Transparency, Trust
A New Paradigm for Assessment
Real Classrooms. Real Impact.
CyberScholar has been piloted in diverse learning settings, ranging from large urban schools to small rural classrooms, enabling students and teachers to co-create learning through AI-driven feedback. These studies reveal how the platform deepens reflection, enhances writing, and reduces teacher workload.
Case Study 1: Empowering Student Writers Across U.S. Classrooms
121 students | Grades 7–11 | 4 schools | ELA, Social Studies, Modern World History
CyberScholar was piloted in beta across four U.S. public schools, exploring how AI can support writing development in K–12 classrooms. The study engaged 121 students and four teachers in Grades 7–11 across English Language Arts, Social Studies, and Modern World History courses. Using Retrieval-Augmented Generation (RAG), CyberScholar integrated teachers’ rubrics, prompts, and instructional materials to deliver feedback directly aligned with curriculum expectations.
Students received detailed and iterative feedback on their writing, with real-time opportunities to refine drafts. The rubric-based scoring made assessment criteria transparent and motivated students to improve their work through visible progress tracking. Teachers observed that CyberScholar fostered metacognitive reflection and self-regulated learning, shifting their roles from evaluators to facilitators of AI-guided feedback loops.
Results showed enhanced writing quality, engagement, and understanding of disciplinary conventions. Students described the AI as a supportive, unbiased “second reader,” while teachers noted its ability to reduce workload and deepen student engagement with feedback. Overall, the pilot demonstrated that generative AI can strengthen formative assessment by embedding feedback into authentic classroom contexts—illustrating a model of cyber-social learning where improvement emerges from interaction among students, teachers, and AI.
Read more:
Castro, Vania, Ana Karina de Oliveira Nascimento, Raigul Zheldibayeva, Duane Searsmith, Akash Saini, Bill Cope, and Mary Kalantzis. 2025. "Generative AI in K-12 Education: The CyberScholar Initiative." arXiv 2502.19422. https://doi.org/https://doi.org/10.48550/arXiv.2502.19422.
Case Study 2: Rural Alaska Learns with AI
17 students | Grades 7–8 | Journal Writing Project
This CyberScholar case study was conducted in a rural Alaskan middle school, exploring how Generative AI can support student writing and metacognitive engagement. The school, serving a small and economically diverse population, integrated CyberScholar—a web-based writing assistant providing rubric-aligned feedback and interactive dialogue features—into a journal writing project for Grades 7–8.
Seventeen students participated, with eight completing all study phases and five engaging in think-aloud sessions. The AI Helper delivered formative, criterion-based feedback using Retrieval-Augmented Generation (RAG), allowing students to converse with the AI to clarify or expand its suggestions. Students valued the immediacy and interactivity of feedback, describing it as “like a friend talking to me.” They appreciated clear, structured guidance and the motivational star-rating system.
Thematic analysis revealed three major benefits: (1) enhanced engagement through interactive feedback loops, (2) improved writing quality via iterative revision, and (3) deeper self-reflection about writing processes. Teachers noted that AI-supported writing fostered independence and reduced grading load, while students demonstrated increased agency and ownership of revisions.
Read more:
Tzirides, Anastasia Olga (Olnancy), Michele Galla, Bill Cope, and Mary Kalantzis. 2025. "Thinking Through AI: Advancing Cognitive and Collaborative Research for AI in Education." EdArXiv. https://doi.org/https://doi.org/10.35542/osf.io/s8hqe