Auditable assessment redesign for teaching readiness in preservice teacher education under generative AI

Generative AI is rapidly reshaping educational practice, yet in preservice teacher education it remains unclear how pedagogical reasoning can be evaluated when AI-assisted outputs become increasingly polished and accessible. This issue is especially salient in programs that prepare future teachers to translate digital tools into classroom-ready pedagogical decisions. This study examined how an undergraduate teacher-education program redesigned assessment to address the gap between AI exposure and teaching readiness under AI-allowed conditions. Using a convergent mixed-methods evaluation design, we triangulated curriculum documents with student, graduate, employer, and course-evaluation surveys to identify recurring priorities for improvement. Across sources, convergent evidence highlighted recurring priorities related to hands-on guidance, practicum and micro-teaching alignment, reusable exemplars, data-informed diagnosis, and explicit boundaries for responsible AI use. These were subsequently synthesized into three linked redesign priorities concerning teaching-facing transfer, auditable process evidence, and clearer AI-use boundaries, which informed an auditable assessment redesign comprising task templates, evidence trails, an analytic rubric, and practicum-oriented mapping. To strengthen the evidence base beyond self-report and document analysis, two raters independently scored 24 sampled final artifacts, yielding acceptable agreement for the rubric total score (ICC(2,1) = 0.77). A micro-teaching evidence pack was also used as an enactment proxy, alongside two cross-institution implementations that provided preliminary evidence of scoring defensibility, classroom feasibility, and cross-site workability. Rather than making causal impact claims, the study offers feasibility-oriented evidence that assessment redesign under AI-allowed conditions can make pedagogical reasoning more visible, more assessable, and more closely connected to practice-oriented performance in preservice teacher education. The findings contribute to current discussions of AI-aware teacher preparation and assessment redesign by showing how process-visible evidence can support more credible inferences about teaching readiness than product-only evaluation alone.