In their recent Policy Options article, professors Fenwick McKelvey, Reza Rajabiun, and Brenda McPhail criticize the CRTC for its December 2021 decision approving Bell Canada’s call-blocking initiative, claiming it failed to conduct a thorough review of that system’s artificial intelligence (AI) component. Their argument will have left your readers with a false understanding of the CRTC’s decision.

The authors would have us believe that the CRTC did not understand the importance and complexity of AI and, as a result, deferred to Bell’s “corporate interest” in approving our application with little oversight. In this telling, the CRTC missed an opportunity to establish a strong set of principles to govern the future use of AI in telecommunications.

This is an incorrect interpretation of events.

There is no question that Bell’s application presented the CRTC with a major public policy issue: how to ensure interveners had sufficient understanding of the application to comment meaningfully on it, while at the same time preserving the confidentiality of our approach to prevent fraudulent and scam callers circumventing it, and continuing to defraud Canadians of millions of dollars annually.

The solution was to ask interveners to sign a non-disclosure agreement (NDA), vetted and approved by the CRTC. They would then have been able to see the essential components of our application and comment meaningfully upon it, while promising to only use that information for the purpose of the CRTC proceeding and to destroy it once the proceeding was over.

Professors McKelvey and Rajabiun were twice provided with the opportunity to review our blocking proposal in detail, including its underlying AI component, in order to comment on it. They declined on both occasions.

The CRTC should be commended for balancing the competing interests at play in this proceeding, as it must under the Telecommunications Act, by adopting the NDA approach. Parties that signed the NDA were able to provide meaningful comment and the CRTC itself was able to examine the AI involved. It is regrettable that the authors chose not to take advantage of the same opportunity.