AI Opens Up Discussion on the Relationship Between Academic Writing and Learning in Higher Education
The widespread adoption of generative AI in academia has dismantled the assumption that writing fluent academic texts demonstrates true learning. Researchers point out that traditional essay evaluations measure linguistic and social background rather than subject mastery.
Generative AI tools and large language models have explicitly called into question the long-standing assumption in higher education institutions that academic essays serve as a direct indicator of learning. Researchers Sioux McKenna and Neil Kramm emphasize that fluent academic writing does not always point to a genuine understanding.
Distinction Between Fluent Writing and Genuine Understanding
The humanities and social sciences have long accepted the fluency of a student's submitted text as direct evidence of their understanding of the subject. However, the ability of large language models to generate well-organized and discipline-appropriate academic texts within seconds challenges this notion.
The fact that a system lacking the capacity to make sense of the world can produce accepted academic texts raises the question of what these texts were measuring in the past.
Linguistic Background and Social Factors
According to the researchers, a student's ability to write in correct academic language is linked more to how familiar they are with the university's discourse style than to how much they know about the subject. Evaluations often highlight academic literacy rather than domain proficiency.
For example, in South Africa, class barriers are further deepened by colonial language hierarchies because university discourse is the students' second or third language.
Need for Change in Assessment Methods
In the face of tools capable of producing graduate-level text, higher education's initial reaction was largely focused on surveillance, policing, and punishment. However, experts argue that rather than penalizing, new assessment models should be designed to make the student's thinking process visible.
To reveal the learning process, drafts featuring visible justifications, oral defenses, real-time concept applications, and structured self-assessments are among the suggested methods.