Academics Reallocate Evaluation To AI Assistants

Prince Sarpong, associate professor at the University of the Free State, argues that human evaluators face finite cognitive bandwidth that degrades marking quality after repeated assessments, citing decision fatigue by the 40th script and gaps in niche expertise. He recommends integrating large language models to handle mechanical verification—machines can pass exams including the CFA—so academics can focus on nuanced judgement.