World Language
A major bottleneck in teaching world languages is the lack of real-world conversational practice. Teachers may not have the time to engage in verbal conversations with every student, which can make it challenging for students to improve their conversational abilities.
With Flint, teachers can design verbal conversations in any language (e.g. French) between the AI and the student, thereby providing each student with conversational practice that is perfectly tailored to their difficulty level.
Here, we’ll explore the use case of a French teacher looking to help their native English-speaking students with conversational skills as well as vocabulary related to food.
Designing an AI tutor in Flint for this use case is quite simple. As you can see below, the teacher has set an ACTFL level, provided a learning objective, and has uploaded a menu of a French restaurant (optional):
Flint uses AI to automatically generate rules for the AI waiter to follow, along with a rubric for the AI to use to give feedback to students.
All aspects of the experience can be customized. For example, we can use the “revise” feature to change the rubric from A-F to a 1-4 proficiency scale.
By starting a session with the AI tutor, students can engage in a verbal conversation in French with an AI waiter. Throughout the course of the conversation, the AI will communicate entirely in French at the ACTFL Intermediate level.
The student can click the microphone button to speak to the AI, and the AI will speak back to them in French. The entire conversation is also transcribed automatically.
Throughout the conversation, the AI waiter will encourage students to order food from the menu provided by the teacher. The AI waiter will politely correct grammar mistakes and vocabulary, while continuing to move the conversation forwards.
Once the student submits their session, they’ll get immediate feedback highlighting their strengths as well as any areas for improvement. This data is available to the teacher, as is the audio recording of the entire conversation.
Other use cases
See all use cases ->
