NEW APPROACHES TO ASSESSING STUDENTS’ INDEPENDENT WORK IN THE ERA OF GENERATIVE AI

Аннотация

Generative AI has made finished text an unreliable indicator of the work behind it. This article aims to systematise the approaches to assessing students’ independent work proposed in response to this problem and to assess their suitability for Uzbek higher education. Sixteen sources were purposively selected, including international publications indexed in Scopus or Web of Science and publications from Uzbekistan. Five approaches emerged: stating the permissible use of AI, assessing the process instead of the product, oral defence of written work, analysis of digital traces from learning systems, and using AI as an assessment tool. A consistent trade-off appears, in that the strongest evidence of authorship requires the most teacher time, while the cheapest methods yield the weakest evidence, and no single approach is sufficient. Two approaches can be adopted in Uzbek universities immediately and without additional resources: clearly stating the permissible use of AI in each assignment and selectively using oral defence for major work. Process-based assessment requires courses to be redesigned first, and trace-based assessment remains a research direction, not a practical recommendation. In philological subjects, where interpretation and aesthetic judgement matter most, approaches built on students explaining and defending their work are better suited than automated scoring. The novelty of the study lies in the first systematisation of these approaches for the conditions of Uzbek higher education.