Forming a dedicated creative AI innovation fund and developing a national AI literacy strategy are among the key recommendations for Canada’s AI strategy in research from OCAD University’s Cultural Policy Hub.

The recommendations come as part of an Oct. 31 submission to Innovation, Science and Economic Development (ISED) Canada. The federal department had invited Canadians to weigh in on the country’s AI strategy, to be developed by the end of 2025.

The Cultural Policy Hub’s submission was developed through consultations, internal research, surveys and recommendations from roundtables that the Hub has collected over the past two years. The goal of the document is “to chart opportunities for the creative industries to be active partners in the development and adoption of ethical AI strategy that reflects the rights, contributions and values of those working within the sector.”

The Hub’s four recommendations include the creation of a dedicated creative AI innovation fund to finance public-private partnerships, and the establishment of a national cultural sovereignty framework mandating explicit consent for data use and metadata regulations.

The Hub also asks for the building of an AI commons to increase access to AI compute, data, tools and education, and the launch of a national AI literacy strategy for the creative economy, ranging from K-12 into post-secondary and beyond.

Further recommendations are split across four categories: cultural sovereignty and creative ambition, sovereign and accessible data, regulation and skills, and training and education.

The first section, cultural sovereignty and creative ambition, focuses on expanding access for creatives to AI technology while defending Canadian and Indigenous cultural sovereignty.

The recommendations include hosting an annual AI and culture summit to showcase Canadian innovation, and integrating Indigenous narrative and data sovereignty principles into federal policies.

The Hub expands further on the issue of cultural sovereignty in the section on sovereign and accessible data, with recommendations about ensuring Canadian data remains under Canadian jurisdiction and that said data and AI technology is affordable to creators.

Alongside an AI commons, the Hub asks for the mandating of “Canadian jurisdiction only” clauses in federal AI procurement, requiring that any data storage for a contract involving cultural content or AI projects occur on Canadian-owned cloud infrastructure.

The section also recommends the allocation of affordable computing resources through a national platform; establishing Canadian-based systems for training AI with cultural data; and investing in public compute and cloud infrastructure co-managed by government and other partners to ensure affordable access.

Moving to regulation, the most commonly proposed measures the Hub heard through its survey included proper labelling and disclosure of AI-created or -altered content, standards to protect creators’ rights, limiting the environmental impact of AI, requiring developers to be transparent about their systems, and data and clarification on ownership of AI-generated content and how it can be used.

Interest holders also shared that text and data mining (TDM) — how AI models can absorb vast amounts of content to improve their pattern recognition — should be subject to the Copyright Act, and that TDM must be voluntarily agreed to by rights holders.

For further regulation, the Hub recommendations include embedding cultural institutions as partners in developing and stress-testing AI frameworks, creating a national AI content registry where cultural institutions can register works and be notified when a developer requests access, and the mandating of transparency on AI models’ data sources, bias mitigation measures and licensing status.

The final section — skills, training and education — focuses on improving AI literacy through a national, cross-disciplinary AI education and training strategy. Along with developing and launching an AI literacy strategy, recommendations include the funding of education campaigns that use the arts to educate Canadians, and the creation of a cultural AI advisory council that shares responsibility on AI with ISED, including representatives from the creative industries, cultural organizations and AI ethics centres.

In addition, the Hub suggests mandating cultural impact assessments, similar to environmental assessments, that will evaluate AI’s impact on creators’ livelihoods; and the creation of a cultural AI fellowship to train Indigenous and marginalized creators in AI tools and creative innovation.

The Hub’s full submission to ISED can be found here.

Image: Unsplash