Integrating Artificial Intelligence into EFL Teaching: Practical Tools for Scaffolding and Feedback

The rapid development of artificial intelligence has fundamentally transformed the landscape of education, offering new opportunities for more personalized, adaptive, and engaging learning experiences. In the field of English as a Foreign Language (EFL) teaching, AI tools are increasingly present both inside and outside the classroom [1]. Rather than restricting their use, educators are encouraged to embrace these technologies as pedagogical assets teaching students not only the language, but also the skills to interact with AI critically and effectively.

This paper presents practical experience of integrating three AI-powered tools into an EFL course at the tertiary level, examining their potential for scaffolding learning, providing formative feedback, and fostering learner autonomy. The tools explored are Google’s NotebookLM, Gemini, and Anthropic’s Claude, each applied in distinct but complementary ways.

NotebookLM: Source-Grounded Content Creation

NotebookLM was used as a research and content synthesis tool. Students uploaded selected academic sources, such as articles, lecture notes, and textbook excerpts, and used the platform to generate structured video overviews, summaries, and targeted presentations on specific aspects of a topic. This approach supports scaffolded reading by allowing learners to interact with complex texts in a guided, manageable way. The ability to limit the tool to pre-approved sources also ensures academic integrity and encourages critical engagement with authentic materials rather than generic AI-generated content.

Gemini: Interactive Quizzes and Formative Assessment

Gemini was employed to generate interactive quizzes based on course content. The tool’s conversational interface and structured output formats allow for the rapid creation of vocabulary checks, grammar exercises, and comprehension tasks. Students found the format intuitive and engaging, while instructors benefited from the ease of adapting quiz parameters to target specific learning objectives. This application aligns with the concept of AI-assisted formative feedback, where learners receive immediate, low-stakes responses to their performance, supporting self-regulation and continuous improvement.

Claude: Building Educational Applications

Claude was used for a more advanced task: creating simple educational applications tailored to the course curriculum. Students and the instructor collaborated to design interactive learning tools, such as dialogue simulators, writing prompt generators, and vocabulary practice apps. This process required learners to articulate their communicative goals in English, negotiate task parameters, and evaluate the output, all of which are authentic language use activities. The experience demonstrated how AI can serve not only as a tutor but as a creative collaborator in the design of learning environments.

Discussion: AI as Scaffolding, Not a Shortcut

A central finding of this study is that the pedagogical value of AI tools depends largely on how they are introduced and framed. When students understand AI as a scaffold, a temporary support structure that helps them reach language goals they could not achieve independently, they engage more thoughtfully and critically [2]. Conversely, unrestricted or unguided AI use risks replacing genuine language production with automated output.

The key pedagogical principle emerging from this experience is transparency: students should know what the tool does, why they are using it, and how to evaluate its output. When AI is embedded into task design with clear objectives, it functions as an effective feedback mechanism, a research assistant, and a generator of authentic language input, all roles that support rather than undermine language acquisition.

Rather than debating whether to permit AI in language classrooms, educators should focus on developing students’ AI literacy alongside their language skills. This means modeling responsible use, building critical evaluation habits, and designing tasks that require learners to go beyond the AI’s output, to analyze, adapt, and create.

Conclusion

The integration of NotebookLM, Gemini, and Claude into EFL instruction demonstrates the practical potential of AI as a pedagogical tool when used intentionally and reflectively. These tools offer meaningful support for scaffolding, feedback, and learner engagement, provided they are embedded within a clear instructional framework. Future research should examine the long-term effects of AI-integrated EFL instruction on language proficiency and learner autonomy.