Open-access research article · 2026, Number 1
Teaching and Assessing Legal Reasoning with Generative AI in Thai Universities
Published · Pages 48–58
Abstract
An accurate legal memorandum prepared with generative artificial intelligence may conceal uncertainty about the proposition on which its advice depends. For a university awarding credit, the difficulty is to establish which legal judgements the student can defend when preparation has been assisted and questioning may itself supply part of the answer. This article combines a critical narrative review with doctrinal analysis of Thai law to develop an assessment sequence comprising a prepared memorandum, observed justification and controlled factual variation. Its central argument attaches independence to the intellectual operation being assessed. Finding a provision, interpreting a provision identified by an examiner and applying a supplied interpretation warrant different attributions of achievement. A fictional licensing problem, followed by Thai administrative provisions and a limited Singapore comparison, demonstrates how source selection, statutory language and evidence determine those attributions. Thai qualification standards and institutional policies provide the setting for examining the proposal’s fairness, workload and treatment of student records. The article also distinguishes the evidence needed to mark present understanding from that needed to investigate misconduct, retention or an improvement caused by teaching. It proposes separate studies of assessment validity and educational effectiveness before wider adoption. By relating credit to the reasoning a student actually performs, the framework gives universities a more precise basis for certifying judgement in advice on which other people may rely.
Keywords: legal education; generative artificial intelligence; assessment validity; legal reasoning; evaluative judgement; Thai higher education