AI-Integrated Learning Across Four Generative AI-Supported Classrooms: Students’ Perceptions, Digital Confidence, Critical Digital Literacy, and Information Evaluation

Generative AI is increasingly used in higher education, but evidence remains limited on how students report multiple digital-learning outcomes across classrooms using different platforms within a common instructional framework. This exploratory posttest-only study describes four intact classrooms using ChatGPT (n = 29), Gemini (n = 29), DeepSeek (n = 28), and Perplexity AI (n = 32) during an eight-session AI-integrated learning program. Because each platform was represented by only one classroom, platform and classroom effects could not be separated; therefore, the study reports descriptive classroom profiles rather than platform-level causal or inferential effects. The four classrooms showed distinct descriptive patterns across the outcomes. The DeepSeek-supported classroom had the highest observed means for AITBLM perception (4.10), Digital Confidence (4.06), and Critical Digital Literacy (4.07), whereas the Perplexity AI-supported classroom had the highest observed Technology Information Evaluation mean (4.21). Across classrooms, AITBLM perception ranged from 3.86 to 4.10, Digital Confidence from 3.84 to 4.06, Critical Digital Literacy from 3.90 to 4.07, and Technology Information Evaluation from 3.89 to 4.21. ChatGPT and Gemini showed nearly identical AITBLM perception means (3.91 and 3.90), while Gemini and Perplexity AI showed comparatively higher TIE means (4.08 and 4.21). These contrasts indicate different classroom response profiles rather than a consistent ranking across all outcomes. The findings are exploratory and should not be interpreted as evidence that one AI platform is more effective than another.