Vanessa John Lobo

My research explores how AI question-answering systems can recognise when they do not have enough evidence to give a reliable answer. Specifically, I am investigating how Retrieval-Augmented Generation systems can assess whether the documents they retrieve from cultural and political collections are sufficient to support an answer, and what they should do when they are not, whether that means abstaining, giving a partial response, or explaining what is missing. The project also examines how these system behaviours affect user trust and understanding.