I, Creative Cyborg: An Autoethnographic Account of Poietic Rumination, GenAI, and the Act of Academic Research
Through an autoethnographic account of writing an MA dissertation, this paper argues that while GenAI serves as a peripheral tool for academic tasks, it cannot replace the essential cognitive core of "poietic rumination," urging researchers to reserve this reflective process for themselves to ensure genuine intellectual engagement.
Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer
In the modern world, the line between human and machine has become increasingly blurred. We have long accepted that our tools shape us, from the way we write with a pen to how we navigate with a GPS. In the realm of higher education, a new kind of tool has arrived: generative artificial intelligence. These systems can write essays, summarize books, and solve problems with a speed and fluency that often mimics human thought. This has sparked a fierce debate among scholars and students: is this technology a partner that helps us think better, or is it a shortcut that stops us from thinking at all? To answer this, researchers have begun to look closely at the actual process of doing academic work. They are asking what parts of research can be handed over to a machine and what parts must remain strictly human. The core question is not just about efficiency, but about the nature of understanding itself. If a computer can produce a perfect paragraph, does that mean it has understood the idea, or has it merely arranged words in a way that looks like understanding?
Ray Grewal, a researcher at the University of the Arts London, decided to investigate this question by turning the lens on his own work. He wrote a master's dissertation on the very topic of how artificial intelligence affects creativity and learning, but he did something unusual: he kept a detailed, honest diary of every time he used a computer program to help him. He recorded every question he asked the machine, every answer it gave, and how he used those answers. He then compared this diary against the final written work to see exactly where the machine helped and where it failed. His goal was to map the boundaries of the human mind when it is working alongside a powerful digital assistant. He wanted to find out if the machine could truly share the burden of thinking, or if it was only capable of handling the heavy lifting of data while the human mind had to do the actual work of making sense of it.
The study reveals that the relationship between the researcher and the machine is not a partnership of equals, but a division of labor between the periphery and the core. On the periphery, the machine proved to be an incredibly useful tool. It acted as a tireless tutor, explaining complex ideas in simple terms when the researcher felt stuck. It served as a research assistant, sifting through hundreds of academic articles to find the most relevant ones, saving hours of reading time. It functioned as a data processor, organizing survey results and creating charts from raw numbers with perfect accuracy. It even helped design the questionnaires and lesson plans needed for the study. In these areas, the machine extended the researcher's reach, allowing more work to be done in less time. It was a prosthesis that made the researcher more capable, much like a telescope helps an astronomer see further.
However, the machine hit a hard wall when the work moved from the periphery to the core of the research process. The study identifies a specific cycle of human thinking that the machine could not replicate. This cycle involves five stages: taking in information, thinking deeply about it, connecting different ideas to form a new understanding, expressing that understanding in words, and finally reflecting on feedback to improve the work. The machine could assist with the first stage by finding information, but it could not truly read or understand the resistance of a difficult text. It could not sit with a confusing idea until it made sense. Most importantly, it could not make the leap of connecting two distant theories to create a new insight. When the researcher tried to combine two different frameworks for the study, the machine enthusiastically agreed with the plan, even though the two frameworks were actually incompatible. It was the human researcher, through the slow, difficult work of thinking, who eventually realized the error and had to discard the machine's suggestion.
The machine also failed at the stages of writing and reflection. It could generate fluent sentences, but it could not write with the intent of transferring a specific thought from one mind to another. When the researcher used the machine to draft sections of the paper, those drafts were often just empty placeholders that had to be completely rewritten to carry the weight of the researcher's own understanding. Similarly, the machine could not provide the kind of feedback that changes a researcher's mind. True reflection comes from the friction of a human critic asking a difficult question or pointing out a gap in logic. The machine, by contrast, tends to be overly agreeable, offering smooth answers that remove the very struggle necessary for deep learning. The study found that relying on the machine for these core tasks would not make the researcher smarter; it would simply bypass the mental muscle-building that happens when a person struggles to solve a problem on their own.
This leads to a crucial distinction in how we should view the "cyborg" of the future. The term cyborg, often used to describe a human merged with technology, suggests a seamless blend. But this research suggests that the most effective cyborg is not the one who merges completely with the machine, but the one who knows exactly where to draw the line. The machine should be allowed to handle the logistics, the data, and the initial search for information. But the human must guard the center of the process: the deep thinking, the synthesis of ideas, and the critical reflection. If the human lets the machine take over the core, they risk losing the very capacity that makes research valuable. The danger is not that the machine will replace the human, but that the human will stop doing the hard work of thinking because the machine makes it look easy.
The findings offer a clear path forward for anyone using these tools. The advice is to name the specific mental work being done and to ask whether the machine is supporting that work or replacing it. If the task is to organize data or find a definition, the machine is a powerful ally. If the task is to understand a complex argument, connect two theories, or decide what a piece of writing truly means, the human must take the lead. The study concludes that the value of artificial intelligence in research lies in its ability to free up time and energy, not in its ability to do the thinking for us. The most important thing a researcher can do is to protect the slow, uncomfortable hours of deep thought, ensuring that the final product is a genuine expression of human understanding rather than a polished arrangement of words. In the end, the machine needs the human to give its output meaning, but the human does not need the machine to think; the human needs the machine only to help them think better.
Drowning in papers in your field?
Get daily digests of the most novel papers matching your research keywords — with technical summaries, in your language.