Notes from the October 7, 2025 meeting # 47


Thanks to everyone who contributed to the last conversation on "Teaching Languages with AI: Opportunities, Challenges, and Policies for the Future." It was fascinating!


This time, our conversation focused on how we can harness various AI tools to support meaningful language acquisition while upholding academic integrity and equity. These tools offers innovative ways to differentiate learning, provide instant feedback, and create authentic practice opportunities. Ignoring AI would risk leaving students unprepared for the tools that increasingly shape their world and that is precisely the reason why this conversation was necessary.

The session began with a deep dive into Talos Learn, an AI platform for student speaking practice, presented by its founder, Immanuel Joy, an experienced AI engineer. The platform allows educators to transform any lesson into a dynamic conversation with an AI avatar by customizing the topic, required vocabulary, specific questions, and the CEFR level (e.g., A1, C1). A key feature for beginners is the ability to restrict the tenses the avatar uses, while for all levels, teachers can select regional accents (e.g., Colombian, Castilian) and control the pace of the voice. The AI is programmed to handle 'Spanglish' or code-mixing by students while maintaining its responses strictly in the target language. The platform's assessment features include immediate grammar checks and corrections and a full transcript for the teacher, highlighting "areas to watch." The platform has a strong data privacy system since it does not require personal data. Thanks, Joe, for sharing this one hour webinar. For more information, contact email: contact@taloslanguages.com. You can find an example of a real-person avatar here. 

For assessment workflow, Joe shared a 'little trick': teachers can use the GoFullPage Chrome Extension to capture a full-page screenshot of a student's conversation transcript. This image, along with a school's specific speaking rubric, can then be uploaded to a general-purpose AI tool like Google Gemini to generate detailed, custom feedback. Joe also shared his Facebook group, Langauge Teaching with AI, and another resource on integrating AI with Padlet for developing all four language skills. Check "Weaving Magic into MFL: Using AI and Padlet to Develop All Four Skills" with this accompanying blog post.

A powerful tool for teacher-created listening materials, Google AI Studio, was demonstrated as a "revolutionary" app that generates native speech audios using real-life voices. Teachers structure a dialogue, paste it into the tool, and use prompts to control the tone (e.g., "angry," "enthusiastic") and pace ("fast" or "slow") to perfectly tailor the task to student level.

Finally, the group discussed NotebookLM, a Google tool accessible via Google Classroom. It generates activities from uploaded resources, and an attendee noted using it to create an English audio overview (podcast) of a challenging text to provide students with prior knowledge. Gemini Storybook was suggested as a successful alternative for creating bilingual content. 

Thanks to everyone who attend the meeting and we look forward to seeing many of you at the World Language Conference in Basel.