ChatGPT at Fulda University of Applied Sciences
AI-supported text creation in the form of ChatGPT-4 is accessible to all members of Fulda University of Applied Sciences via the HAWK interface. You can access this advanced AI via the following URL: ki-chat.hs-fulda.de
Please note that you must be within the VPN of the Fulda University of Applied Sciences during use and need your fd ID for access. No personal data will be stored or passed on.
The HAWKI interface was developed by the University of Applied Sciences and Arts Hildesheim/Holzminden/Göttingen (HAWK). It enables dialogue with the AI and offers various functions:
1. conversation: A chat area similar to ChatGPT that allows quick access to any task.
2. virtual office: Here you can manage conversations with fictitious experts, which can help you familiarise yourself with non-specialist areas and make specific enquiries to real experts.
3rd learning space: The learning spaces are designed to help people understand the various possibilities of AI and learn what makes a good prompt. HAWKI is a prototype and will be continuously developed. There are plans to add further functions to the AI. In this video you will get a short introduction to LLM'S and HAWKI.
Overview - What do I need to know about ChatGPT?
In the presentation by Prof Dr Weßels (FH Kiel) What is ChatGPT and how does it work? - And what similar tools are there? You will get a comprise overview of the functions and possibilities of ChatGPT. Learn more about how this advanced AI was and is being developed and how it can help you with your tasks. If the whole video is too long, we recommend the chapter "How does GPT-3 work?" from minute 16:33.
Technology - How does ChatGPT actually work?
Dr Thomas Arnold (University of Darmstadt) goes into more detail in his presentation ChatGPT for non-computer scientists: Keys to Understanding Artificial Intelligence and its Applications in University Teaching goes deeper into the technical aspects of ChatGPT and offers a detailed insight into how generative AI works. Whether you're a tech enthusiast or just curious about the technology behind ChatGPT, this video will help expand your knowledge of ChatGPT. Check out the chapters to see which topic interests you most.
A prompt is a request, question or instruction that is addressed to an AI model such as ChatGPT to generate a desired output. The quality and relevance of the output depends heavily on the way the prompt is formulated. There are basic recommendations that have been established in recent months to get the best out of an AI. In his presentation, Prof Dr Kipp (HS Augsburg) will shed light on How do I tell my AI? Background and Principles of Prompting in ChatGPT, Prof. Dr. Kipp (Augsburg University of Applied Sciences) highlights the importance of prompting, also known as prompt engineering, to control the dialogue with the AI.
Below you will find recommendations for prompting from the AI campusprompt laboratory:
Recommendation 1: Allocate a role to the AI, this will significantly influence the type, scope and degree of complexity of the responses. Example prompt: "You are a friendly and helpful university didactics expert at a German university of applied sciences who gives me effective, specific and concrete feedback on a new teaching format that I have developed. You have high expectations and believe that I can achieve them."
Recommendation 2: Use priming to complement the role allocation and influence the content as well as the form and style of the response: "You give feedback in a direct and clear way, asking me questions that encourage me to first explain the feedback in my own words and then describe how I might respond to it."
Recommendation 3: Provide structure: "Write your answers in German and at university level. Note that my specialist field is the humanities. Be polite to me. Format your output using Markdown for headings and highlighting."
Recommendation 4: Limit the length of the output: "Keep your answers short and avoid further details or references."
Recommendation 5: Precise wording makes the answers more targeted and less generic: "First ask me for the title and a brief description of the teaching format. Don't answer for me. Your role is only that of an advisory service for university didactics. Don't manage the conversation until I answer."
Recommendation 6: Use markers to enable the AI to better understand the transmitted prompt, place them in front of the instructions: "# Your role", "# Your task as a university didactics expert", "# Instructions for your answers", "# Language, level and subject context" etc.
Recommendation 7: By sequencing tasks, you guide the AI step by step so that the AI is actively encouraged to segment and refine.
Recommendation 8: Avoid ambiguity by avoiding filler words or adjectives such as "quite", "rather" or "approximately".
