Exploring the future of learning with Gen AI

 A webinar of the keynote and panel discussion from The Centre for the Study of Higher Education at University of Melbourne. It is an in-person event with zoom of  the opening keynote and panel session. 

Associate Professor Tim Fawns (Monash University) presents the keynote - Beyond cognitive offloading: student as a complex phenomenon. 

Notes of the presentation:

The presentation began with a welcome from Associate Professor Cochrane and an acknowledgement of country and an introduction to A/P Fawn.

Began with defining cognitive offloading - which might not be a bad thing! A sustained lack of critical thinking, through using technology, is one way to look at it. Current studies on cognitive offloading etc. have not perhaps been conducted validly. A student using AI, does not cause any form of brain damage. It is difficult to work out what students are thinking and we must not assume what students are thinking.

We move to assumptions when students use AI. We need to find out why and how students use AI. Students' practices, contexts and temporalities matter. One use of AI does not tell us much, however, intentionality matters - what are they using AI for?? 

In learning, collaboration matters. However, humans are a collective species. We need to work with others to achieve objectives. Individual knowledge matters but they need to also use this in conjunction with the milieu they are in. Dialogue with students is important to find out what they are doing with AI.

Cognitive offloading is 'the use of physical action to alter information processing requirements of a task so as to reduce cognitive demand (Risko & Gilbert, 2016). If you offload enough, the reduced task becomes the task. We offload all the time, due to the volume of information overload we encounter.

Human learning can be distinguished from 'cat learning (J. Dorn). We need to be more precise as to what we are looking at and how to better understand the values we need to protect with regard to human learning. Therefore, we cannot offload cognition that has not been thought of yet. You cannot know what would have been thought otherwise. The thing you overload to does not think what you would have thought. Offloading enables and requires (different) thinking, depending on your goals. 

There is both beneficial and bad effects of cognitive offloading. Short/long term effects are difficult to pin down. Cognitive offloading came about through work on transactive memory. A shared cognitive system where groups or couples collectively encode, store, and retrieve information. ' not traceble to any individual. Distributed cognition is the processes distributed across members of a social group, between internal and external resources, and across time (Hutchins, 2000). Therefore, it is cognitive distribution, 'remixing'.

Shared the work undertaken on student perspectives on AI in HE - AIinHE.org. The complexity and diversity of what is happening is a defining factor of the study, carried out over the last few years. Over 8000 responses show the main use to be editing/writing, brainstorming, teaching how to something, searching/researching, getting feedback (all over 50%). When an assessment task restricts AI, students think they are able to learn less deeply. 19% of students feel that being able to use AI in assessments, helps them to learn more deeply.

As an summary of using AI to undertake summaries, students found that it saved time, helped understand key concepts, manage heavy reading loads, engage with complex or unfamiliar (over 50%). Most students say they read the original text, as well as the summary.

Therefore, there is a big, messy and diverse mix, including resistance, low trust of outputs, varying confidence in their ability to use AI and worries about learning (Jung, 2025) AI is embedded in different ways. AI feedback seen as less trustworthy, less relevant, clearer, more 'objective' and less emotionally fraught (Henderson et al., 2025). Students were concerned about dependency, needed to trust themselves and their processes and were developing moral and ethics (Bearman, 2025).

Optimism, guilt, scepticism, relief, moral boundaris, fears ---- complex moral negotiation (Oberg et al. 2026). Things 'stick' to AI including alarm, distrust, enthusiasm and hope. 

Therefore, many students are worried about becoming lazy and taking ahortcuts. Overall, thoughtful. 

Shared work at Monash with a focus on evidence of learning. (Fawns, Boud, & Dawson, 2026) pm targeted learning to focus not only on product, but the performance, process and practices. Important to help students leverage off AI. However, the practices are still emergent and need to be better understood.

Questions on assessments include - are learning environments appropriate for encountering students and what they can do? Do assessment tasks allow demonstration of learning outcomes via a range of proxies? Are the conditions appropriate to generate evidence of learning? Are the evaluation methods appropriate to the evidence?

Therefore, AI relate beliefs and practices are as diverse as those of staff. AI in education is an even bigger, and more emotional and personal challenge. Students need educators who understand all of this and help them navigate. Competing demands, the competitive nature of HE, eroding trust between students and institutions are perhaps more important. 

Q & A ensured. 

Panel discussion followed (I will need to catch up on these in the recording) was moderated by Dr. Shannon Rios with Laura Chambers (board director at Mozilla), Dr. Solange Glasser (University of Melbourne) and Jim Hsiao (University of Melbourne).