the path for AI adoption may change and the acuity of each particular risk that Naqvi identifies may way and wane, she argues that the core goal of human responsibility and communication should be placed centrally in any governance framework. She believes that as this trend progresses, advisors can serve as key advocates for the kind of governance that might give them a trust advantage.

“I think [advisors] need to just continue to press the point that AI may change the inputs into advice, but it should not change the duty of care. And at least legally, that still continues to rest with the individual. So fiduciary responsibility continues to sit with the individual. And that's why the ethics and professional integrity remain as important as ever,” Naqvi says. “The specific steps to take will vary by organization, will vary by jurisdiction, because this will be a messy transition with uneven adoption. I'm reluctant to give a single silver bullet for how to do that. But, I think keeping professional integrity and the fiduciary responsibility at the heart and at the core of everything that's done is a good North Star.”