This dashboard explores how French users engage with generalist chatbots, examining whether AI primarily serves as an automation tool, an augmentation tool, or fulfills other roles in human-AI interaction.

The Bunka.ai team analyzes data from the Compar:IA dataset, the most comprehensive collection of conversations from the French Ministry of Culture's AI comparison platform (comparia.beta.gouv.fr), launched in October 2024. This dataset offers particular value because it captures genuine, unconstrained usage primarily in French, revealing natural interactions rather than controlled scenarios. With over 175,000 authentic questions and responses from more than 59,475 distinct users engaging with 30+ different conversational AI models of various sizes (both open-source and proprietary) it provides unprecedented insight into real-world AI usage patterns.

To examine user conversations systematically, the Bunka.ai team has developed specific analytical dimensions with unique classifiers: Task Detection identifies activities like creation, information seeking, and recommendation seeking; Topic Detection categorizes content across areas such as arts and culture, education, and coding; Language-Level Detection distinguishes between conversational, formal/professional, and academic/advanced communication styles; and AI Interaction Mode differentiates between augmentation and automation approaches.

The analysis draws insights from a representative sample of 25,000 conversations, offering a comprehensive view of how French users interact with state-provided AI services.