When Quebec Fiction Brings Generations Together
At the Quebec City Convention Centre, the 12th Comiccon brought together thousands of popular culture, comic book, and science fiction enthusiasts. Among the guests of honour this year, Quebec actor and screenwriter Claude Legault drew a multigenerational crowd that came to celebrate one of the defining figures of Quebec television and film. As reported by Radio-Canada, the creator witnessed firsthand the enduring public affection for cult works that continue to span generations, from the sci-fi comedy Dans une galaxie près de chez vous to the police drama series 19-2.
This widespread enthusiasm highlights a reality often overlooked in technology discussions: the resilience of local imagination. Whether through the absurd adventures aboard the spaceship Romano Fafard or the raw dialogue of Montreal patrol officers, these productions reflect a rhythm, humour, and vocabulary deeply rooted in Quebec society. They demonstrate that culture cannot be reduced to textbook grammar; it thrives on distinct turns of phrase, shared references, and a singular brand of self-deprecation.
Yet as artificial intelligence systems claim to simulate conversation, generate stories, and assist human creation, a fundamental question emerges for citizens: how do these new engines handle our collective memory and cultural figures?
The Limits of Large Models Facing Cultural Specifics
When a French-speaking user asks an artificial intelligence agent to draft a speculative scene in the voice of a Quebec spaceship crew or interpret everyday colloquial expressions, the built-in limits of foundation models quickly surface. Most large language models (LLMs), whether built by California tech firms or other global conglomerates, are trained on vast text corpora dominated by English and normative European French.
This documentary asymmetry creates noticeable cultural and linguistic biases. In a scientific paper published on arXiv by research teams from Université Laval and Concordia University, scholars demonstrated the major difficulty language models encounter when adapting to regional French dialects without thorough retraining. Mainstream architectures systematically smooth away regional characteristics, treating Quebec idioms as syntax errors or replacing them with Anglo-American or Parisian syntactic calques.
As Destiny Tchéhouali, co-chair of the Quebec Research Chair on Francophone Artificial Intelligence and Digital Technologies at UQAM, points out, cyberspace faces a form of cultural algocracy where transnational platform algorithms render minority creations invisible. For its part, Quebec's Commission de l'éthique en science et en technologie notes that algorithmic biases are not merely a technical performance challenge; they represent an ethical issue. By prioritizing statistical averages from global datasets, generative AI risks homogenizing creative expression and sidelining key elements of Francophone diversity.
GoIA: A Comparative Window for Citizens
It is within this framework of transparency and cultural ownership that the GoIA application fits inside the sovereign ProductivIA platform. Designed as an accessible, fully no-code entry point for the public, GoIA offers an interface where anyone can interact with artificial intelligence without complex installation or restrictive subscriptions.
The defining feature of GoIA lies in its multi-model architecture: rather than locking users into a single system whose biases remain hidden, the tool allows them to query and compare multiple language models simultaneously using the same prompt. A curious citizen can test how well a model understands a line written by Claude Legault, submit a speculative scenario, or compare how different engines interpret Quebec irony.
In particular, the application enables users to evaluate commercial systems alongside open-source alternatives and Matania, a sovereign Quebec provider. By observing lexical accuracy, response speed, and contextual nuance side by side, users build genuine digital literacy. The exercise demonstrates why a model hosted on Quebec servers and fine-tuned on local data reacts differently from a distant system trained on Silicon Valley standards, all while ensuring that conversation data is never repurposed for advertising or profiling.
Looking Ahead
The public attachment witnessed at events like Quebec City Comiccon proves that fictional heroes are far more than commercial products; they are lasting cultural anchors. As artificial intelligence makes inroads into screenwriting, historical outreach, and cultural transmission, safeguarding our distinct voice within these tools remains a major societal concern. Developing open and comparative application environments is an essential first step toward ensuring tomorrow's technologies reflect the depth of our stories rather than flattening them.