Recommendation 9: Use complex prompting strategies such as Mega-Prompting or Prompt Creator. The mega-prompt developed by Rob Lennon passes through six sections and comprises a comprehensive preparation of the AI system for its task. With the Prompt Creator, the AI is used to write a prompt, which we then give back to the AI. As some of the example prompts are too long, you can read them here or directly in the Prompt Lab.
When using Chat-GPT, you should bear the following aspects in mind:
- Chat-GPT has no knowledge and cannot think logically - it is purely text-based and relies on probabilities
- The stored data is not up to date, currently a data set is stored until December 2023
- Chat-GPT may have errors in content - results must always be examined - this tool is for experts only!
- We train ChatGPT with our data - do not upload sensitive, private data under any circumstances
- Chat-GPT was trained with texts that are also based on prejudices, opinions, values and norms - these may also be reflected in our results
In recent years, AI-powered copywriting has emerged as one of the most innovative technologies in text production. With the help of artificial intelligence (AI), it is possible to produce high-quality texts without having to invest a lot of time or effort. AI-supported text creation enables people to put their ideas and thoughts into words more quickly and efficiently.
Since December 2022, one such tool, ChatGPT, has been freely available and has quickly become synonymous with all machine-generated text tools. There are many concerns about the use of these technologies in and at universities, which have also been the subject of heated debate in the university landscape and the press. In particular, the production of student writing is seen as a problem. However, as ChatGPT is not really "intelligent", but only composes texts on the basis of statistical probabilities, this challenge can be met in various ways.
1. transfer reference
- Set tasks with a concrete transfer reference, preferably to existing practical placement partners. ChatGPT has hardly any information on this and gets lost in easily recognisable phrases.
- Make reference to current topics. ChatGPT's database is not up to date. Very current topics are unknown to the system.
2. source analysis
- ChatGPT does not know any facts! When assessing the work, check the sources, especially literature references and usage in the text. ChatGPT can generate deceivingly real references, but these are almost always "made up". Citations and their sources are generated!
- If the bibliography was created by the students, the literature is correct, but the references are generated within the text. ChatGPT has no knowledge of the sources.
3. style analysis
- Machine-generated texts are amazingly error-free. Both spelling and grammar are almost perfect. The more specialised the subject matter or the more demanding the transfer performance requirements, the more clichéd the text.
- Machines have no opinion of their own. Examine whether the texts contain new insights or reflective parts.
In general, the systems can currently only reproduce content. There is no creativity involved. However, it is precisely this creativity that characterises the writing of academic texts. Therefore, avoid tasks that only require reproduction. Despite all reservations, these systems can also be used sensibly. They take over the creation of standard texts and can offer alternatives and improvements to texts. It is also possible to assess texts according to predefined criteria. Furthermore, these systems can point out and explain errors in your own work, especially in programming. ChatGPT & Co can therefore be used beneficially in teaching. It is important for students to have the competence to know the potential and limits of the software and to be able to take them into account when using it. In this way, they become part of academic research and are also subject to the requirements of "good academic practice".
ChatGPT also presents new challenges and opportunities in the area of examinations. In order to provide you with the best possible support, we have created a guide to help you find your way around.
Artificial intelligence offers a wide range of possible applications in teaching. In addition to being used by instructors themselves to organise their teaching, AI can also be used didactically by students as part of the teaching-learning process. A distinction can be made between different levels of use: Production, interaction and reflection. Below you will find some examples of the respective application scenarios:
As part of the highly recommended prompt laboratory of the KI-Campus, an open prompt catalogue with tried and tested prompts from the field of university teaching has been set up. There you will find successfully used prompts to reuse and try out.
If you have any questions or comments about the possible applications of ChatGPT, prompting strategies, etc., please contact the Central University of Applied Sciences e-learning laboratory. You are welcome to drop by during our office hours. This takes place on weekdays from 15:00 to 16:00 in the BBB room of the e-learning laboratory. Outside of the e-learning office hours, you can reach the e-learning laboratory centrally via the following e-mail address ✉ elearning@hs-fulda.de or via the ticket system of the IT Support of the Fulda University of Applied Sciences.
