A new interdisciplinary study shows that learners are happy to share their learning data at first. But group discussions change this willingness.

The digital transformation of the education system is creating increasing amounts of learning data. But what role do social interactions play in the decision to share this data with others? Recently published in the journal Humanities Social Science Communications, a study led by Dr. Louis Longin from LMU’s Chair of Philosophy of Mind, LMU, and researchers from Technical University Munich (TUM) sheds light on the matter. It turns out that learners become considerably more cautious around data-sharing decisions after they discuss these decisions with peers.

“In different professional circles, we talk a lot about the ethical use of data, but there has been hardly any research into how individuals can be better integrated into these decisions,” explains Longin. “Through this interdisciplinary collaboration, we had a chance to bring interactive decision-making into an educational context. We wanted to understand how group discussions and contextual factors affect decision-making.”

The research team carried out an experiment with 60 test subjects. In different phases – both individual and after group discussions – these participants rated how acceptable they found the sharing of their learning data under various conditions.

Awareness of context through group reflection

The results show that group discussions sharpen learners’ awareness of different contexts where their data are collected by educational technologies. Do learners care, for example, who is collecting their data? The participants scarcely distinguished between different data sharing contexts before the group conversation, but it was the discussion that caused them to be more cautious. Researchers also observed that the discussions were biased. When learners discussed data sharing with private companies, the talk focused on the benefits, whereas when the same data were shared with government entities, learners tended to discuss their concerns about the data flows.

“The effect of integrating interactive decision-making into otherwise top-down approaches of informed consent is important. More data are collected to support learning with AI and equipping learners with tools for better decisions is essential” emphasizes TUM-Professor Oleksandra Poquet.

Participatory practice for responsible educational technology

According to the researchers, the results have far-reaching consequences for educational technologies. “Our proposed method of interactive informed consent could start addressing power disbalances in data collection practices,” reckons Longin. “Through the integration of group discussions in decision-making processes about data consent, educational institutions can promote a learner-centered approach that enables meaningful engagement with decisions about one’s data.”

Moreover, this research fundamentally contributes to the development of a paradigm shift away from a purely technical consideration of data protection toward a participative model, which actively involves learners in decisions about their data. It opens questions about critical literacy needed to properly engage in these decisions.