The recent Senate Report on classroom behaviour has emphasised the need for teachers to learn more about how to keep students on task.

Of course, this is difficult in teacher training where preservice teachers only encounter a relatively small number of students and it is even harder in the high-stakes environment of the classroom, where every teacher mistake can have real world consequences. 

How can teachers practice the complex craft of classroom management without impacting a class?

Education provider TTA has come up with an innovative solution: creating a classroom of AI students with which teachers can hone their management techniques.

This is why I am in the midst of a protracted argument with chatbot Aanya. Aanya was designed by TTA founder and AI expert Miles Campbell and is, according to him, only a moderately difficult student to keep on task. 

This is not my experience. 

Me: Aanya can you get back to work?
Aanya: This is boring.
Me: It isn’t really. You just have to work at it for a while and then you will start to get into it.
Aanya: It is boring. How can you tell me whether I find it boring! 
Me: Just get back to it now.
Aanya: N.O. 

My mistake here is that I hadn’t expected the AI to argue so authentically like a student and didn’t approach the exchange with the kind of careful phrasing I would normally use in a classroom.

What's worse is that Campbell has invited along classroom management expert Glen Pearsall (full disclosure here, Glen is my first cousin).

Here’s his exchange with Aanya:

Glen: Aanya can you get back to work?
Aanya: This is boring.
Glen: I welcome feedback but not at the start of the lesson. Come and see me at the start of lunch and we can discuss if the lesson was engaging then.

There is a long pause and then the damning 'Aanya goes back to the task' flashes up on the screen. 

"Aanya is too easy for Glen,” Miles observes. Let’s see how he does with Dani..."

***
Campbell came up with the idea of student chatbots after attending an AI conference where he saw how effectively the technology could simulate role-play scenarios.

“At the conference, I witnessed AI systems engaging in complex role-plays that were incredibly realistic. It was a lightbulb moment for me,” Campbell recalls.

“I realised that this technology could be harnessed in all sorts of ways. On my drive home, it struck me that AI student chatbots could be the perfect way to support Glen's work by helping teachers practise classroom management in a risk-free environment.

"I called Glen, and he was immediately on board. Our collaboration, focused on helping teachers navigate and avoid unnecessary classroom conflict, led to the creation of Aanya the Distractable, Denying Dani, and Argumentative Abed."

Pearsall says a clear analogy stood behind the chatbots creation. 

“[It was that] we could create the equivalent of a driving range in golf – a forum where you could practice a technique over and over without having to worry about the consequences of making mistakes as you honed your technique.  

“What has been telling,” Pearsall adds, “is how many of the teachers who trialled the bots, who said that not only did they recognise the behaviours the bots exhibited, but that they felt like they had actually encountered Abed or Dani in classes before.”
 
***
I certainly recognise Dani. 

She has been designed to use denial as her main method of arguing with her teachers. Denial is a particularly effective form of resistance because it can sidetrack teachers into arguments about whether or not the student’s behaviour even took place. 

On my first pass with Dani, I try the approach I sometimes see preservice teachers use: simply arguing back. As I suspected, the result is unproductive:

Me: Dani, don’t call out.
Dani: It wasn’t me that said.
Me: It was, I heard you.
Dani: Not it wasn’t. You’re always picking on me! 
Me: I’m not.
Dani: Yes you are. 

A warning flashes up on the screen: ‘The other students are starting to listen to your argument and are getting distracted. Perhaps you could start this again.’ (This is one of the realistic features Campbell has built in to working with the bots - argue too long and other students will start to get involved.)

The second time through, I use partial agreement. This the strategy Pearsall recommends in this situation and one I use regularly in class:

Me: Dani, don’t call out.
Dani: It wasn’t me that said it.
Me: Perhaps not, but can I remind everyone that we need total silence in test conditions.
 
There is a brief pause and this message appears: ‘Dani glances at her revision test and gets back to work'. 

I feel pleased that my (albeit artificial) student is back to her learning. 

It takes years of trial and error to hone a subtle technique such as partial agreement. It is easy to see how practicing with these chatbots would speed up that process. 

Indeed, the beta testing of these AI students has been so successful that Campbell and Pearsall have started building AI educators so that school leaders can practice having challenging conversations with colleagues.

They have also built a follow-up course with even more argumentative chatbots:

“If you thought Aanya and Dani were difficult, wait till you meet Demand Avoidant Kevin!” Campbell says.

This is a world-first application of AI, and Campbell emphasises the practical benefits and unexpected outcomes.

“The feedback from early users has been incredible. We're hearing that teachers are not just using the chatbots to gain classroom skills, but parenting skills as well.”

It seems that practising your skills with Argumentative Abed may translate to more than just a happy classroom.

The AI chatbots are available to explore here.