Artificial Intelligence Across the Educational Assessment Lifecycle: Human-Centered Approaches to Design, Scoring, Feedback, and Validation

Artificial intelligence is rapidly transforming the educational assessment lifecycle. Generative AI, large language models, learning analytics, educational data mining, and adaptive technologies are increasingly being used to generate assessment tasks, analyze learner responses, automate scoring, diagnose learning needs, and provide personalized feedback. Emerging technologies are also empowering novel assessment formats such as conversation-based assessments and human-AI collaboration tasks. Recent studies illustrate the potential of AI-generated assessment items and LLM-based scoring, while also demonstrating the need for careful psychometric evaluation and contextual validation (Crossley et al., 2026; Isley et al., 2026).These developments create opportunities to move assessment beyond the production of scores. AI-supported systems may capture evidence of learners’ reasoning, problem-solving strategies, misconceptions, and developing knowledge states. Process data, log data, eye-tracking, and other multimodal evidence may also provide a richer understanding of how learners engage with assessment tasks and arrive at their responses (Chen et al., 2023). Automated systems can additionally provide immediate and individualized feedback, although existing evidence indicates that the effectiveness of such feedback varies substantially across learners, tools, and instructional contexts (Fleckenstein et al., 2023).However, technical accuracy or agreement with human ratings does not by itself establish that an AI-supported assessment is valid, reliable, fair, transparent, or educationally meaningful. Valid assessment requires evidence supporting the intended interpretation and use of scores, as well as attention to accessibility and fairness across learner populations (AERA, APA, & NCME, 2014). AI-supported assessment introduces additional concerns related to algorithmic bias, model instability, boundary violation, limited explainability, data privacy, and excessive reliance on automated judgments (Baker & Hawn, 2022; Messeri & Crockett, 2024).This Research Topic examines how AI can support the educational assessment lifecycle while maintaining sound measurement principles, meaningful human oversight, and clear educational purposes. It seeks to connect research in educational measurement, psychometrics, artificial intelligence in education, learning analytics, educational data mining, assessment design, and the learning sciences.We invite empirical, conceptual, methodological, theoretical, review-based, and design-based contributions that examine the use of AI in assessment design, administration, scoring, interpretation, reporting, feedback, and instructional decision-making. We are especially interested in studies that move beyond demonstrations of technical capability and examine the quality, meaning, fairness, and educational consequences of AI-supported assessment. Submissions should clearly describe the assessment purpose, the role assigned to AI, and the evidence used to support resulting interpretations or decisions.Topics of interest include, but are not limited to:1. AI-Supported Assessment Design and Development● Generative AI and large language models for assessment task and item development● AI-assisted development of rubrics, scoring criteria, simulations, and performance assessments● Human-AI collaboration in assessment design, task refinement, and quality assurance2. Automated Scoring and Assessment Decision-Making● Automated essay scoring and evaluation of open-ended, multimodal, and performance-based responses● Human-AI collaboration in scoring, moderation, interpretation, and assessment decision-making● Validity, reliability, generalizability, interpretability, and reproducibility of AI-supported scoring and assessment systems3. Adaptive, Diagnostic, and Multimodal Assessment● Adaptive assessment, cognitive diagnosis, knowledge tracing, and learner-state modeling● Use of process data, log data, response sequences, eye-tracking, and multimodal assessment evidence● Methodological advances for analyzing complex, process-oriented, and multimodal assessment data4. AI-Generated Feedback and Learning Processes● AI-generated formative, personalized, explanatory, and dialogic feedback● Learner uptake of AI-generated feedback, including reflection, revision, self-regulation, and transfer● The consequences of AI-supported assessment for teaching, learning, motivation, and educational opportunity5. Fairness, Human Factors, and Responsible Implementation● Bias, fairness, accessibility, multilingual assessment, and assessment for neurodivergent learners● Teacher and learner trust, agency, assessment literacy, and human oversight● Privacy, data governance, accountability, and responsible institutional implementationContributions are welcome across K–12 education, higher education, teacher education, professional and workplace learning, informal learning, and large-scale assessment. The scope encompasses formative, summative, diagnostic, adaptive, classroom-based, performance-based, and standardized assessment contexts. Interdisciplinary work connecting AI with educational measurement, learning sciences, assessment design, and responsible technology implementation is particularly encouraged. Topic editors: Bo Pei, Yizhu Gao, YI-HSIN CHEN, Tongxi LIU Source: https://www.frontiersin.org/research-topics/84568/artificial-intelligence-across-the-educational-assessment-lifecycle-human-centered-approaches-to-design-scoring-feedback-and-validation