Allena
A simulated classroom, populated by students brought to life by artificial intelligence, where teachers can practise handling difficult situations before they meet them for real.
Allena (Italian for "train" or "practise") is a classroom simulator for teacher training. The idea is simple: pilots train in a simulator before they fly, doctors practise on cases before they see patients; teachers, by contrast, learn to manage a class almost entirely on the job, with real classes and real students. Allena aims to offer a gym where making mistakes costs nothing. It is a work in progress: for now it is described here, not yet available.
How it works
The teacher enters a simulated class of 26 students and faces a scenario: the start of a lesson, a conflict, a class that isn't following. The teacher speaks to the class and the students react, each in their own way.
- Eight students have a full personality, built in layers: character, motivation, their relationship with school, with classmates, with the adult in the room. The others are "extras" with a lighter profile, who bring the class to life.
- The class has a memory: attention, mood and relationships among students change during the lesson according to precise rules, not at random. So the same choice by the teacher produces consistent effects, and you can understand what caused them.
- Relationships matter: friendships, groups and tensions among classmates shape how the class responds.
- Ways of communicating: every move by the teacher is read through Eric Berne's Transactional Analysis, which describes the roles and psychological "games" that arise in relationships, the classroom included.
The debriefing
At the end of the simulation Allena provides a debriefing: what happened, which moves helped the class and which pushed it away, what could have been done differently. This is where the real learning happens, like watching the replay after a match.
How I'm building it
Allena is the most ambitious of my projects, so I'm developing it with a rigorous method, Spec-Driven Development: first the principles, then the specifications, the plan and the tasks, and only at the end the code. The students are animated by artificial intelligence models, but their behaviour is governed by a deterministic rule engine, because a simulator has to be believable and repeatable. The first phase includes a full class, one scenario and the debriefing; later on will come scenarios with just a few students, such as one-to-one help sessions or catch-up groups.