Reasonable Doubt About Readiness: Lessons from Using MarkGPT (a custom GPT) to Grade and Provide Feedback on Law Assessments

The emergence of Open AI’s ChatGPT in late 2022 triggered considerable debate among academics both within Law Schools and more widely across Higher Education. Initially, debate focused on students’ use of ChatGPT, and later iterations of Generative AI (GenAI), in creating their academic work. Fears of cheating were rife and, to some extent, have proven well-grounded. However, part of the discussion moved to academic use of GenAI, and one element of this was its use to grade student work and provide feedback. For some, Generative AI has potential to be quicker and unbiased, and for others, GenAI marking is part of the insidious creep of the technology, risking academic redundancies. This article emerges from the MarkGPT project, which used ChatGPT to grade and feedback on student work which had been completed the previous year. The project compared the GenAI marking and feedback with the moderated grades from the module teams, and independent third markers. The project concludes that, as of Summer 2025, ChatGPT lacks the capacity for consistency (with itself, and with human markers), and accuracy in grading, but that a thematic analysis of the feedback showed significant similarities with the human markers. ChatGPT then, is not quite there. Yet.