KI Frequently Asked Questions
Artificial Intelligence (AI) is an umbrella term for technologies aimed at analysing, automating, or simulating human cognitive processes. Generative AI refers to technical systems that are capable of independently creating content such as text, images, music, or videos. Based on an input—known as a “prompt”—these systems generate corresponding outputs, for example text in natural, human-like language. To achieve this, the models are trained on large datasets and generate content based on probabilities.
Large language models (LLMs) are based on probabilistic calculations that predict which word is most likely to follow another. As a result, the texts they generate are not always factually accurate. While general and well-known information is often reproduced reliably, the likelihood of errors increases with more specific topics (so-called “hallucinations”). These inaccuracies are typically presented in a linguistically convincing manner, making it essential to critically evaluate the content.
Paderborn University provides an AI Chat portal that enables the use of an AI-based chatbot. Through this service, teaching staff and students can submit queries to language models developed by OpenAI (ChatGPT). In addition, various open-source models (such as Codestal, LLaMA, and Mistral) are available. Only user inputs (prompts) are transmitted via an API; no personal data is shared. Students and teaching staff can activate the service via the service portal. It can be accessed within Paderborn University's network or via VPN.
The suffixes indicate where your prompts are processed. In this video, you will learn about the differences between the language models offered in the AI Chat.
Teaching staff should review the learning objectives of their course to determine whether and how AI applications can be meaningfully integrated. A useful approach is to compare what students learn without AI tools and how the learning objectives change when such tools are introduced. It is important to communicate transparently to students the reasons for using or restricting AI applications and to establish clear, context-specific guidelines. As the technology is still relatively new, it also offers opportunities for teaching staff and students to exchange ideas and collaboratively develop appropriate practices.
Prior to the start of the course, you should inform students that access to the AI chat portal can be activated via the university’s Service Center. This enables free, data protection–compliant access to AI language models designed for use in higher education. This information should be communicated at an early stage, for example via PAUL/PANDA or during the first session, to ensure timely access for all students. Where appropriate, brief instructions or a link may be provided. It is also recommended to clearly indicate in the course description on PAUL that the use of AI tools is mandatory in order to prevent later fundamental discussions.
There are specialized AI detection tools designed to identify AI-generated texts. If you are considering using such a tool, please feel free to contact us. Your contacts are Christina Flotmann-Scholz and Iris Neiske. Alternatively, you can attend the Digital Teaching consultation hour, which takes place every Monday from 9:00 to 9:30 a.m. via Zoom.
As part of their courses, teaching staff generally determine which aids are permitted (e.g. calculators) and, where applicable, which are not (e.g. programmable calculators). They may also regulate the use of AI applications for text generation. Any deviation from these requirements by students constitutes an attempt at academic misconduct.
Students must ensure that their use of AI-based writing tools is disclosed transparently in their work. The exact form of this disclosure should be agreed upon in consultation with the teaching staff. Different approaches may be adopted, such as a general statement on the use of the tool, a description of specific functions (e.g. support with structuring or language revision), or detailed references within the text, based on established citation guidelines. Further information can also be found in the Guide on the Use of AI in Term Papers and Theses.
Examination work that goes beyond highly structured formats such as multiple-choice exams—for example, term papers—is subject to copyright protection. If the data entered is stored and reused, as may occur when using AI-based writing tools, entering such examination work constitutes unauthorized reproduction and thus a violation of copyright, unless the students have given their explicit prior consent.
In addition, from the perspective of examination regulations, examination work may only be evaluated by the designated examiners themselves. Autonomous evaluation by software is not permitted.
Uploading photos, videos, or audio recordings, or entering information about individuals, is not permitted without their consent. Under Article 4(1) of the GDPR, personal data means any information relating to an identified or identifiable natural person. A person is considered identifiable if third parties are able to clearly assign the information to a specific individual, for example through direct identifiers such as names, personalised email addresses, full addresses, or photo and video recordings. Identification may also occur indirectly through the combination of different pieces of information.