Dr Zafar Khan Safdar
Walk through a Pakistani university today and the scene looks familiar. Students sit in classrooms, computer labs are busy, professors assign essays and supervisors review theses. Yet beneath this routine, something fundamental has changed. Work that once required days of reading, searching, discussing and thinking can now be produced in minutes. ChatGPT, Gemini, Claude and other generative AI tools have quietly entered academic life.
That change is not necessarily bad. In a country where libraries can be poorly stocked, research material expensive and classrooms overcrowded, AI can be remarkably useful. It can explain difficult concepts, help students find their way through unfamiliar subjects, organise ideas, translate material and make knowledge more accessible. The problem begins when assistance becomes substitution, when a student no longer uses technology to understand something but uses it to avoid understanding it.
University life in the 1990s had a certain innocence to it. An assignment was almost a small expedition. Books were not a click away. We searched library shelves, wandered through second-hand bookshops for forgotten titles and waited patiently for books already borrowed by someone else. Photocopying was neither easy nor cheap, so every useful page mattered. We depended on the books we could find, our teachers’ lectures and the notebooks we filled by hand. We read, underlined, took notes, compared arguments and then began to write, often discovering what we truly thought only after wrestling with the words on the page. It was slower, certainly. It could be frustrating and sometimes exhausting. But there was something beautiful about that slowness. We did not simply receive knowledge. We went looking for it. We carried books, filled notebooks, argued with ideas and slowly formed our own. The journey was difficult, but the journey itself was education.
Today, a student can ask an AI system a question and receive an essay with an introduction, argument, references and conclusion within seconds. A computer science student can generate code with a prompt. A graduate student can produce a literature review without moving between dozens of books and journals. A researcher can feed pages of material into a machine and receive a concise summary almost instantly. The gain in efficiency is extraordinary so is the temptation.
The danger is not simply plagiarism. Universities have always dealt with students copying from books, classmates or the internet. The deeper problem is more difficult to see: a student can submit original-looking work without having done the intellectual work that produced it. The essay may be polished. The grammar may be perfect. The argument may even be sophisticated. But who did the thinking? That question matters because a university education is supposed to produce more than correctly formatted answers. It should teach students how to confront difficult questions, find evidence, recognise weak arguments, challenge assumptions and defend a conclusion when someone disagrees. Those abilities are not downloaded. They develop through reading, confusion, failure, revision and repeated attempts to make sense of something difficult.
If that struggle is outsourced, the paper may become better while the student becomes weaker.
This matters particularly in Pakistan, where higher education has long struggled with rote learning and assessment systems that can reward reproduction more readily than inquiry. Generative AI has arrived in a system already vulnerable to shortcuts and made those shortcuts almost frictionless. Ask a student to write 3,000 words on democracy and a machine can produce a respectable answer almost instantly. Ask that student to explain why one argument is stronger than another, defend the sources, challenge an opposing position or respond to an unexpected question, and a very different picture may emerge. The first task can measure the ability to obtain an answer. The second begins to reveal whether learning has taken place. That distinction should change the way universities assess students.
The answer is not to ban AI. Nor should universities build academic integrity systems around unreliable AI detectors. They should redesign assessment so that students must demonstrate ownership of their thinking. Written assignments can be accompanied by presentations, viva voce examinations, drafts, research diaries and classroom discussions. A student submitting a thesis should be able to explain its argument, defend its methodology and identify the evidence supporting its conclusions. A computer science student should understand the code submitted under their name. Faculty members, meanwhile, need training to teach responsible AI use and to distinguish between technology that sharpens thinking and technology that replaces it. There is an irony here. AI may force universities to rediscover what education was always meant to be.
The real challenge is not that students have too much information at their fingertips. It is that we are beginning to mistake access to knowledge for learning itself. Generative AI can summarise a debate, connect arguments and produce polished prose in seconds. But education happens in the slower work behind the answer: reading deeply, asking difficult questions, confronting ideas we disagree with, changing our minds and learning why we believe what we believe. A university degree should mark that intellectual journey, not merely a student’s ability to generate convincing answers with better prompts. If technology allows students to bypass the struggle through which thought is formed, universities may produce graduates who can write anything but understand very little. The purpose of education is not to give a student a machine that can answer every question. It is to build a mind that knows which questions are worth asking.
The writer is Ph.D in Political Science and visiting faculty at QAU Islamabad. His area of specialization is political development and social change. He can be reached at [email protected] and on X @zafarkhansafdar.
