‘The Trust Gap is a Real One’: Experts Advise Feds to Course Correct on AI
Federal Members of Parliament (MPs) are currently grappling with the complexities of regulating artificial intelligence (AI). This involves weighing the potential threats AI poses to Canadian jobs, cybersecurity, and data sovereignty, alongside its economic opportunities.
Study Launched by the Standing Committee on Industry and Technology
At the behest of the Bloc Québécois, the Standing Committee on Industry and Technology has initiated a study to explore how AI should be regulated, considering a multitude of factors. The issues at hand include:
- The scientific challenge of ensuring AI models are honest.
- The legal question surrounding data sovereignty.
- The political question of AI multilateralism.
- The immediate debate regarding the establishment of a dedicated AI committee to monitor the fast-evolving technology.
Call for Public Consultation
Experts have responded positively to the idea of creating a government AI committee, yet a recurring theme has been the federal government’s failure to engage in adequate public consultation on AI issues. Colin Bennett, a professor emeritus at the University of Victoria, stated, “The trust gap is a real one, and it needs to be closed.”
A recent KPMG study revealed that while 50% of Canadians surveyed approve of or accept the use of AI, nearly 80% express concerns regarding potential negative outcomes. Furthermore, 75% believe that AI regulation is necessary.
Critique of Recent AI Consultations
The “what we heard” report produced by Canada’s AI ministry, which utilized AI to analyze responses, was criticized for lacking genuine insight. Michael Geist, Canada Research Chair in Internet and E-Commerce Law, noted that the consultations felt more like “what we wanted you to think we heard.” He emphasized that “if we’re going to have confidence in these consultations,” they must extend beyond a mere 30-day effort.
Recommendations for Improved Engagement
Bennett and Professor Yoshua Bengio from the University of Montreal advocated for the use of citizen assemblies to gather feedback from Canadians who will be affected by AI technologies. They argue that AI’s pervasive nature necessitates a fundamentally different consultation process.
Data Sovereignty Concerns
Discussions also centered around ensuring Canadian sovereignty over its data. Geist suggested that robust privacy laws should be prioritized, rather than merely focusing on Canadian ownership of companies. He pointed out that smaller, Canadian-owned firms often lack the capacity to provide necessary services at scale, leading to potential vulnerabilities.
Geist elaborated, “As long as a company has connections to a foreign country, such as the United States, Canadian privacy laws may not be sufficient to guarantee their application.”
The Challenges of Autonomous AI
The meeting also addressed concerns regarding “agentic” or autonomous AI. Bengio highlighted that current AI systems are “not reliable and trustworthy” due to their training methods, which aim to imitate human behavior. He added that these systems often remain opaque, making it impossible for companies to guarantee their intended behavior.
Privacy Issues and Current Investigations
Bennett pointed out ongoing investigations by the Office of the Privacy Commissioner into various AI applications, emphasizing the need for a healthy skepticism towards emerging technologies. Notable cases include:
- Investigation of ChatGPT for the non-consensual use of Canadians’ personal data.
- Concerns regarding xAI’s Grok related to the display and sharing of inappropriate images.
- Scraping of images by Clearview AI for facial recognition systems used by law enforcement.
Conclusion: A Call for Multilateralism
The AI regulation study was initiated by the Bloc Québécois and has involved input from a diverse range of experts. Bengio emphasized the importance of multilateralism, arguing that Canada should collaborate with like-minded middle powers to establish AI guidelines that reflect shared values and concerns.
He concluded that Canada must strive to lead in safe, competent AI development to ensure it remains a player on the global stage rather than becoming an afterthought. Alongside national laws and international treaties, it is crucial for researchers to design AI that adheres to legal and moral standards.
