Since 2022, a lot of general things have been said and written about generative artificial intelligence and the consequences of its spread. But I’m particularly interested in the specific cases in which AI provokes conflicts. Although…in this case we don’t yet know whether and how ChatGPT was actually used.
The Technical University of Munich has rejected an applicant for a master’s degree course. Reason: He used AI to write his application essay. The rejected party sued, and the proceedings show what questions AI can raise, as this SZ article summarizes.
Among other things, the question is how to recognize AI texts. The TUM refers to an examination software. But – as with many AI applications – no one knows how this ominous software works, what criteria it uses to recognize AI texts, and how accurate it is.
The reference to the alleged perfection of the essay is funny:
the “brevity and density of content” of the text, which differs greatly from the essays of other applicants, as well as the “outstanding linguistic quality of the text with the absence of any breaks in content or logic,” said a spokesman. In short: the essay was too good to be real.
So when AI reaches perfection, we don’t want it. The question arises: what for then?
(The fact presented by the university that the student’s application essay read completely differently last year is worth considering.
The article raises another problem: Not only do universities have different standards for how they evaluate (or even encourage) the use of AI by students. Many chairs within a university also have different standards. In short: there is arbitrariness, which is why AI may not only be a blessing for students, but also a curse.