OSC grapples with promise, risks of AI

Artificial intelligence is already used widely in the Canadian capital markets — including the advisory business — highlighting the growing risk of tech-generated advice that’s biased and flawed.

The Ontario Securities Commission (OSC) published a report, developed alongside Ernst & Young LLP, examining the state of AI in the capital markets. The report found that AI adoption is at an “intermediate” stage.

“Capital market participants are currently using AI to enhance their existing products and services rather than creating new ones,” it said.

The deployment of AI is most mature in supporting the advisory and customer service functions, along with trade surveillance and enhancing operational efficiency and accuracy, the report found.

Firms are using AI “to provide automated customer support functions and support client-facing advisors with information, analysis, and recommendations,” the report said. “This includes improving the quality of data available by detecting patterns, trends, and anomalies with greater precision.”

The technology can also be used to facilitate the segmentation of clients into higher-value and lower-value categories based on revenue and profitability.

“Once this segmentation has been completed, AI can be used to generate customized reports and recommendations for low-touch clients, providing tailored services at a lower cost. AI can also be used to process information about high-touch clients from multiple sources and types of data to generate insights about client profiles, activity and preferences to provide insights for advisors and individuals in sales and trading,” it said.

The technology’s increasing use to facilitate customer service and advisory functions carries its own set of risks, the report said.

“Even where AI tools are used to provide guidance to advisors, there is a fear that advisors may become reliant on these tools to the point where they are unable to recognize that poor or biased advice is being recommended based on flawed data. The extent to which consumers understand these risks is also an important factor to consider,” the report said.

The technology’s use is less advanced in areas such as asset allocation and risk management, it noted.

“While large hedge funds are using AI for research, economic analysis, and order execution, its use for trading and asset allocation are otherwise limited. Risk management shows varying degrees of AI adoption,” the report said.

“More advanced and sophisticated applications of natural language processing are being actively explored but challenges remain. Issues related to data quality, privacy, fairness, explainability and interpretability as well as staffing challenges, retention of developers and changes to cultural and operating models will need to be thoughtfully addressed to fully benefit from AI,” it said.

These challenges also raise issues for regulators — including the potential for malicious use — which will require collaboration to ensure “responsible innovation,” OSC CEO Grant Vingoe said in a release.

To that end, the OSC called for capital market participants to share their feedback on the use of AI in the investment industry through its OSC IdeaHub platform.

James Langton

James is a senior reporter for Advisor.ca and its sister publication, Investment Executive. He has been reporting on regulation, securities law, industry news and more since 1994.