Recent studies, preprints, and journal articles on learning and AI, selected by our editors. Each entry links to the original paper; preprints have not been peer reviewed.
A cluster-randomized experimental study of a generative AI intervention designed to support students' internal feedback on their writing. The authors examine the roles of evaluative judgment and student-AI interaction in writing improvement. Relevant for educators weighing how AI feedback tools fit with students' own self-assessment.
A classroom study of 65 secondary students compared a Socratic writing chatbot, which questions arguments without generating essay text, against a control chatbot sharing the same interface and model. Students using the Socratic tool engaged more in idea-level dialogue, and moves from discussed ideas to actual revisions were associated with score gains. The authors propose coordinating reflection with revision as a design target for AI writing support.
An analysis of AI guidance documents from 35 states and Puerto Rico, comparing their priorities and tracing who wrote them and what they cite. The authors find near-universal attention to privacy and academic integrity, urgent workforce framing without matching guidance on capacity or accountability, heavy reliance on nonprofits like TeachAI, and no parents listed as contributors. It is useful context for leaders working within or writing such guidance.