Scaffold or substitute? First-year students’ and lecturers’ perspectives on generative AI in academic writing at a Ghanaian university

In the quiet hum of a university library or the glow of a laptop screen late at night, a new kind of helper has arrived for students everywhere. It is not a tutor who sits beside you, nor a library book you pull from a shelf. It is a generative artificial intelligence, a type of computer program that can read your prompt and write paragraphs, explain complex ideas, or fix your grammar in an instant. For decades, learning to write in a university setting has been about more than just putting words on a page; it is a process of thinking, struggling with difficult concepts, and slowly building a unique voice that belongs only to the writer. This process, often called academic writing, is how students learn to think like scientists, engineers, or historians. Now, with these powerful tools available at the click of a button, a fundamental question has emerged: does this technology help students climb the mountain of learning, or does it carry them up the slope so they never learn to walk?

This question sits at the heart of a new study conducted at a public university in Ghana, where researchers wanted to understand how this digital helper is changing the classroom. The study focused on the very beginning of a student's journey, looking at first-year students and the lecturers who teach them. The researchers were not interested in whether students were violating academic integrity, but rather in how they were using these tools to learn. They asked a simple but profound question: is the artificial intelligence acting as a scaffold, a temporary support that helps a student build their own understanding, or is it acting as a substitute, doing the thinking for them so the student never has to? To find the answer, the researchers gathered a large group of sixty-one students and eleven lecturers. They held fourteen group discussions with the students and sat down for individual interviews with the teachers, listening carefully to their stories about how they write, how they learn, and how they teach.

The students who spoke to the researchers described a relationship with artificial intelligence that was both helpful and worrying. Many said they used the tools to make sense of difficult topics. One student explained that they would ask the computer to explain a complex subject as if they were teaching a twelve-year-old, which helped them grasp the basics before trying to write their own essay. Others used it to organize their scattered thoughts or to check their grammar after they had already written a draft. In these moments, the technology felt like a supportive tutor, offering immediate feedback and clarity when a human teacher was not available. However, the same students admitted that the line between help and replacement was thin. When deadlines loomed and time ran out, some confessed to typing a question into the computer and submitting the answer it generated without reading it or understanding it. They knew this was risky, but the pressure to finish the work was strong.

A recurring theme in the students' stories was the fear that they were trading short-term success for long-term ability. They noticed that while the computer could produce a polished essay in minutes, they often could not answer the same questions on an exam without the tool. One student described this feeling as "false hope," noting that they could submit a perfect assignment but would fail if asked to do the same work independently. They realized that if they relied on the computer to think for them, their own ability to think, research, and write would weaken. They expressed a strong desire for clear rules and training, wanting to know exactly how to use these tools responsibly rather than being told simply to avoid them. They saw the technology as a powerful resource that needed guidance, not a forbidden object.

The lecturers shared a similar ambivalence, though their concerns were rooted in their role as assessors of student learning. They acknowledged that artificial intelligence had arrived to stay and that trying to ban it entirely was futile. They recognized that the tools could help students organize their ideas and improve their writing structure. However, their primary worry was that the technology was obscuring the student's own voice. When a student submits an essay, the lecturer is trying to assess the student's mind, their ability to argue a point, and their understanding of the subject. If the computer writes the essay, the lecturer is grading the machine, not the student. They reported that some assignments looked perfect on paper, but when they spoke to the students about the content, the students could not explain what they had written. This disconnect made it difficult to know if the students were actually learning or just producing a convincing performance.

Because of this uncertainty, many lecturers described changing how they teach and test. Some stopped giving take-home assignments where students could work alone at home, fearing that the work submitted was not their own. Instead, they moved toward in-class writing, oral exams, or practical demonstrations where they could see the student thinking in real time. They wanted to ensure that the work being graded was truly the student's own. While they supported the idea of teaching students how to use artificial intelligence responsibly, they faced a practical problem: the curriculum was already full, and there was little time to add new lessons on technology. They also worried that even if they taught the rules, students might still choose to use the tools in ways that bypassed learning. The lecturers felt caught between the need to adapt to a new reality and the duty to protect the integrity of their students' education.

The study suggests that the answer is not to ban the technology or to let it run wild, but to find a middle path where it supports learning without replacing it. The researchers found that the key difference lies in who is doing the thinking. When the student uses the tool to clarify a concept, organize their own thoughts, or fix a sentence they wrote, it acts as a scaffold. It helps them build their own understanding. But when the student uses the tool to generate the entire argument or the whole essay, it becomes a substitute. In this case, the student skips the hard work of thinking, and the learning does not happen. The study highlights that this is not just a technical issue of detecting plagiarism, but a deeper issue of how students learn to think and express themselves.

Ultimately, the research points to a need for a new kind of education. Students need to learn how to use these tools critically, understanding when they are helpful and when they are a trap. They need to know how to keep their own voice at the center of their work. Lecturers need support to redesign their assignments so that they can see the student's thinking process, not just the final product. The study concludes that artificial intelligence is becoming a permanent part of the academic world, and the goal is not to fight it, but to shape how it is used. If done right, it can be a powerful partner in learning. If done wrong, it risks turning students into passive consumers of information who can no longer think for themselves. The future of writing in universities depends on finding that balance, ensuring that the technology serves the student's mind, rather than the other way around.