Difficult conversations with parents are part of teaching, and they can be particularly challenging for new and early-career teachers who have had little opportunity to experience them before entering the classroom.
In a groundbreaking project, researchers at La Trobe University in Melbourne are exploring whether social robotics and AI can help address a significant issue of work-related violence and aggression in Australian schools. Their approach gives teachers a safe environment to practice challenging parent-teacher conversations with an AI-based social robot before encountering similar situations in real life.
For the study, Furhat robot was used to create Fletch — a character mimicking disgruntled parents, designed by researchers specifically for these training scenarios.
Led by Professor Therese Keane at La Trobe University's School of Education, the project looks at how AI-supported simulation can better prepare new and early-career teachers for challenging interactions with parents.
To make those interactions reflect situations teachers actually encounter, the researchers interviewed 20 experienced educators about difficult conversations they had faced throughout their careers.
Their experiences helped shape scenarios around issues including student grades, behavior, bullying, chronic lateness and suspension.
Rather than working through these situations as fixed scripts, teachers can now experience them as conversations.
For the project, the Furhat social robot becomes Fletch, an AI-based parent that teachers can sit across from and talk to.
During each scenario, Fletch responds to what the teacher says and can adjust its behavior as the conversation develops. If the situation isn't handled well, the parent can become increasingly difficult or unreasonable, giving participants an opportunity to practice de-escalation and experiment with different ways of responding.
Usually lasting four to five minutes, interactions are brief but purposefully demanding. Participants have described feeling exhausted after completing them, and reflecting some of the emotional pressure these conversations can create in real life.
Using Furhat gives the researchers a physical, face-to-face element that a text or screen-based simulation cannot provide. The robot combines spoken conversation with a human-like face, eye contact and expressive behavior, while AI allows the character and conversation to respond dynamically to each participant.
This creates a controlled environment where the same type of difficult situation can be explored repeatedly, without requiring another person to perform the role of an angry parent each time.
Fletch is being developed primarily to support pre-service and early-career teachers at La Trobe University, but the project is already moving beyond university training.
Several schools in Victoria have signed on to use the technology with existing teachers, with an initial rollout planned across schools in Melbourne.
The researchers are also considering other situations where this type of embodied AI training could be useful.
Professor Keane said other uses for Fletch were already being explored, such as difficult conversations between teachers, as well as between teachers and students. Further along, Fletch could also have a role as a classroom assistant, helping identify students not paying attention.
The project provides a fresh approach to practicing one of the more challenging aspects of teaching by fusing authentic educator experiences with AI and social robotics.
