Human-centred learning analytics: Four challenges in realising the potential

The notion of Human-Centred Learning Analytics (HCLA) is gaining traction as educators and learning analytics (LA) researchers recognise the need to align analytics and artificial intelligence (AI) technologies with specific educational contexts. This has led an increasing number of researchers to adopt approaches, such as co-design and participatory design, to include educators and students as active participants in the LA design process. However, some experts contend that HCLA must go beyond stakeholder participation by also focusing on the safety, reliability, and trustworthiness of the analytics, and balancing human control and algorithmic automation. While the adoption of human-centred design (HCD) approaches promises considerable benefits, implementing these practices in data-intensive educational systems may not be straightforward. This paper emphasises the critical need to address specific ethical, technical, and methodological challenges tied to educational and data contexts, in order to effectively apply HCD in the creation of LA systems. We delve into four key challenges in this context: i) ensuring representative participation; ii) considering expertise and lived experiences in LA design; iii) balancing stakeholder input with technological innovation; and iv) navigating power dynamics and decision-making processes.