According to Cotton and Shipway, University of Plymouth in an academic paper. About an eight minute read.
Comment: interesting paper, good discussion of the capabilities of these models, closes with an outline of strategies University’s can adopt to cope. They do raise an interesting point about the quality of text and TurnItIn detection of AI written submissions, however anyone with even basic skills in prompt engineering can insure generated text is increasingly chaotic, however at that point hallucinations become a problem. Prolly best for lecturers to craft better papers at the outset.
