Start with Learning, Not AI

Artificial intelligence is changing what teachers can create. AI can certainly save teachers time, generate resources and streamline planning. But its greatest educational value lies elsewhere. The more important question for language educators is:

How can AI help teachers create better learning for students?

Tools such as ChatGPT, Claude, Gemini, Codex and emerging AI agents can generate activities, analyse materials, create code, build interactive resources and provide feedback. These capabilities are impressive, but technology alone does not improve learning.

The value of AI depends on the pedagogy behind it.

Start with learning, not AI

A useful starting point is to identify the learning behaviour students need.

Do they need to:

• retrieve vocabulary without prompts?

• recognise language through listening?

• construct sentences independently?

• apply familiar language in new combinations?

• receive feedback and try again?

• revisit previously learned language over time?

Only after identifying that learning need should AI enter the design process.

This reverses a common approach to educational technology. Instead of asking, “What activity can this AI tool create?”, teachers can ask:

“What learning process needs to happen, and how could AI make that process more effective?”

Use AI to design practice, not just content

One of AI's most promising roles in language education is not content generation but practice design.

A traditional resource may contain ten questions. An AI-supported learning tool can create repeated variations of the same language pattern while controlling vocabulary, difficulty and support.

For example, students learning pets might repeatedly combine:

number + measure word + adjective + animal

The combinations can change while the underlying language remains familiar.

Students are therefore not simply completing more questions. They are repeatedly retrieving and recombining knowledge.

This aligns with established findings from the Science of Learning: learning becomes more durable when students retrieve knowledge rather than continually re-read it, and when practice is distributed rather than concentrated into a single lesson (Dunlosky et al., 2013; Roediger & Karpicke, 2006).

The question for teachers becomes:

Could AI help turn a one-off classroom activity into a reusable retrieval routine?

Protect the thinking

AI makes answers extremely easy to access. That can also become its greatest educational weakness. If students immediately receive translations, model sentences or corrections, the technology may remove the very thinking that produces learning. Effective AI-supported activities can deliberately delay assistance.

A student might:

listen → think → respond → reveal → compare → try again

rather than:

ask → receive answer → copy

This principle has shaped a number of interactive learning tools developed for Yeah Chinese, including Rainbow Noodle, I Draw, I Recall, Pet Parade, and topic-based apps for topics such as pets, weather, daily routines and clothing.

The technology varies, but the principle remains consistent:

students should do the cognitive work before the technology provides the support.

Before adding an AI feature, it is worth asking:

Does this feature increase student thinking - or remove it?

Use AI to control cognitive load

AI can generate almost unlimited language. Beginning learners do not need unlimited language. They need carefully selected language at an appropriate level. This is particularly important in secondary language classrooms, where students are often managing unfamiliar vocabulary, grammar, pronunciation, script and meaning simultaneously.

Cognitive Load Theory reminds educators that working memory is limited and that instructional design should minimise unnecessary processing (Sweller, van Merriënboer, & Paas, 2019).

AI therefore becomes powerful when teachers use it to constrain, not simply expand.

An AI-supported activity might deliberately use:

• only vocabulary already taught;

• one target sentence structure;

• limited visual information;

• optional rather than permanent scaffolding;

• pinyin or transliteration that can be removed;

• short feedback rather than lengthy explanations.

The important design question is:

What does the learner need to see now - and what can be withheld until later?

Use AI to create variation without constant novelty

Language learning requires repetition. Students, however, quickly recognise repetitive activities. AI makes it possible to preserve the learning objective while changing the surface of the task.

The same vocabulary might appear through listening, speaking, categorising, drawing, sentence construction, translation, retrieval games or short communicative challenges. This provides variation without continually introducing new language.

For teachers, this creates an important possibility:

Instead of using AI primarily to create the next new resource, could it be used to generate better ways for students to revisit what has already been taught?

Keep the teacher as instructional designer

AI agents will continue to become more capable. The teacher's role therefore becomes more - not less-important.

AI can generate language. It cannot independently determine which language deserves the greatest classroom attention.

AI can create hundreds of activities. It does not automatically know whether those activities produce useful retrieval, unnecessary cognitive load or shallow guessing.

AI can provide feedback. It still requires pedagogical judgement to decide when feedback should appear, how much should be given and what the learner should do next.

The most productive relationship is therefore not:

AI replaces part of the teacher.

It is:

AI increases the teacher's capacity to design learning.

Teachers bring curriculum knowledge, knowledge of learners, assessment literacy and pedagogical judgement. AI expands what can be built around that expertise.

A useful starting framework

Before using AI to create the next language-learning activity, consider three questions:

1. What should students retrieve, understand or produce?

2. What thinking should happen before support is provided?

3. What should happen after a student succeeds—or makes an error?

These questions shift the focus away from the novelty of AI and towards the quality of learning it enables.

The opportunity is not simply to make existing worksheets faster.

It is to rethink what language practice can look like when teachers can rapidly design, test and refine interactive learning experiences.

That may ultimately be one of AI's most valuable contributions to language education.

References

Dunlosky, J., Rawson, K. A., Marsh, E. J., Nathan, M. J., & Willingham, D. T. (2013). Improving students' learning with effective learning techniques: Promising directions from cognitive and educational psychology. Psychological Science in the Public Interest, 14(1), 4–58.

Roediger, H. L., III, & Karpicke, J. D. (2006). Test-enhanced learning: Taking memory tests improves long-term retention. Psychological Science, 17(3), 249–255.

Sweller, J., van Merriënboer, J. J. G., & Paas, F. (2019). Cognitive architecture and instructional design: 20 years later. Educational Psychology Review, 31, 261–292.

---

At YEAH CHINESE, we’re dedicated to helping you bring dynamic, interactive, and effective language lessons to your classroom. Want more creative ideas, resources, and community support? Subscribe to our newsletter today to join a growing community of language teachers and get access to even more tools, tips, and activities designed to make learning exciting for your students. ❤️

Subscribe to Our Newsletter