The following collection of materials shows just how broad the topic of "Generative AI in teaching" is. In addition to further links, which include handouts on the topic and a list of various AI resources, you will find recorded presentations that deal with specific aspects of the topic in more detail.
- The learning platform for artificial intelligence:
https://ki-campus.org/ - ChatGPT is just the beginning:
https://hochschulforumdigitalisierung.de/de/blog/ChatGPT-erst-der-anfang - ChatGPT - a milestone in AI development:
https://www.forschung-und-lehre.de/lehre/chatgpt-ein-meilenstein-der-ki-entwicklung-5271 - ChatGPT for non-computer scientists:
https://www.youtube.com/watch?v=-c8ogAwX6KI&t=6s - What are we here for? A value-based reflection and discussion on #ChatGPT in #research and #teaching:
https://www.youtube.com/watch?v=oSjdx-hadYY - ChatGPT and the future of learning: evolution instead of revolution:
https://hochschulforumdigitalisierung.de/de/blog/chatgpt-und-die-zukunft-des-lernens-evolution-statt-revolution - What does Artificial Intelligence mean:
https://www.ki-konkret.de/was-ist-ki.html - AI Resources
https://www.vkkiwa.de/ki-ressourcen/
The Hochschulforum Digitalisierung has published a dossier on the topic of "Generative AI" as well as an annotated collection of links, which can be found under the following links:
Dossier Generative AI
What role do ChatGPT and other generative AI tools play for the future of university teaching? What does the rapid development mean for skills acquisition and examinations?
ChatGPT in the university context - an annotated collection of links
Prompt Lab - Generative AI in university teaching
The Hochschulforum Digitalisierung (HFD) and the KI-Campus, in collaboration with experts, offer a comprehensive experimental space for university members with the "Prompt Lab".
Lectures Virtual Competence Centre VK:KIWA
AI Writing Tools: Implications for examination practice at universities
Dr. Anika Limburg (Head of the LehrLernZentrum of the RhineMain University of Applied Sciences)
Dr. Isabella Buck (Language Workshop at the LehrLernZentrum of the RhineMain University of Applied Sciences)
Lectures of the dghd series "AI in Higher Education".
What is ChatGPT and how does it work? - And what similar tools are there?
Prof. Dr. Doris Weßels, Kiel University of Applied Sciences
27.01.2023 takes place from 14:00 - 15:00 hrs
ChatGPT for Non-Computer Scientists*: Keys to understanding artificial intelligence and its applications in university teaching
Dr. Thomas Arnold, TU Darmstadt, Department of Computer Science
10 February 2023, 13:00 - 14:30
What are we here for? A value-based reflection and discussion on ChatGPT in research and teaching
Prof. Dr. Gabi Reinmann
22 February 2023, 12:00 - 13:00
ChatGPT as a cognitive tool - How can we promote the acquisition of 'higher order skills' in students?
Prof. Dr. Christian Spannagel
13 March 2023, 12:30 - 14:00
Teaching and testing with text-generating AI tools
PD Dr. Ulrike Hanke
20 March 2023, 12:00 - 13:30 hrs
Lectures of the HessenHub Impulse Series "AI for Education: How Artificial Intelligence is Changing the Higher Education Landscape".
Participation in the series is free of charge and limited to a maximum of 300 participants per event. The series is organised by the HessenHub sub-projects at the Frankfurt University of Applied Sciences, the Technische Hochschule Mittelhessen and the Justus Liebig University Giessen and is primarily aimed at teaching staff at Hessen universities.
Registration is via a web form at the Frankfurt University of Applied Sciences and is possible up to two days before the respective event begins. All events take place as online formats via ZOOM. Individual events will be recorded.
You can also find all the information on the flyer for the impulse series.
Artificial Intelligence: Current Developments and Implications for the Higher Education Sector
Date
29.03.2023 - 15:00 - 16:30 hrs
Speakers
Prof. Gabriela Alves Werb (Frankfurt UAS)
Ph. D. & Dr. Sarah Malewski (Frankfurt UAS)
Description
This talk will provide an insight into the current state of AI applications and their uses. We explain how these applications learn and address their potentials and limitations. We also highlight their practical implications for research and teaching and discuss them from different perspectives with the participants.
