Skip to content Skip to footer

THKI GPT-Lab

TH Cologne offers its students and employees their own access to a Large Language Model: The THKI GPT-Lab. Based on an API interface, lecturers, academic staff, students and administrative staff can send requests to ChatGPT – free of charge and in compliance with data protection regulations.

In a nutshell: What is the THKI GPT-Lab?

Logo: © ZLE

TECHNICAL REQUIREMENTS

– Network of the TH Cologne, if necessary VPN connection
– Laptop, smartphone, or tablet
– Browser, no app required

The THKI GPT-Lab is a didactic interface based on the OpenAI API. The TH Cologne’s license sends centralized requests via the API without user-related data to ChatGPT from OpenAI. The Large Language Model is a powerful machine learning model that understands and generates natural language. It can analyze and understand texts, answer questions, write texts, and even hold human-like conversations. By training on huge data sets, it develops an understanding of language patterns and can then generate new texts that resemble human writing styles.

The necessary API interface was developed by colleagues at the HAWK Hildesheim/Holzminden/Göttingen, who make their entire didactic interface freely available via GitHub. In our adaptation, it is now technically possible for our version of ChatGPT to be used exclusively by university members with campusID.

Range of Features

  • Text creation and comprehension
  • Iterative prompting
  • Learning-based approach
  • Multilingual skills
  • Menu-controlled conversations

Uses

  • Lecture
  • Seminar
  • Group work / student projects
  • Academic work
  • Exams
  • Consultation hours

Step by Step

All university members can log on to https://ki.th-koeln.de/ with their campusID and their individual password. The THKI GPT-Lab is easy to access via any browser and no installation is required.

The default setting is the GPT 4 model, which offers improved performance but may be slower than older models.

The THKI GPT-Lab is divided into two application areas. In the Conversation area, you can use a chat, similar to the one you know from ChatGPT, to quickly get started with any question or task. In the Learning Space, you will understand the various support options of the THKI GPT Lab in everyday university life and learn what characterizes an effective prompt. You can learn more about scientific work, organization, creativity, and the respective cases of application.

At this point, we would like to explicitly point out that the learning space is still in trial operation. This means that the conversations stored there do not yet specifically relate to the TH Cologne context. Answers with false correlations are to be expected, as the corresponding data sets are missing. The evaluation process showed that a quick introduction is preferable to one that is complete in terms of content.

Although the protection of your personal use is guaranteed by the central API interface, we strongly advise against entering personal, sensitive, and university-internal data into the interface.

The opportunities and risks of artificial intelligence, particularly for university teaching and studies, have already been presented and discussed at our university. By providing the GPT Lab, TH Cologne wants to give its university members the technical basis to explore, reflect on, and evaluate the possibilities and limits of AI in a self-determined manner.

The Center for Academic Development (ZLE) is running the project on behalf of the Executive Board in close collaboration with Campus IT (CIT). There is a staggered introduction for the different status groups at TH Cologne. Initially, all professors, teaching staff for special tasks, academic staff and all heads of department, all employees of Campus IT, the university library, the Academy of Advanced Studies, and the Competence Center can log in with their campusID and use the service. The GPT Lab will be open for students from April and May. Administration employees are expected to be able to use the service from mid-May.

Links & Literature

We appreciate your feedback!

When developing this offer, we tried to include you as a teacher in advance. We would therefore be delighted if you would help us to further improve our offer. Please let us know what you would like, what bothers you or what you particularly like. Would you like to contact us directly? Then write to us at lehrpfade@th-koeln.de!