Generative AI for Language Learners

Can Generative AI Become a Better Study Buddy?

This eight-week research project explored how university students use Generative AI while learning a new language. Our team of four UX researchers examined where the experience helps, where it breaks down, and how AI could better support deeper language learning rather than simply generate answers.

Background & Challenge

Students increasingly use tools such as ChatGPT, Google Translate, and Quizlet to summarize content, explain concepts, generate examples, and prepare for exams. But language learning requires more than correct answers: it depends on grammar, tone, sentence structure, context, and real communication. We asked how AI could become a study buddy for language learners instead of another answer generator.

Research Process

We used a mixed-method approach: secondary research into AI in education, language learning, intelligent tutoring systems, conversational AI, and human-AI collaboration; an initial survey of usage frequency, preferred tools, confidence, and study behaviors; and semi-structured interviews with Spanish learners at Thomas Jefferson University from medical, animation, and undergraduate programs.

What We Learned

Students did not replace traditional resources with AI. They combined AI with professors, classmates, textbooks, dictionaries, YouTube, Google Translate, and Quizlet. AI was valued for speed: generating examples, explaining vocabulary, translating words, preparing dialogue, and creating flashcards. Yet human guidance remained essential for grammar, sentence structure, speaking practice, feedback, and clarification.

Trust and fragmentation

Participants rarely accepted AI responses without verification. They cross-checked with professors, friends, textbooks, and online sources. Learning also became fragmented as students switched between ChatGPT, Google Translate, Quizlet, textbooks, instructors, YouTube, and dictionaries during a single study session.

Journey & Persona

The learning journey moved from previous experience with traditional methods, through quick AI assistance and practice, into verification and reflection. Justin, a 21-year-old medical student learning Medical Spanish, represents the need for reliable explanations, context-aware examples, speaking practice, personalized feedback, and confidence for future patient interactions.

Storytelling & Core Insights

Justin’s story captured a recurring concern: “Am I really learning, or am I just getting answers faster?” Four themes surfaced across the research. AI feedback was often superficial, showing what was correct without explaining why. Responses could be inaccurate or overgeneralized. Context, tone, and situational awareness were often missing. And fragmented tool-switching interrupted learning rather than integrating it.

The AHA Moment

The breakthrough was not simply that AI could be inaccurate. Students were not asking AI to replace teachers; they wanted a better learning companion. They needed tools that explain reasoning, adapt to learning level, understand context, support conversations, and build confidence.

Opportunity Areas & Concept Direction

Future AI-assisted language learning should explain rather than only answer; personalize vocabulary, difficulty, and explanations; adapt to conversational context; encourage active practice through dialogue and reflection; and build trust with confidence indicators, reasoning, and references. We envisioned an AI Study Buddy with personalized paths, grammar explanations, context-aware conversations, interactive speaking practice, adaptive feedback, guided reflection, and progress tracking.

Impact & Reflection

This study deepened my skills in research planning, interview moderation, qualitative synthesis, affinity mapping, journey mapping, storytelling, insight generation, and opportunity identification. It also reinforced that UX research is not about validating assumptions—it is about uncovering the human needs behind behavior. The findings can guide future educational AI systems toward more transparent, personalized, and supportive language learning.