Students Require More Than AI Skills: They Need Responsible AI Literacy

As students return to Canadian colleges and universities soon, many will enter classrooms where generative artificial intelligence (GenAI) is already part of everyday academic life. Many might use tools such as ChatGPT and Gemini to brainstorm, summarize readings, revise writing and prepare job applications.

The pressing question is no longer whether students will use AI. It’s what universities will teach them about using it with judgment and accountability.

This question has gained urgency since Canada’s government launched its new national artificial intelligence strategy, AI for All. The strategy connects AI with public trust, economic opportunity and Canadian sovereignty. It also includes a National AI Literacy Initiative aimed at post-secondary students and educators.

However, learning to use an AI system is not the same as becoming AI literate. Students need more than the ability to write effective prompts or complete tasks faster. They need accountable AI literacy: the ability to explain why they used AI, assess what it produced and take responsibility for the resulting work.

I have taught professional communication and worked on curriculum innovation that connects classroom learning with changing professional practices. My recent research on prompt engineering has led me to a clear conclusion: teaching students how to communicate with AI can be valuable, but only when it develops rather than displaces human judgment.

Good prompts don’t guarantee good answers

Prompt engineering refers to how we devise and revise instructions to obtain a useful response from GenAI. A strong prompt may identify the task, audience, context, relevant evidence and desired format. These are familiar principles in communication education. Someone writing a policy brief, news release or workplace report needs to consider the audience and purpose before deciding what to say and how to say it.

Prompting can be a legitimate part of learning. A systematic review of prompt engineering in higher education found that structured prompting can support learning when students deliberately formulate, test and refine their interactions with AI. Other researchers describe prompt design as a rhetorical process shaped by purpose, context and audience, rather than a set of technical tricks.

Instructors might, for example, ask students to compare several prompts, explain why they revised them and evaluate how those changes affected the results. Used this way, prompting makes parts of a student’s reasoning visible.

Yet a polished response can still be false, biased or poorly supported. GenAI systems can produce falsehoods with unwarranted confidence. They may fabricate sources, obscure uncertainty or generate claims that students lack the subject knowledge to assess.

Therefore, we should be cautious about relying on AI systems as dependable tutors or collaborators. These systems can produce confident answers without bearing responsibility for whether those answers are true.

That responsibility remains with the person who uses the material.

What accountability requires

Calling for “accountable” AI use raises an obvious question: accountable to whom and for what?

Students are answerable to the people who read, evaluate or rely on their work. These include instructors, classmates, employers, clients and members of the public. In professional settings, AI-assisted communication may also affect patients, employees, customers or communities that had no role in deciding whether AI would be used.

Responsibility begins with factual accuracy, but it does not end there. Students should be able to explain how they used AI in an assignment, how the AI’s claims were checked, whether confidential information was entered and why the final work meets the standards of the course or profession.

UNESCO’s AI competency framework for students combines practical knowledge with ethics, human agency and responsible citizenship.

Students need enough knowledge of a subject to recognize weak or misleading outputs. Consequential claims require verification rather than acceptance simply because they sound authoritative. When disclosure is required, students also need to identify AI’s contribution while remaining clearly responsible for the final work.

AI literacy also includes knowing when not to use AI. Entering private information into a commercial platform, outsourcing high-stakes decisions or automating an assignment designed to develop a foundational skill may be inappropriate even when the technology is capable of doing so.

Students should not carry the burden alone

Accountability should not mean transferring every risk to individual students. Universities have a role in setting clear expectations and designing assessments that make students’ reasoning visible.

Instead of relying mainly on AI-detection software, instructors can also ask students to document revisions, defend their choices and reflect on where AI did and did not improve their work.

One pilot study on incorporating GenAI into educational assessments recommends instructors distinguish between tasks in which AI is prohibited, permitted or actively incorporated. This gives students clearer guidance than either blanket bans or vague permission.

Universities, colleges and schools need to consider which platforms they bring into classrooms. Requiring students to use a commercial AI product may expose their data and deepen universities’ dependence on private technology providers.

AI companies, in turn, remain responsible for system design, data practices and claims about what their products can reliably do. Teaching students to verify outputs does not absolve technology providers of responsibility for faulty systems.

Communication education belongs at the centre

Canada’s AI literacy agenda therefore extends beyond computer science programs, workplace training and tutorials produced by technology companies.

Communication and other humanities and social sciences programs already teach students to ask who is speaking, what evidence supports a claim, which audience is being addressed, what has been omitted and whose interests are served. These questions are central to evaluating AI-generated material.

They also matter in the workplace. An OECD study of changing skill demands in Canada found that occupations highly exposed to AI increasingly require communication, social, language and digital skills.

Preparing students for AI-shaped workplaces does not mean replacing these capacities with technical training. It means teaching students how to apply them when AI becomes part of the work.

Canada needs graduates who can use AI effectively. It also needs people who can challenge unreliable outputs, defend their decisions and recognize when automation is the wrong choice.

If the AI for All strategy is to build public trust rather than simply accelerate adoption, Canadian students will need more than technical fluency. They will need judgment to decide which answers to accept, which systems to question and when AI should not be used at all.