Artificial intelligence (AI) is set to fundamentally change the world of work. To prepare students, Macromedia University is now radically incorporating the technology into its curriculum. But how is this changing students’ work and performance? An analysis of student papers reveals unexpected findings.
Catharina Wobbe feels her life has been made easier. “I believe we’ve reached a time when AI can be seen as just another colleague, even if it’s not physically in the office,” says the 25-year-old, who is in her sixth semester of Media and Communication Management at the private Macromedia University in Munich. AI is “incredibly helpful,” especially during brainstorming, Wobbe adds.
Her university is pushing for AI integration into the curriculum. However, not all students are enthusiastic. A university-led analysis of 10,318 project papers and 590 bachelor theses from the summer semester of 2023 found that around a quarter of students rejected the use of AI—despite explicit recommendations. Moreover, according to the data, there is no performance advantage. The study shows that using AI does not lead to better grades.
ChatGPT was released in November 2022, and by the following summer semester, AI was on the curriculum at Macromedia University. Initially, its use was only recommended; later, in some modules, it became mandatory. The university is serious about this. Students who refuse to use AI risk losing points. This makes Macromedia, which offers programs in media and humanities across eight German cities, one of the first universities in Germany to adopt the new technology so radically.
The goal is to prepare students well for their future careers, says Joschka Mütterlein, Vice President for Teaching and Faculty Development and author of the study. Universities should actively engage with AI, as it will “never be as bad as it is today.” It is also clear that the professional world will not remain as it is today.
The analysis of the project papers, however, shows that this message has not reached all students. Mütterlein divides the students into five groups: the hesitant, the pragmatic, the innovators, the cheaters, and the confused. The innovators are those who use AI the most, but they account for only 2.4 percent of the submitted papers in which the technology was used extensively.
There is also no measurable advantage from using AI. The failure rate for papers involving AI remains at 19.4 percent, consistent with the overall rate. The incidence of plagiarism has not changed with the use of AI either. Only 0.5 percent of students fall into the category of cheaters who try to disguise their use of AI. The conclusion is clear: The technology does not improve performance; the responsibility for quality work still lies with the students themselves.
Younger generation more easily influenced
Since AI use became mandatory in selected courses, the requirements for a paper include, for example, that at least two pages must be written by AI and a AI-generated image must be used. Five points are awarded for each. In proportion, this accounts for ten percent of the overall grade. Despite these requirements, only 72 percent of students used AI in such cases. This means a quarter of the students accepted a point loss and boycotted the requirement.
According to Mütterlein, there are several reasons for the reluctance to use AI: Some students simply enjoy writing themselves. Others were not aware of the option to use AI. “Of course, there are those who are not tech-savvy or simply don’t feel like it. Not everyone is innovative just because they are young.”
Student Wobbe offers another explanation: The younger generation may be “tough and brave” and stand up for things clearly. However, many young people today are more sensitive. “We are more easily influenced, especially by social media. Many universities have banned the use of AI from the start.” When a topic is immediately met with skepticism, it often results in rejection. Incentives and support systems, like those commonly practiced at the university in Berlin, can be helpful.
Matthias Jaroch from the German Association of University Professors and Lecturers also acknowledges that “AI can deliver great results, but it depends on the prompts.” Prompts are the instructions or inputs given to the AI. But what makes a good prompt? “A prompt is the question or command I give the AI,” says Wobbe. It is important to provide as many details as possible, ideally integrated into a complete sentence. The quality of what you offer the AI determines the quality of what it produces.
“We want a reflective use of AI”
Although AI simplifies information retrieval, there is no guarantee of the usefulness of the result. Because AI sometimes misunderstands things or is uncertain, it is important to learn how to critically assess information, as AI cannot do this, says Jaroch. At Macromedia, AI is not intended to replace students’ work but to serve as a potential resource. “We want a reflective use of AI; students must ask themselves: Is the source reliable, or do I need to research the matter myself again? To this end, we are changing our entire study architecture,” says Mütterlein.
Students are responsible for evaluating the produced information, which is where it gets tricky. If the AI provides incorrect information and students use it as a legitimate source, they will lose points.
The question of whether the traditional term paper is outdated is also raised by Matthias Jaroch. After all, AI can write entire texts on its own. Whether texts are original or generated is difficult to verify. In the end, it comes down to a dispute between the student and the professor. Especially in the humanities, there is skepticism about the use of AI; in computer science, the approach is somewhat more open.
Macromedia University does not see AI as a threat to traditional examination formats. The goal is to work with the technology rather than demonize it, says Mütterlein. AI can be directly quoted, or its text used, but the “production” of the generated text must remain traceable. There are specific guidelines for the introduced AI directory.
The prompt, the AI response, the student’s evaluation, and the final part used must be disclosed. Understanding how AI operates is crucial here because only those who can handle it confidently will get good results—and possibly even better grades than before.
