Senior Lector I Michael Farina on AI, Teaching, and Language Learning

To nominate an FAS faculty member to be featured in this series, please email fas.dean@yale.edu.

With the use of artificial intelligence (AI) on the rise, Michael Farina couldn’t help but wonder how AI was being used at Yale. As Senior Lector I in Italian Studies, he saw the technology’s use climbing each semester.  

Seeking an answer from Yale students themselves, Farina took to the college social media app Fizz, where students can anonymously post messages, create polls, and interact with other peers on campus. There he searched for student polls on AI: how often they are using it, and what they are using it for.

One poll, with responses from 2,000 Yale undergraduates, revealed that 86 percent of students used AI to help with coursework on a daily or weekly basis. 81 percent of respondents said they used ChatGPT on or to prepare for their finals.    

While some may see AI as a threat to academic integrity, Farina sees it as a powerful tool for learning. He has been teaching Italian at Yale since 2005, and also researches Italian language pedagogy, AI pedagogy, translation, Italian medieval and renaissance literature, and the history of Italian cooking.  

His interest in Italian studies sparked from a two-year study abroad to Italy as an undergraduate at the University of Connecticut, as well as a personal endeavor to strengthen ties to his ancestors’ culture. After graduating with a BA in Philosophy and Italian, he went onto pursue an MA in Italian Language and Literature.

Farina’s drive to learn Italian and embrace the culture was an expression of his natural curiosity—a trait that has served him well as AI has become increasingly prevalent in the classroom.  

AI in the classroom

Since 1998, Farina has taught his language classes using a flipped-classroom methodology. This pedagogical strategy has students engage with course material before class, then apply their knowledge in class through interactive activities like discussions and collaborative work. This differs from the traditional class structure in which students usually receive information for the first time during class through lectures. Flipped-classroom methodology frees up class time for activities that involve critical thinking, allowing students to develop and apply their knowledge and skills more independently.  

After experimenting with AI, Farina found that Language Learning Models (LLMs) can be great tools for facilitating the learning process. Supported by an Instructional Innovation Grant from Yale's Center for Language Study, he used ChatGPT to create custom chatbots, known as TutorBots, for Italian: one for differentiated language learning, and another for vocabulary acquisition. By being available to students 24/7, the chatbots improve the accessibility of course materials and help students practice grammar, oral proficiency, and interviews on their own time—strengthening the flipped-classroom methodology.

“Because they're able to do that now with basically a personalized tutor, in class we're able to really use what they're learning in a much more beneficial way,” he said.  

Farina has since helped expand the availability of language chatbots for other areas of study. During the Instructional Innovation Workshop at Yale’s Center for Language Study last May, Farina introduced more GPT TutorBots, available for students and faculty under the Yale Open Education Resources. Currently, the TutorBots are available in 11 languages: Chinese, Hebrew, Arabic, German, Italian, Korean, Hindi, Turkish, Spanish, Japanese, and French.  

“AI is revolutionizing the workforce, and our students at Yale need to become the most proficient users of AI in the world,” Farina said. “Our students are expected to be the future leaders of the world, and researchers, so if AI is going to be used in the workplace, we here at Yale need to incorporate it into what we're doing, so that students have very strong AI literacy,” he added. 

Sharing lessons on AI with fellow faculty

Farina's lessons on AI usage aren't limited to teaching students how to use AI. He has also been spreading his knowledge among colleagues and faculty members, including with a mini-course offered as part of Faculty Academy, a program offered by the FAS Dean’s Office initiative Scholars as Leaders; Scholars as Learners (SAL2). The courses allow FAS faculty to teach and take short courses on an array of topics from colleagues and peers.  

In February, Farina led a workshop titled “Teaching and Learning with ChatGPT,” which gave faculty insight into how students interacted with and used AI in their studies.

"Some [faculty] are, like me, enthusiastic about it and want to incorporate it right away. Others are very, very hesitant and don't want to use it. And then when you show them what it does and how students are using it, a light bulb goes off, and they very often change their perspective,” Farina said of the workshop.

“I think students and professors and people in the workplace need to approach AI with a heuristic principle. They have to look at AI as, ‘how do I use AI to learn with it?’” he said. “You can easily have it just generate an answer. But rather than just do that, how do I use these tools in such a way that I'm learning more as I go?”

Farina has done further research on AI-assisted language learning and vocabulary acquisition, and has supervised two Senior Projects in the Department of Computer Science that explored medieval manuscript transcription using Large Language Models.  

Farina credits his investigative drive to his students, highlighting the reciprocal nature of teaching. He has not only shared knowledge with his students, but is constantly learning from them as well.  

“I love teaching, and that's what gets me out of bed. Getting into the classroom, talking with my students, helping them, and guiding them to learn Italian—that's what I love to do more than anything.”