Article by H.A. Siriwardana

Ms. Anitha Siriwardhana Senior Lecturer - Faculty of Business Lyceum Campus

AI has entered lecture halls, sits beside students during assignments, whispers in examination corridors, and drafts first paragraphs of essays submitted since 2023. Yet many universities continue to treat AI as either a threat to be policed or a novelty to be tolerated. Neglecting entirely the urgent pedagogical question of the decade. “Are we teaching our learners to communicate with AI effectively, ethically, and with intellectual authority the academy demands?” The answer, overwhelmingly, is ‘NO’ and the cost of that omission is already visible in the quality of AI-assisted outputs flooding universities worldwide.

Prompt engineering (practice of instructing large language models with precision, clarity, and intent) is not a technical skill. It is, at its core, a rhetorical competency. It draws on skills that academics spend careers developing in their students. Ability to frame a question, to construct an argument, to communicate a specific need without vagueness. A lecturer who has spent thirty years teaching students to write proper research questions has, without realising it, been preparing learners for exactly this. "Students who vaguely request an essay get a vague essay. Students who command one get something worth arguing with."

Researchers in computational-linguistics have noted a striking pattern across major LLMs in widespread use today. They say that LLMs such as ChatGPT, Google Gemini, Claude, Perplexity, DeepSeek and others seem to work well with assertive prompts. Some scholars, drawing on communication theory and gender studies literature, have described this assertive mode as a more "directive" or even stereotypically "manly” in style of prompting. Imperative in tone, unambiguous in expectation, specific in scope, and intolerant of vagueness. "Summarise the key tensions in postcolonial theory for a third-year undergraduate with no prior background, in under 400 words, using three concrete examples" will consistently outperform "Can you maybe explain postcolonial theory a bit?". This is NOT merely a stylistic preference; it reflects something structural about how transformer-based systems are built. They are trained to match the specificity and confidence of instructions received. Here, the pedagogical implication is quite direct. Assertiveness training, which has long lived in communication programmes, needs to be reframed as a foundation for responsible AI use. It is not a soft skill. It is, increasingly, a prerequisite today. The challenge for university academics, then, is not simply to tell students that better prompts give better results. The harder work is designing structured learning experiences that actually build assertive prompting as a practised, transferable skill.

Here is the uncomfortable bit. Many universities are still, in practice, preparing their students for the AI of 2023. The technology has already moved on twice. The trajectory from general AI (systems that answer questions) to agentic AI (systems that carry out multi-step tasks on your behalf) is no longer a matter of speculation. Tools that browse the web autonomously, write and execute code, manage calendars, and handle workflows are already in routine use. The next phase, which some practitioners have begun describing as Artificial Generative Intelligence, involves models that do not merely generate content but adapt continuously to the behavioural patterns of individual users over time. "The model learns you… and what it learns, it uses. Even when you do not ask it to." This is a development that should give every academic a pause. As LLMs become increasingly personalised, shaped by patterns of how users tend to think and write; the same prompt typed by two different users will produce meaningfully different outputs. A lecturer who sets an identical prompt for a class of forty students and expects equivalent results is working on an assumption that is no longer technically accurate. The model’s output is not a neutral document. It is, in a very real sense, a conversation, one that has been inflected by everything the system has inferred about the person on the other end. This raises questions that no institution has yet satisfactorily resolved; about academic integrity, about the equity of AI-assisted assessment, and about what it even means to treat an AI-generated text as a scholarly source. These are not hypothetical concerns. They are already showing up in practice, and they will intensify with technology development.

It also reframes what responsible AI use actually means. It is no longer a matter of citation and disclosure. It requires that students understand the nature of the system they are working with, that it is not a search engine nor a neutral tool. It is a probabilistic, pattern-sensitive system whose outputs are shaped by training data, user history, the assertiveness of the prompt, and the accumulated biases of the architecture itself. Teaching students to use AI responsibly means teaching them to interrogate the output with the same scepticism they would apply to any source, and to understand that the output they receive is, partly a meaningful reflection of themselves.

So, for universities “AI bans” is a sailed ship. They way-forward is to teach prompting as a discipline. Structured, practised, and, held to intellectual standards. Assertiveness training, critical AI literacy, disciplinary prompt design, and structured reflection on AI output quality should sit alongside academic writing in every undergraduate programme. The faculty member who incorporates prompt engineering into their module is not waving a white flag at technology. They are doing what good educators woul have always done. Meeting students at the edge of a new language and insisting that they learn to use it with purpose. If we are preparing graduates to think critically, communicate precisely and act ethically in their professions and we are not teaching them to do all three when they speak to an AI, then we are not preparing them at all.

“The prompt is the new paragraph. The question is whether we have the courage to teach it like one."