Assessment Authenticity in Generative-AI-Rich Learning Environments

Evelyn R. Harrison

Assessment Authenticity in Generative-AI-Rich Learning Environments

Author(s)	Dr. Evelyn R. Harrison
Country	United States
Abstract	Generative artificial intelligence has altered a foundational assumption of higher-education assessment: that a polished submitted artifact provides reasonably direct evidence of the student's underlying knowledge and capability. Contemporary generative systems can support planning, explanation, translation, coding, editing, data interpretation, multimedia production, and extended writing, making it increasingly difficult to infer learning solely from a final product. Current assessment-reform guidance consequently places greater emphasis on learning assurance rather than detection. The Australian Tertiary Education Quality and Standards Agency states that forming trustworthy judgments about learning requires multiple, inclusive, and contextualized forms of assessment and that institutions should redesign assessment to capture authentic demonstrations of student capability rather than invest primarily in detection mechanisms. This paper develops an Assessment Authenticity and Learning Assurance Framework (AALAF) for generative-AI-rich learning environments. Assessment authenticity is conceptualized through two related but distinct dimensions: professional authenticity, concerning the extent to which a task reflects meaningful disciplinary or workplace practice, and evidentiary authenticity, concerning the extent to which assessment provides trustworthy evidence that the claimed learning belongs to and can be demonstrated by the student. This distinction is necessary because recent research indicates that authentic assessment is not automatically resistant to generative-AI substitution. Kofinas, Tsay, and Pike found that the level of conventional assessment authenticity did not by itself determine the capacity to safeguard against or identify inappropriate generative-AI use. The proposed framework integrates contextualized tasks, process evidence, staged checkpoints, explicit AI-use conditions, source and output verification, reflective justification, oral or practical defense, and strategically located secure demonstrations of learning. A transparent simulation compares five assessment designs, ranging from a conventional unsupervised take-home essay to a combined artifact-plus-oral/practical-defense model. The simulated authenticity and learning-assurance score rises from 42 for the conventional take-home essay to 92 for the artifact-plus-defense model. These scores are methodological illustrations and do not represent empirical measurements of student performance.
Keywords	assessment authenticity, generative artificial intelligence, authentic assessment, academic integrity, learning assurance, assessment validity, AI literacy, oral defense, process assessment, higher education
Field	Engineering
Published In	Volume 3, Issue 6, November-December 2022
Published On	2022-11-02

View / Download PDF File

About AJAMIR Fees & Payment Current Issue Publication Archive	Submit Research Paper Track Submission Status Publication Guidelines Publication Ethics Peer Review & Plagiarism	Join as a Reviewer Editors & Reviewers Reviewer Referral Program Get Reviewer Membership Certi.	Website/Journal Policies Usage Policy Content Policies Privacy Policy

Contact Us		+1-872-302-3851	editor@ajamir.com

American Journal of Advanced Multidisciplinary Innovation and Research

A Widely Indexed Open Access Peer Reviewed Multidisciplinary Bi-monthly Scholarly International Journal

Assessment Authenticity in Generative-AI-Rich Learning Environments

Share this