Bringing abstract concepts to life with GenAI

We co-teach into an interdisciplinary post-graduate unit called Innovation to Influence System Change, where students work in teams on real world problems like nature positive cities, sustainable transport or health and wellbeing in the community.

Students come up with policy or innovation that will create systemic change within these themes. These are intended to be as realistic as possible, but what we find is that students sometimes get fixated on designs where the logic of how they work in the real world isn't going to be all that effective, or they design solutions that are really only going to work for people like themselves. So we ask them to use a tool called Theory of Change, which makes them articulate what's going into their project or innovation, and how that will logically create outcomes in the system and identify the gaps and assumptions so that the idea can be improved.

But students often find it quite abstract, a bit technical and a bit dry. So we really wanted to bring it to life by using AI simulations.

We approached Joel Moore and the ATLAS team, who have been creating an AI simulation platform here at Monash, to create an avatar that would help guide the students through their theory of change process.

Meet our resident alien Abuzaharen, who comes from a distant planet called the Xanaxia!

We created a conversational “Socratic” chat bot trained on both literature and practice guides that were open access for the Theory of Change methodology.

So rather than the students asking it questions and then the bot giving answers, it was the other way round, the bot was trained and given rules on how to both inquire with students about their understanding of theory of change first, and then step them through the process of articulating the logic of their policy innovation or the proposal that they were working on.

It asked them to explain their rationale, and it resisted giving them easy answers or offering them shortcuts. We wanted them to go the long way round and give us the best idea that was possible and this AI implementation supported that.

We realised this was a chance for students to engage with AI, not as a task completion tool or a shortcut, or a grammar editor, that are common uses right now, but as an AI tool that asked them to think more critically.

Student journey

	Students receive a link to the platform and then a log in.
	They open the simulation, and Abuzaharen pops up on the  screen. The students then click a button and start speaking into the laptop.
	Abuzaharen will respond, both by talking back to the students and changing its expression to fit the context. You could see the students responding to that. They might have initially thought it was going to be a written communication, but once they saw how effective the responses were, they really got into it - comparing and sharing information and strategising the questions to pose to Abuzaharen.
	The session lasted for 20 minutes or until the students chose to end it.
	A report which summarises the session and provides useful information, such as, highlighting gaps in their logic and assumptions that they maybe hadn't considered, can then be downloaded.

Also, one of the things we did was that we used the Atlas platform in the class itself, with the students getting into their project teams to use it, whereas previously  the ATLAS platform was largely used by students individually and outside class. It was great to see how the students huddled around the computers, engaging with Abuzaharen and each other and having fun.

In our case, we built Abuzharen around the theory of change, but you could customise the avatar and theme to your unit needs. The verbal communication and also the changing of emotions of the avatar are really creative and engaging aspects.

The ATLAS team has a pretty amazing set up where you essentially fill in a template with some guardrails that include:

• What kind of training material?
• What kind of behaviour do you want the bot to have?
• What is its character and backstory?
• What will it say directly to students?
• What won't it say?

The ATLAS team took what we provided and made a prototype and then gave it to us to do some testing. We invited our colleagues from the Monash Sustainable Development Institute education team to come along and test the simulation and give it a good investigation. Some took an easy approach, some really tried to push it. Some of the feedback was in-depth, some of it was simpler, like the change in the tone of voice. We gave that feedback to the ATLAS team and they took it on board, made the adaptations and it turned out to be a really good platform.

We just want to give a big shout out to Joel, Amna and the ATLAS team for working with us on this project. It was a great experience for us. We really enjoyed the creative and collaborative nature of it, and it has taken us so much further on our journey using AI in the classroom.

Don't be afraid to introduce AI into your own classrooms. We have seen the potential of it. We have seen how it can be effective and impactful. So give it a go and enjoy the journey yourself!