Cited by Explainable Automatic Grading with Neural Additive Models

The future of standardised assessment: Validity and trust in algorithms for assessment and scoring DOI

Cesare Aloisi

European Journal of Education, Год журнала: 2023, Номер 58(1), С. 98 - 110

Опубликована: Янв. 17, 2023

Abstract This article considers the challenges of using artificial intelligence (AI) and machine learning (ML) to assist high‐stakes standardised assessment. It focuses on detrimental effect that even state‐of‐the‐art AI ML systems could have validity national exams secondary education, how lower would negatively affect trust in system. To reach this conclusion, three unresolved issues (unreliability, low explainability bias) are addressed, show each them compromise interpretations uses exam results (i.e., validity). Furthermore, relates trust, specifically ABI+ model trust. Evidence gathered as part validation supports four trust‐enabling components (ability, benevolence, integrity predictability). is argued, therefore, barriers limit extent which an AI‐assisted system be trusted. The suggests addressing unreliability, bias should sufficient put par with traditional ones, but might not go far fully reassure public. achieve this, it argued changes quality assurance mechanisms will required. may involve, for example, integrating principled frameworks assessment policy regulation.

Язык: Английский

Процитировано

Can AI support human grading? Examining machine attention and confidence in short answer scoring DOI