Beyond AI Detection

“Madam, I only used ChatGPT to improve my grammar.”

This was one of the responses I received when I asked a student about an assignment that seemed unusually polished. As I continued the conversation, it became clear that the student had not copied the entire essay from ChatGPT. Instead, the student had used AI to brainstorm ideas, reorganise paragraphs, improve sentence structure, and correct grammatical errors before submitting the final version.

This experience made me reflect on a question that many educators are currently debating.

At what point does using artificial intelligence become learning, and at what point does it become academic misconduct?

Across universities in Malaysia and around the world, artificial intelligence (AI) has become part of students’ daily academic routines. Applications such as ChatGPT, Gemini, Copilot, and Claude are now being used to summarise journal articles, explain difficult concepts, generate programming code, translate texts, improve writing, and even prepare presentations. According to UNESCO, the rapid advancement of generative AI has fundamentally changed how knowledge is accessed and produced, requiring education systems to rethink both teaching and assessment.

In response, many universities have introduced AI policies, while others have invested in AI detection software to safeguard academic integrity. Although these initiatives are well-intentioned, I believe they address only part of the issue. The real question is not whether students are using AI but whether they know how to use it responsibly.

As an English language teacher, I see both sides of AI almost every day. Some students use ChatGPT as a learning companion. They ask questions, request explanations, compare different writing styles, and seek feedback before revising their own work. These students remain actively involved throughout the learning process.

Others, however, simply copy the first AI-generated response without questioning its accuracy. I have encountered assignments containing references to non-existent works, quotations from authors who never wrote them, and examples that are completely unrelated to the given task. When these errors are pointed out, many students are genuinely surprised because they assumed that if AI sounded convincing, it must also be correct.

This is precisely why AI literacy has become more important than AI detection.AI is remarkably good at producing fluent language, but fluency should never be mistaken for accuracy. Unlike humans, AI does not understand information in the way we do. Instead, it predicts the next most likely word based on patterns learned from massive amounts of data. As a result, it occasionally generates information that appears credible but is entirely fabricated, a phenomenon widely known as AI “hallucination”.

Students, therefore, need to develop a new set of competencies. They must learn how to evaluate AI-generated responses critically, verify information using reliable academic sources, recognise potential bias, write effective prompts, and use AI ethically. These competencies are increasingly recognised as essential future skills. The OECD has similarly highlighted AI literacy as a critical competency for the future workforce, emphasising the importance of critical thinking, digital judgement, and responsible AI use alongside technical proficiency.

Perhaps the biggest misconception is that AI itself makes students lazy. From my observations, this is not entirely true. AI simply magnifies the learning habits students already possess. A diligent student uses AI to deepen understanding, generate alternative perspectives, and improve the quality of work before making independent decisions. A disengaged student uses the same tool merely to complete a task with minimum effort.

The technology is identical. The outcome depends entirely on the learner. This distinction is important because it changes how universities should respond. Instead of investing all our efforts in catching students who use AI, perhaps we should invest more in teaching them how to use AI wisely.

Assessment design also deserves reconsideration. Traditional assignments that primarily require students to reproduce information are becoming increasingly vulnerable in the age of AI. Rather than asking students only what they know, educators should design assessments that require students to explain how they arrived at their conclusions, justify their decisions, reflect on their learning process, and apply knowledge to authentic contexts. These are tasks that cannot simply be outsourced to AI because they demand human reasoning, personal experience, and critical judgement.

This shift is also consistent with the aspirations of the Malaysia Education Blueprint (Higher Education) 2015–2025, which emphasises producing holistic, entrepreneurial, and balanced graduates equipped with higher-order thinking skills, digital competencies, and lifelong learning capabilities. AI literacy naturally complements these aspirations by preparing graduates not only to use emerging technologies but also to evaluate them responsibly.

Equally important, lecturers should not be expected to navigate this transformation alone. Universities must provide continuous professional development that equips educators with practical strategies for integrating AI into teaching, designing AI-resilient assessments, and fostering ethical AI practices in the classroom. AI should become a catalyst for innovation in teaching rather than a source of anxiety.

The future workplace will not ask graduates whether they used ChatGPT during university. Instead, employers will expect graduates who can think critically, communicate effectively, solve complex problems, and make sound judgements while working alongside AI technologies.

Ultimately, higher education should not aim to produce graduates who avoid AI, nor graduates who depend entirely on it. It should produce graduates who know when to trust AI, when to question it, and when to rely on their own human judgement. Perhaps, it is time for universities to shift the conversation. Instead of asking, “Can we detect AI?”we should begin asking,”Are we developing AI-literate graduates who are prepared for the future?”