Teachers working on the front lines of education have likely heard students actively discussing AI recently. A colleague once asked me with concern, “What should I do if students use AI to do their homework?” In fact, the emergence of generative AI (such as AI) is quietly transforming the classroom. We might as well take a rational approach, combine it with our own teaching experience, and explore the principles, limitations, and potential of this new tool together.

First, it is important to understand how AI works. Simply put, AI is a model trained on large amounts of language data, capable of generating seemingly reasonable responses based on prompts. However, it does not truly “understand” content, nor does it have the ability to judge whether something is true or false.

In other words, AI is good at imitating how humans use language to express knowledge, but it may still make mistakes in facts or logic. For example, when someone asked AI a simple math question, “Which is greater, 9.11 or 9.9?”, even elementary school students know that 9.9 is greater than 9.11. Yet AI might confidently answer that “9.11 is greater.” Generative AI is not omnipotent—it can make surprisingly absurd mistakes, and both teachers and students must act as gatekeepers.

Given this, what potential does AI have for integration into education? In practice, it can serve as a powerful teaching assistant.

In language classes, teachers can use AI tools to generate questions and guide students. For example, I once tried using an AI tool to create reading comprehension questions for a text and found that the perspectives it offered were quite novel, helping to stimulate students’ thinking. At the same time, teachers can ask AI to provide summaries of paragraphs or extended explanations of a text as a starting point for classroom discussions. When teaching writing, AI tools can act as an “idea generator”: when students get stuck, teachers can guide them to ask AI for writing ideas or structure, and then have students critically evaluate, select, and revise the output themselves, thereby building their confidence and organizational skills in writing.

In math classes, AI tools can play the role of a problem-solving assistant. When preparing lessons, teachers can ask AI to generate various types of practice questions, quickly building a question bank. When grading assignments, if there are complex calculation processes, teachers can input students’ solution steps into AI to help check for possible errors or provide hints. However, it is important to remember that AI is not particularly strong in numerical calculations. As mentioned earlier, it has even made mistakes in comparing decimals, so teachers must always double-check whether its answers are correct.

From the above examples, we can see that AI has great potential in teaching: generating questions, assisting with grading, providing multiple perspectives, and even acting as a dialogue facilitator. All of these can reduce teachers’ workload and enhance students’ learning experience. However, it is equally important to pay attention to the risks and limitations involved.

First is the issue of plagiarism and academic integrity. Students today are very clever and may directly copy AI-generated answers into their assignments, making it difficult to assess their true level of understanding. Is banning AI the solution? Probably not that simple. A better approach is to educate students to take responsibility for their own learning: helping them understand that generative AI may sometimes produce biased, discriminatory, or incorrect information, and therefore its outputs should never be copied without thinking, nor fully trusted. Instead, students should be guided to treat AI as an assistant rather than a shortcut for cheating. For example, students can use AI to check the grammar of their writing and then revise their sentences accordingly, or use it to get hints for the next step in solving a math problem, while still completing the final steps and verification themselves. In this way, students develop both self-review and critical thinking skills while using AI.

Second is the issue of accuracy and data bias in AI-generated answers. As mentioned earlier, AI may sometimes confidently provide incorrect answers, and its knowledge base may not always include the latest curriculum standards or local cultural context. If teachers do not carefully review the content it generates, incorrect information may be brought into the classroom. Therefore, when applying AI in teaching, teachers must take on the role of a “supervisor”: testing and verifying the information in advance, asking questions in different ways when necessary, and ensuring the reliability of the answers. If AI responses are found to be inaccurate or inappropriate, teachers should proactively explain the issues to students and even use these mistakes as teaching materials to develop students’ ability to evaluate information. As for potential bias—for example, certain assumptions reflected in AI’s examples or wording—teachers should point these out in a timely manner and remind students to maintain a diverse and inclusive perspective when interpreting AI-generated content.

With the rise of generative AI, the role of teachers is also evolving. It is gradually shifting from being a transmitter of knowledge to a designer and facilitator of students’ learning experiences. With AI assisting in lesson preparation and answering questions, teachers can devote more time to observing individual learning differences and guiding students’ thinking. In this process, teachers must also uphold professional ethics. For example, when students make extensive use of AI, there is a responsibility to adjust assessment methods accordingly—reducing reliance on standard-answer-based testing and instead adopting open-ended questions or project-based inquiries that encourage students to use AI to support rather than replace thinking. At the same time, teachers should remind themselves that no matter how powerful AI becomes, it cannot replace the care and value guidance that teachers provide. Especially at the primary and secondary school levels, teacher-student interaction and emotional connection remain fundamental to education. AI should only serve as a supporting tool, not the main driver of education. Teachers must also adhere to ethical boundaries when using AI, such as respecting students’ original work (not allowing AI to replace creative assignments) and protecting students’ personal data by avoiding unnecessary input into AI systems. These are all emerging ethical considerations in modern education.

Finally, establishing correct AI literacy among teachers is essential in order to guide students to grow healthily in the AI era. Teachers themselves need to continue learning, understanding the basic principles and limitations of AI, and recognizing that it is a tool rather than an authority on knowledge. AI literacy should also be incorporated into the curriculum—for example, introducing how AI works, discussing its potential biases and ethical issues, and teaching students how to interpret and verify AI-generated results. This helps cultivate digital citizenship and enables students to make responsible choices when using AI. In the AI era, students who only rely on tools without developing independent thinking may become the first to be replaced by AI. Our role is to help them balance “using AI effectively” with “maintaining independent thinking,” so they can stand firm in the tide of technological change.

Generative AI brings both challenges and opportunities. We should embrace this new technology with a proactive yet cautious attitude. In the classroom, we can boldly experiment with AI-assisted teaching activities while maintaining professional judgment and avoiding blind reliance on technology. Start small—for example, using AI to assist in grading vocabulary exercises or generating a few extension questions for next week’s math lesson. Don’t be afraid—after a few attempts, you will find the approach that works best for you.