Quebec’s national library, Bibliothèque et Archives nationales du Québec (BAnQ), is embarking on an ambitious project to create a database that will house cultural and governmental content aimed at improving artificial intelligence systems’ comprehension of Quebec’s unique society, culture, and Indigenous languages. Following the completion of a feasibility study earlier this year, the library has initiated an experimental phase for this databank, which is designed to address significant gaps in the data available for training AI models pertaining to Quebec.
Addressing the Data Gap
The move comes in response to growing concerns that existing generative AI technologies often fail to accurately represent Quebec’s societal nuances, economic landscape, and rich cultural heritage. Valérie D’Amour, who spearheaded the feasibility study, expressed optimism about the initiative’s potential. “All scenarios are a little bit on the table right now,” she stated. “We have a lot of ideas and we want to validate the possibilities with cultural stakeholders, as well as with data owners and providers, who will be involved in the discussions.”
BAnQ has clarified that the forthcoming platform will not serve as a public distribution channel for creative works. Instead, access to the curated data will be tightly regulated to ensure responsible use. Marie Grégoire, BAnQ’s president and CEO, underscored the necessity of integrating Quebec-related references into AI models, stating, “That means having Quebec references, whether in small models or large models, whether they come from research or from the business community.”
Learning from Global Examples
Similar approaches have been adopted in other regions, such as Sweden, where extensive collections of Nordic-language texts have been compiled to aid in the development of generative AI models for Scandinavian languages. BAnQ intends to begin its project by leveraging its own collections, gradually expanding to include data sourced from additional entities. This initiative aligns with a 2024 report from Quebec’s innovation council, which highlighted the scarcity of Quebec-specific data in AI training datasets.
Destiny Tchéhouali, co-holder of a research chair focused on French-language AI and digital technologies, emphasised the importance of greater representation of Quebec culture within AI frameworks. He warned, “We run the risk of reproducing linguistic biases and cultural biases. And when we also talk about Indigenous peoples, we run an even greater risk of all these biases.” He sees the proposed database as “strategic infrastructure” that would assist in establishing guidelines for identifying and cataloguing local content within today’s AI systems.
Protecting Creators’ Rights
As BAnQ develops the database, concerns regarding copyright have arisen within the cultural sector. However, Grégoire believes that the new platform could actually enhance protections for creators compared to the current landscape. “Right now, it’s a bit like the Wild West,” she remarked. “Data is being harvested for free, and that should not be the case.” The database aims to act as a centralised gateway, facilitating better compensation for creators whose works are utilised.
Despite these assurances, some artists remain apprehensive about contributing their work to AI training systems, fearing it could ultimately jeopardise their livelihoods. Maxime Harvey, a postdoctoral researcher at the National Institute of Scientific Research, conveyed the prevailing concern: “Even if artists earn income from it, they are still feeding the beast that will eventually be used to replace contracts they may lose because of AI.”
Project Timelines and Funding
The feasibility study outlines a timeline for the platform to become operational by 2029, although D’Amour indicated that this schedule will be re-evaluated post-experimental phase. The projected budget for the initiative over the next five years is estimated at nearly $10.5 million, encompassing both operational and capital expenses. To support this effort, BAnQ has already secured $340,000 from the Quebec government for the feasibility study and an additional $750,000 to fund the current experimental phase.
Why it Matters
This initiative by BAnQ represents a significant step towards ensuring that Quebec’s cultural narratives are accurately represented in AI technologies. By creating a dedicated database, the province aims to combat the underrepresentation of its society and Indigenous languages in the digital realm. As AI continues to influence various sectors, the need for a more inclusive approach becomes paramount. This project not only seeks to safeguard the rights of local creators but also aspires to weave the rich tapestry of Quebec’s culture into the fabric of artificial intelligence, ultimately fostering a more nuanced understanding of diversity in the digital age.