Exploring AI Conversation in Learning Central: Roleplay and Reflection

Sonia Maurer is a Learning Technologist in the Learning and Teaching Academy, with a strong interest in the ongoing enhancement of Learning Central and helping colleagues to use it more easily and effectively. 

Have you noticed the new AI Conversation feature in Learning Central? I’ve been experimenting to see what it can offer, and I’d like to share a few thoughts and examples that might spark ideas for your own modules. 

What is AI Conversation in Learning Central? 

AI Conversation allows you to create a simulated dialogue between a student and an AI character. You design the persona, set the scenario, and let the students interact. 

There are two modes available: 

• Roleplay, where students interact with a character and a scenario you define. 

• Socratic Questioning, which prompts students to think more deeply through guided inquiry.  

The course instructor can then review all the interactions that have taken place and optionally assign grades. You can enter a reflection question where the student can provide their thoughts and feedback at the end.  

NB AI Conversation is always framed as a type of assignment. (You need to bear this in mind, although it doesn’t have to be graded.) 

My First Experiments: Jane Austen Bot vs Careers Advisor

My first experiment was with a Jane Austen persona. I had a cozy chat with her about morality and women’s rights and ended up explaining about film and TV! At first, it was an oddly moving little fantasy, with her responding in pseudo-19th-century style. But I gradually lost the will to believe in the prose it was generating – maybe it was a tall order! (Just so you know, she happened to be hallucinating during her final illness, so had the impression of happily communicating with 2025!)  

A second experiment—a careers advisor—was more successful. The AI asked pertinent questions and gave plausible advice. However, it didn’t seem very natural to present this kind of activity as an ‘Assignment’. 

Language Partner 

There’s also exciting potential for language learning. I tried an Italian conversation partner, with the AI playing an experienced tutor who provides corrections and explanations when desired. I specified that the student could conclude the conversation by typing ‘Ciao’, and the AI would then summarise a list of 5 important learning points from the exchange.  

Professional Roleplay 

Some of the most useful applications are likely to be simulated professional situations – think teacher-parent, nurse-patient, police-witness etc. Basically, any dramatic situation you can dream up, to allow a student to practise in a safe space. These scenarios have the potential to be very useful rehearsal – but will require careful testing and honing.   

Advanced Ideas: Hidden Characters and Dual Conversations 

I recently watched a brilliant presentation by Geona Edwards of the University of Phoenix, who’s been pushing the boundaries of AI roleplay. (You can access it from this page.) 

One of his examples involved a three-way conversation, with the student playing a therapist and the AI playing both the client and a ‘professor’ character who was available for advice or feedback at any point. Another involved a hidden literary character—Atticus Finch from ‘To Kill a Mockingbird’. However, instead of playing himself, the AI was given a different public-facing name and instructed to reply as Atticus Finch. This demonstrates how using a well-known literary character (etc) can give you a ready-made shortcut to a profile. 

Geona’s advice? Start by asking: What experience would I really like the students to be able to have, and why? Then a scenario and character might just fall into place. 

Socratic Questioning: A Work in Progress 

The Socratic mode is designed to help students think more deeply, through structured questioning. However, in default mode it’s frankly a little too aggressive with its questioning! But you can control the intensity by telling it specifically how to behave. For example, you can tone it down by asking it to acknowledge and validate the student’s responses, and engage with their thoughts, before launching the next question. This could be a useful way of getting students to think around issues and foster critical thinking.   

Where Next? 

These may be the first steps in something that could become quite big in Teaching and Learning. And these scenarios are already quite powerful, as they have full access to the LLM behind the scenes. But we’re still in the early stages… 

If you are curious to investigate how we can use this now, please have a play with AI Conversation in Learning Central – and feel free to get in touch with the AI Community on General | AI Community @ CU | Microsoft Teams, or myself via maurers@cardiff.ac.uk.