In dialogue with ChatGPT on the potential and limitations of AI for evaluation in education

Abstract

The development of Deep Learning (DP) techniques has enabled Artificial Intelligence (AI) applications in a wide range of economic, political and cultural areas, opening up new perspectives (OECD, 2022; Ia.italia.it, 2023) that present enormous opportunities but also pose great risks and ethical questions (Harari, 2018). As a result, the interest in AI’s educational applications has also increased (UNESCO, 2021; 2022; Goksel, Bozkurt, 2019; Panciroli et alii, 2020; Pham, Sampson, 2022). Following pioneering works that consider natural language processing algorithms as expert interlocutors, this contribution presents and discusses the results of a conversation with ChatGPT – an evolved chatbot prototype – on the opportunities and limitations of AI in supporting evaluation processes in education.