Why you should still write yourself at university
Artificial intelligence means that previous teaching and examination methods as well as competence expectations at universities need to be renegotiated. Why should students even bother with a laborious writing process when the AI can generate texts on command? We shed light on writing at universities as a learning object and tool for thinking and demonstrate its immense value for educational processes – also in combination with artificial intelligence.
The importance of writing in university teaching
What functions does writing fulfill for the acquisition of competences at university?
Texts are the central means of communication in science. At the same time, writing texts is a thinking tool that should not be underestimated, as it can help to generate new ideas, network knowledge, and train critical thinking. This means that writing not only transports knowledge and serves information, but also fulfills an epistemic-heuristic function. Writing – including calculations, creating drawings, CAD, and other visualizations – is therefore important for all subjects in order to simultaneously develop subject-specific content and train and deepen technical thinking.
In addition to term papers and theses, which train academic writing, other types of texts such as portfolios, seminar and experimental protocols, internship, practical and reflection reports, documentation, and other practical written examination forms are also important at TH Cologne. Text types are patterns that writers use as a guide when designing the form and content of a text and that readers apply to a text as an expectation of its form and content. By reading and writing different types of texts, students learn how questions are asked and problems solved in their subject area. Your knowledge of text types will also enable you to adapt more quickly to new communication situations in both academic and professional practice. Writing in the university context in particular is characterized by a complex cognitive process that requires students to develop individual (writing) strategies. Academic work and writing require an interplay of different sub-skills (information, reading, and writing skills, but also generic sub-skills such as self-regulation or digital skills) as well as knowledge about content, language, or addressees (see figure). Thus, scientific writing aims at higher levels of learning such as analysis, synthesis, and evaluation and must be learned and taught step by step, without and in the future also with AI.T From our point of view, both should be integrated into teaching, not as additional teaching material, but as a tool in teaching practice, which specifically trains technical and metacognitive skills and thus represents a benefit for all parties.
To what extent does the use of AI in writing change the acquisition of competences?
The use of large language models (LLM) and AI-generated texts shifts the skills required for academic writing: Writers no longer write texts from scratch, but instead instruct language-processing AI to generate text and check, assess, and revise the AI-generated text elements. This phenomenon is already known in the sociology of technology (Rammert, Schulz-Schäfer, Latour, and others) and cybernetics as a shift from executive actions to human-machine interaction, as illustrated by the use of navigation devices: the machine calculates a route specified by a human. The person checks whether the planned route is realistic and meets their needs and then follows the machine’s suggestions. There is potential in this shift of competences, but it also harbors uncertainties and risks. The purpose of academic writing – even with AI – remains a person’s cognitive interest, i.e. the epistemology of the writing process. For writing at university and subsequently in professional life, the use of the THKI GPT-Lab can open up new opportunities for competence development, also concerning future skills (such as technological, transformative, or ethical competences), but this must also be practiced and reflected on together with students.
How can AI be used beneficially and responsibly for academic writing?
The writing process is shaped by academic conventions, individual strategies, and previous writing experience, but also by the tools and technologies available to create text: Notebooks, computers, and now machine intelligence can influence and relieve cognitive processes, freeing up capacity for other mental tasks. It can therefore be assumed that AI tools will influence writing and thus also scientific practices and vice versa. The extent to which academic practices and concepts of authorship will change cannot yet be predicted. The basis for scientific texts continues to be the principles of good scientific practice: The author is responsible for their own text and must deal transparently and critically with their own and others’ research and with the way the results are achieved. This makes it all the more important for you as a teacher to have students practice technical writing in your teaching so that, for example, the importance of complete and correct citation of sources, the correctness of calculations, the conformity of technical drawings and CAD representations to standards, etc. becomes clear to them. It is up to you to convince students that this is not a boring compulsory exercise or a superficial phenomenon to be neglected: Users are not only responsible for the formal correctness of the text generated by the machine. Of course, people also remain the responsible authority, for example to check,
- whether a research question in a term paper is suitable,
- whether the connection between several working hypotheses is correct,
- whether the chosen method is suitable for answering the research question,
- whether individual statements in the generated text are correct or incorrect – this also applies to visualizations–,
- whether generated statements are comprehensible for the audience and meet the expectations of the respective readers, i.e. the text type,
- whether the structure and arguments are coherent,
- whether sources specified by the machine exist at all,
- whether the sources relevant to the topic are included in the text, and
- whether the sources cited meet scientific and possibly even ethical standards.
As things currently stand, it is still essential for students to be able to master all components of the writing process without AI tools in order to use them in a targeted and reflective manner. AI tools can make the writing process easier or more efficient in many areas. Students can explore the familiar AI tools under your guidance concerning the following: Obtaining suggestions for finding topics/generating ideas, ideas for structuring texts, feedback on the self-written text in terms of spelling, punctuation, layout, citation rules, possibly expression, or even content.
One risk could be that students are overwhelmed and not trained in academic writing, especially when writing their first academic paper during their studies (in some subjects in the first semester, in others only with the final thesis). You could then have entire texts generated by ChatGPT and copy them without critically checking the formal and content-related aspects mentioned above. This raises the question for lecturers as to whether students who are now enrolled in a degree course, just like their fellow students in higher semesters, still need to be able to write a scientific text without generative AI. Alternatively, the question arises as to whether teaching and examination methods need to change fundamentally, for example by incorporating an oral examination or defense of written work. This could remain the basis for checking other people’s texts for their scientific accuracy and ultimately being able to critically assess and evaluate the information provided. This competence remains one of the essential skills that students acquire during their studies and need for their later professional life, especially as citizens of a democratic society. What does this then mean for teaching and learning at the university?
Instructions for practice
The further development of AI is very dynamic and it is therefore advisable to stay up to date. At the end of the article, you will find recommendations for current publications and videos on the topic. At the moment, a lot of research projects and their results are still pending in order to be able to make well-founded recommendations for practical application. As things currently stand (March 2024), however, we would like to give you the following five practical suggestions.
Any questions?
Feel free to contact us by mail to lehrpfade@th-koeln.de!
Links & Literature
Header-Image: © memorystockphoto/Adobe Stock