What value does human assessment of essay submissions still have in the age of AI? Our group leader, Professor Deborah Talmi, has recently won ai@cam funding to address this important question, collaborating with Cambridge colleagues Dr Alexandru Marcoci, Dr Ruth Walker, and Isla Fay, and Roni Tibon (University of Nottingham), Yael Benn (The Manchester Metropolitan University), and Maryam Abo-Tabik (University of Central Lancashire).
Higher education has been unsettled by the potential for AI to disrupt conventional assessment and marking practices. Large Language Models already produce passable university essays and sophisticated summative and formative feedback; their performance improves exponentially. Given the importance of student essays in assessment, universities must prepare for the challenges that this innovation poses to their core mission.
The sector has mainly responded by addressing the risk to the integrity of assessments. From informing students on responsible use of AI, revising coursework to decrease reliance on AI, to switching to examinations and viva voces, to date, efforts mostly target the student side of the equation. By focusing on the academic evaluators of submitted work we address a related but separate issue that will acquire increasing importance in the coming years: the complementary roles of human and AI assessment.
This initial project focuses on assessment of Psychology essays, where excellence requires a combination of creativity and originality together with skilful application of rigorous research methodology and meticulous references to the established scientific literature.
Comentarios