Rethinking AI in Canada: A Public Utility Approach for Generative Technologies

Rethinking AI: The Case for CanGPT as a Public Utility in Canada

As generative artificial intelligence (AI) tools like ChatGPT and Google Gemini continue to revolutionize the digital sphere, the primary discourse around their implications in Canada has predominantly emphasized commercial innovation. However, there’s a pressing question that often goes unexamined: What if AI were developed as a public utility rather than as a commercial service?

Drawing from Canada’s rich history with public service media, such as the CBC and Radio-Canada, we can explore a framework for developing AI that serves the public interest. This approach resonates with the growing calls for an AI policy rooted in public benefit.

The Knowledge Commons: AI’s Foundation

Commercial AI models are largely predicated on the assumption that user-generated content available online can be accessed to train their algorithms. This heavy reliance on global cultural knowledge, which treats the internet as a “knowledge commons,” must be scrutinized.

AI systems would not exist without public data, much of which has been utilized without adequately giving back to the public domain. For instance, Canada played a significant role in the evolution of AI—one notable example being the automated translation efforts that began in the 1980s when Canadian parliamentary transcripts were anonymously shared with IBM to aid early translation algorithms. Imagine if, today, Canada took the initiative once again to develop AI with intentionality and care.

CanGPT: A Vision for Public-Service AI

Countries like Switzerland, Sweden, and the Netherlands are pioneering public AI models to create services for their citizens. Canada is also exploring this avenue with its internal tool, CanChat, but a broader national initiative is necessary.

By building a public-service AI, we could follow the precedent set by earlier technologies like radio and television. Public broadcasters were established to ensure that emerging communication platforms served democratic values. A similar paradigm could benefit AI. The Canadian Parliament could sponsor the development of CanGPT, an AI model that utilizes public domain materials and datasets, harnessing the multilingual archives of CBC/Radio-Canada as a foundational resource.

Establishing Democratic Boundaries for AI

Creating a public AI like CanGPT would invite much-needed discourse about the ethical ramifications and potential limitations of AI technology. Generative AI has already been linked to challenges such as deepfake pornography and other forms of technology-driven harm. Currently, these boundaries are largely dictated by private tech companies, which limits democratic engagement in important decision-making processes.

A publicly governed CanGPT would enable Canadians to engage in discussions regarding the ethical and social framework surrounding AI through democratic institutions, rather than leaving these discussions solely to private companies.

The Importance of a Public Model

Adopting a public AI framework represents a significant departure from the government’s current, infrastructure-heavy approach. Canada’s existing strategy, with its multi-billion dollar investments into large-scale AI initiatives, risks entrenching dependence on American tech firms and undermining our capacity to foster public-interest AI.

Furthermore, a public model could lead to the development of energy-efficient AI systems that run on smaller, local machines, reducing environmental footprints. This would contrast sharply with the government’s current approach focused on massive data centers, which have been criticized in light of potential environmental impacts.

Envisioning a Sustainable Future for AI

Building CanGPT undoubtedly presents challenges, including questions of funding, continual updates, and achieving competitive performance against commercial AI counterparts. However, this initiative would spark a collective national dialogue on the social role of AI, regulatory frameworks, and how public institutions can shape our digital infrastructure.

Envisioning a public AI model not only highlights the potential for delivering improved services but also paves the way for alternatives to the subscription-based models offered by large tech corporations.

Conclusion

The concept of CanGPT may initially seem unconventional, but it underscores a critical juncture in Canada’s approach to public service media and digital sovereignty. As we consider the future of AI, it’s essential to imagine pathways that prioritize public good over commercial gain. The exploration of a public AI model like CanGPT opens up exciting possibilities for harnessing AI’s promise in a manner that benefits all Canadians.