Future Skills in the Age of AI
Date
20.04.2023 - 16:00 - 17:30 hrs
Speakers
Prof. Dr. Michael Guckert (THM)
Description
New developments in the field of artificial intelligence, such as the prominent example of ChatGPT, enable completely new applications in business, administration and society. An active creative approach to artificial intelligence is thus increasingly becoming an essential competence. In particular, new opportunities and challenges arise for teaching and studying. The lecture will discuss answers to the following questions:
- What skills do students and teaching staff need in dealing with ChatGPT and Co.
- What skills will students and teaching staff need in the future?
- How can AI help to teach students the skills/key competences that are important for the future in a digitalised world?
- Can AI support students in their individual learning process?
ChatGPT et al. - Implications of Natural Language Processing for a Sustainable Examination Culture
Date
09.05.2023 - 16:00 - 17:30 hrs
Speakers
Dr. Anika Limburg (HRSM)
Description
The implications of tools like ChatGPT for higher education are currently much discussed. While some universities have already banned the use of such tools in the examination context, elsewhere they are recommended in the sense of human-machine interaction. At the same time, the search for innovative examination formats and modalities has begun, which should take technological developments into account and ensure the future viability of the higher education examination system. This lecture will highlight both the problems and fears, for example of a loss of examination fairness and the infiltration of learning processes, as well as the first promising approaches with which AI tools can enrich higher education teaching and examination systems.
The lecture will discuss the following questions, among others:
- What impact does AI have on examination culture: examination formats, performance, assignments, assessments?
- What advantages does it offer teachers in the context of examinations (creation of MC questions, assessment matrix)?
- Where might dangers arise in relation to the examination culture at German universities?
- What does ChatGPT mean for copyright and plagiarism?
Artificial Intelligence in University Teaching: Didactic Implications and Constructive Use in Teaching Practice
Date
12.06.2023 - 16:00 - 17:30 hrs
Speakers
Prof. Dr. Christian Spannagel (PH Heidelberg)
Description
At the end of 2022, the release of the AI system ChatGPT shook up teaching staff at schools and universities: The language model is able to provide answers to questions in natural language, for example when working on homework or solving exams. Yet this powerful tool is only the beginning of a development that is 'rolling towards' the education system. Calls for a ban on such systems quickly became loud.
The lecture will deal with the constructive use of AI systems in teaching beyond bans. The following questions will be answered:
- How do learning goals change against the background of the availability of AI systems?
- How can students and teaching staff use AI productively?
- And how can problems and challenges be met?
Digitality for all!
Date
26.06.2023 - 16:00 - 17:30
Speakers
Prof. Dr. Kati Hannken-Illjes (Philipps University Marburg)
Description
Digital competences and the ability to reflect on the conditions of digitality are fundamental for all fields of study and an important prerequisite for most fields of activity. Using the example of the MarSkills study area at the Phillips University of Marburg, the presentation shows how these offers can be included in the curriculum in such a way that all bachelors students can use them. A special focus is on the Marburg module as an interdisciplinary project module.
ChatGPT and Consorten - Blessing or Curse for University Teaching. Answers from the perspective of examination law and didactics
Date
12.07.2023 - 16:00 - 17:30 hrs
Speakers
Prof. Dr. Ingo Striepling (OTH Regensburg)
Description:
AI like ChatGPT is on everyone's lips at the moment. Colleagues are streaming live tests based on old exams and certifying that the AI would have passed. The same applies to aptitude tests. Is this the brave new world? Actually, the idea of having others work for you is not new. What is explosive now, however, is the rapid speed with which passable results can be generated. What impact will this have on our courses and examinations? Can we stick to the current formats? How can we prevent the use of ChatGPT, for example? Or should we better ask: How can we guide our students to use AI intelligently? We want to approach the answer to these questions in the context of inputs and discussion.