On the 29th May we had the second UAL AI Reading Group where we discussed the article, Artificial intelligence addiction: Exploring the emerging phenomenon of addiction in the AI age by Liqaa Habeb Al-Obaydi and Marcel Pikhart. We started with a short summary of the article by Mark Robinson, Learning Technologist at LCF. Mark’s introduction to the paper framed AI addiction as an emerging and under-researched issue, particularly in education. He noted that only a small number of studies currently explore AI use in education making it difficult to draw firm conclusions.
He explained that the paper identifies features of AI systems, especially dynamic feedback loops as potential drivers of repeated use and addiction. These systems adapt to users in real time, which can make them engaging in ways that differ from earlier digital technologies. Mark also highlighted the role of anthropomorphic design, where AI mimics human behaviour increasing the likelihood that users can develop emotional connections to chatbots.
Dependency or addiction?
One of the starting points for our discussion was a issue noted in the article that there is some evidence to suggest that levels of AI “addiction” among students remain relatively low, particularly for highly motivated learners. Rather than addiction in a clinical sense, I suggested that we might think about softer concept, ‘dependency’. AI tools are not necessarily creating compulsive behaviour, but they are becoming an easy default especially for tasks like brainstorming or summarising. The danger may not be addiction but the subtle erosion of independent thinking when AI becomes the “go-to place.”
Seeing AI as a person not a machine
A second point we discussed is the growing role of “human like” AI design. When chatbots appear conversational, empathetic, or even friendly, they can encourage deeper engagement. This can lead to what someone (Mark?) identified as parasocial interaction, users relating to AI both as a tool and as a social presence. This dual perception, AI as both a tool and “friend” can led to increase emotional attachment, especially when AI is always available 24/7.
Cognitive Offloading
During the discussion several of us shared concerns that students are increasingly offloading thinking onto AI. Sometimes referred to as cognitive offloading. Examples ranged from summarising teaching materials without fully engaging with them, to generating ideas instead of developing them independently. This pattern risks bypassing the productive struggle that underpins learning. In fact this struggle is essential to the creative process and is not always easy. As someone put it, this could narrow thinking and reduce imagination.
Emotional Attachment
We spent a little bit of time discussing how how individuals, especially younger users are beginning to turn to AI for personal advice and emotional support. Chatbots constant availability makes them very appealing. This raises questions about fake empathy. Because AI is a machine it simulates understanding without genuinely possessing it but this may still feel meaningful to users. Again as someone said these behaviours echo broader societal shifts already seen with social media like much fewer face-to-face interactions and increasing reliance on social media communication. AI may be less a radical break than an intensification of these trends?
Responsibility, and Regulation
We finally got onto the question of who should take responsibility for the safe use of AI. Should educators manage AI use, or should this sit with technology companies? The group recognised that AI systems are often designed to maximise engagement which mirror patterns seen in social media. And like social media I think the consensus was that AI companies should be taking responsibilty for this, whether that actually happens is another question!
To conclude
Its so difficult to summarise and draw any conclusions to what was a really interesting (and enjoyable) conversation but its clear that AI addiction is an under researched topic. Behaviours around dependency, cognitive offloading, and emotional engagement are already here. What we need to do as educators is to try and understand them more before they become fully embedded into or behaviours.
Next reading group
Friday 26th June 12 to 1pm
We’ll be discussing the article The ‘Tragic Dilemma’ of GenAI in Education: An Exploratory Human-AI Conversation authored by Agnieszka Palalas, Mark Pegrum, Charlotte N. Gunawardena, Rachel Siow Robertson & Jarrod Ventura. Postdigital Science & Education 2025
Get your ticket (UAL Staff only)
Leave a Reply