Exécution de notre outil de synthèse sur un nouvel article

Résumé : Universities are under pressure to build generative artificial intelligence (GenAI) literacy quickly, yet few documented interventions address generative AI specifically or can be deployed outside the curriculum at short notice. This paper documents AI Literacy Lab, a ten-hour extracurricular workshop programme designed and taught by a student research society at the Jagiellonian University in Kraków in 2026 and evaluates what it achieved. Thirty-three students enrolled, 85% of them from one social sciences and humanities faculty, and 67% reported no regular use of AI coding assistants. The programme ran as five two-hour modules covering large language models, everyday study workflows, prompt engineering, academic method and ethics, and AI-assisted programming, delivered to two parallel groups, one on campus and one online, with open educational resources and shared API access. Evaluation combined a baseline survey (N = 33) with a post-workshop survey (N = 23, 70% of those enrolled and 85% of the 27 who attended at least one session) using retrospective then-now self-ratings. Tool repertoire broadened from ChatGPT as the dominant tool to a median of four tools including locally hosted models, and 16 of 23 respondents described a concrete independent application afterwards. Self-rated competence rose on all five constructs by mean increases of 1.78 to 2.52 points on a five-point scale, with every Wilcoxon test remaining significant after Holm correction. No objective knowledge test was used. The paper treats these gains as bounded: across 115 individual comparisons not one participant reported a decline, a pattern the retrospective format produces by construction, and an independent pre-post design would not. A comparison against the concurrent baseline is reported as a partial check on recall.