Examining Committee
Professor Moa Johansson, Chalmers Tekniska Högskola
Docent Mathias Creutz, Helsingfors universitet, Finland
Docent Henrik Gyllstad, Lunds universitet
Substitute if member in the committee will be missing:
Professor Christina Lindqvist, Göteborgs universitet
Opponent
Docent Senja Pollak, Jozef Stefan Institute, Slovenia
Chair of the Public Defence
Professor Elena Volodina, Göteborgs universitet
Abstract
There has been an increasing push for the use of automated systems in the last few decades. Educational contexts are no strangers to this push, as AI has found its way to both teaching and assessment. Multiple organisations dedicated to standardised language testing have already deployed automated language assessment systems. However, the use of algorithmic automation in educational settings is considered high risk due to concerns regarding fairness and accountability.
In this doctoral thesis, Ricardo Muñoz Sánchez studies the concept of fairness in the context of automated language assessment. This is explored through two core strands of research: language modelling for automated essay classification and bias and fairness in AI. These two strands are woven together into a rich tapestry, ultimately leading up to a crucial question: can automated language assessment be fair and, if so, under which conditions?
