Search for:

When artificial intelligence (AI) emerged as a viable option for use in financial services, experts hailed its transformative potential while also warning of the risks it could pose to markets, institutions and consumers. Since then, AI has permeated much of the financial industry and significantly changed some aspects of how its businesses operate. Not as radical transformation of the industry, however, but as a steady trickle of new technologies and applications across the sector. Considerable risks also remain, as does uncertainty about regulatory approaches to AI.

We spoke with four experts to explore how AI use is likely to evolve in the financial industry in the years to come. They are optimistic that banks, insurers, and other institutions will find new ways of deriving value from it, including in back-end operations. At the same time, they have concerns about weak governance of AI, as well as the implications that incumbents’ control of data will have on competition. Familiarity with the technology may now be widespread, but much remains unclear about its ultimate impact on the industry.

Use cases today and tomorrow

Unsurprisingly for an emerging technology field, AI use has not met all the expectations that financial industry experts initially had for it. In investment management, for example, share trading has benefitted less from AI than some early predictions. One factor, according to Peter Hafez, chief data scientist of RavenPack, a provider of news and analytics tools to the financial industry, is that signal construction algorithms used in trading were already sophisticated, and AI models have not moved construction on very much.

Portfolio optimization, by contrast, has benefitted enormously from AI, says Hafez. “Some big players are generating hundreds of thousands of alpha streams out of thousands of data sets. By applying machine learning and AI techniques they reduce or cluster all of these alphas and produce an overall portfolio strategy out of them.”

Whereas financial services providers have thus far mainly sought to generate value from AI in client-facing areas such as chatbots and in investment management, in the future Hafez expects companies to apply it more heavily to improving administrative processes and controls. “For example,” Hafez says, “I expect to see greater application of AI techniques to areas such as risk management and fraud detection.”

Fraud management is one of the most promising AI use cases in the financial industry, says Bradford Newman, partner at Baker McKenzie and a leader of its AI and ML Practice.

Mastering unstructured data will bring big improvements in fraud management, says Andrew Rear, an insurance technology investor and adviser. “Fraud detection models are very good today at working with large sets of structured data but still struggle with large unstructured data sets. That should soon start to change,” Rear says.

More broadly, Newman believes AI will permeate every facet of financial services provision within the next five years. “The industry is nowhere close to realizing the benefits that AI can bring to its companies and to consumers.”

The good citizen AI

According to Nicole Schepanek, founder and managing partner of Aureus Capital, a private equity firm, improvements in model accuracy will do as much or more than new use cases to help companies generate new value from AI.

Other experts are similarly hopeful about AI use to combat climate change. Newman sees it being applied more widely in the coming years to impact investing, in which fund managers seek to gain social or environmental benefits from their investments, as well as financial gains. Developing algorithms that identify portfolio investment targets with high-potential carbon reduction technologies is a prime area for AI use, Newman says.

Hafez believes that the industry can further sustainability by using AI to highlight the incidence of “greenwashing” by companies, where they exaggerate progress in reaching ESG (environmental, social and governance) milestones. However, Hafez says, this will take some time: “Good algorithms require high quality, consistent data, and that’s currently far from assured in ESG reporting.”

Lingering constraints

Data challenges—revolving around quality and availability—and scarcity of specialist talent are likely to remain constraints on AI value-generation in the financial as in other industries. And algorithm explainability continues to be elusive. “Companies are finding ways to increase the visibility of what’s driving decisions in their models,” says Hafez, “but finding the right balance between explainability and model performance is still a significant challenge for most.”

Limited explainability remains a major legal risk for institutions that, according to Newman, is exacerbated by weak AI governance.

A step on the way to remedying this is for all large financial institutions to appoint a chief AI officer, advocates Newman.

For Schepanek, unambitious management is the primary constraint on AI value-generation. “The industry will fail to exploit new use cases if managers focus more on the potential risks they pose, rather than the growth they can bring.”

Regulation, competition and innovation

Thus far, financial regulators have largely refrained from intervening directly in institutions’ use of AI. Data privacy rules have been the main point of reference, as well as the normal systems and controls around technology risk that licensed financial institutions are expected to have in place. Could that change? Several regulators have published discussion papers on AI use (such as those of the UK and Germany), high-level principles (Hong Kong, Singapore, and others) or even proposed regulations (the European Union).

Andrew Rear, like our other experts, believes direct intervention is coming, although he thinks it will be later rather than sooner:

A less likely target of regulatory intervention, but probably more problematic for AI-enabled competition and innovation, is the dominant access that large financial institutions are likely to enjoy to the data that feeds models and drives their improvement. “Only a small handful of dominant market players will have the capital to invest not just in collecting and analyzing that data, but in actually doing something with it,” says Newman.

If that stifles financial industry innovation, in the long run AI will prove less disruptive and less impactful than many investors, policymakers, regulators and other industry stakeholders have hoped it would be.

Expert interviews were conducted with:


Bradford Newman is a litigation partner resident in Baker McKenzie's Palo Alto Office and Chair of the North America Trade Secrets Practice. According to Chambers USA, Brad is a "recognized authority on trade secrets cases" who "is valued for his tenacious, intelligent and thoughtful approach to trade secrets matters." Bradford regularly serves as lead trial counsel in cases with potential eight and nine-figure liability, and has successfully litigated (both prosecuting and defending) a broad spectrum of trade secrets cases in state and federal courts throughout the country. He routinely advises and represents the world's leading technology, banking, professional service, manufacturing and commerce companies in connection with their most significant data protection and trade secret matters. Bradford is the author of Protecting Intellectual Property in the Age of Employee Mobility: Forms and Analysis, a comprehensive treatise published by ALM that offers authoritative guidance on legal risks and practical steps companies can take to protect their IP and remedy IP theft.

Write A Comment