Developed by Audrey Picard at ESCP London Campus, Amélie uses Socratic dialogue to help students engage more deeply with feedback.
Amélie Socratic AI Tutor, an educational AI project developed at ESCP Business School, has been shortlisted for the AI Innovation in Education Award at the QS Reimagine Education Awards 2026.
Led by Audrey Picard, Associate Head of Languages and AI Coordinator at ESCP's London Campus, the project explores a simple question: what if generative AI helped students think through their mistakes rather than correcting those mistakes for them?
First developed in 2024 and tested in French courses at ESCP London Campus, Amélie is designed to turn written feedback into an active learning process. Instead of rewriting a student's work or supplying the correct answer, the AI tutor asks questions, offers targeted hints, and requires students to explain their reasoning before progressing.
Even though we are facing an unprecedented learning revolution with the evident risk of cognitive offloading, there is a way for AI to bring something positive for our students' progress. With the right boundaries, framework and instructions, teachers can harness the tools for a better learning experience and outcomes. This nomination has already started some fruitful conversations with colleagues, an exchange of practice that will nourish further experiments.
AI London Campus Coordinator, ESCP Business School
An AI tutor that keeps students thinking
Generative AI can correct or rewrite work in seconds, allowing students to reach a better answer without necessarily understanding their mistakes. Amélie is designed to avoid that shortcut.
Using students' original work and teacher feedback, the chatbot asks them to identify errors, guides them towards their own reformulation, and requires them to explain their reasoning rather than supplying the answer. The aim is to preserve the intellectual work involved in learning from feedback while using AI to provide personalised guidance when students need it.
The process also gives educators greater visibility into how students learn. Conversations with Amélie are recorded and incorporated into assessment, allowing teachers to see the reasoning behind students' answers and identify where further support may be needed.
Early evidence from the classroom
Amélie has been developed iteratively over two academic years with postgraduate students studying French at ESCP London Campus, with plans to extend the project to all postgraduate French classes during the 2026–27 academic year.
Early results are encouraging, with students using the more structured version of Amélie showing improved performance between coursework and final examinations. Given the small cohorts involved so far, further testing will help assess these results as the project develops.
Student feedback has also been positive, suggesting that although working with Amélie takes longer than asking a conventional AI tool for a correction, the process encourages students to reflect more carefully on their mistakes.
Keeping people at the centre of educational AI
Amélie keeps distinct roles for the student, AI and educator. The student produces and justifies the answer; AI provides individualised guidance; and the teacher retains oversight of learning and assessment.
The next phase of the project will explore ways to give educators greater visibility over student interactions while reducing the time needed to review transcripts. This includes plans for a dashboard summarising student reflections and highlighting interaction patterns for teachers to examine.
AI in service of learning
The project forms part of ESCP's wider work on the responsible use of generative AI in education, including ChatGPT Edu, training and classroom experimentation.
Amélie's principle remains simple: use AI to support students' thinking, rather than do the thinking for them.
Relevant links
Campuses