You must always adhere to good academic practice when using AI tools to support your learning and develop your work. This includes completing your studies with integrity and, where appropriate, indicating when you have relied on the benefits of AI. Within a course, lecturers will specify whether and for what purpose AI tools may be used.

An example exercise that encourages reflection on the output of generative AI could look like this:

► In an assignment, students are asked to enter prompts on a specific topic into a generative AI tool. The output is to be critically examined for criteria such as structure, false information or inconsistency. On the one hand, this promotes an in-depth examination of a topic, while on the other, it highlights the possibilities and limitations of generative AI.

Prompts are inputs in the form of one or more sentences in a Large Language Model (LLMs) such as ChatGPT to record a dialog. This is called prompting or prompt engineering. The more precise the input into a generative AI tool is, the better the results will be. It is advisable to make several attempts to generate the desired result.

The following tips can help you with prompting:

• Give the AI clear instructions and sufficient context: In the course of this, take on a role yourself.
  ► Example: "Create a business plan to found a startup business. This is a company to develop an app for students in German-speaking countries. I will be the managing director of the company and need the business plan to ask for financing. The business plan should contain the following points: Executive Summary, product description and an analysis of the market and competitors."

• Ask the AI to take on a specific role: This way, you allow the tool to vary in style and influence the output to adapt it to your ideas.
  ► For example, start your prompt with the sentence: "Write me a speech on the topic: What is the relationship between the unemployment rate and inflation? The speech should be written in the style of an expert in the field of economics with many years of professional experience."

• Define key points that the AI should definitely mention: You can also specify which information you want to exclude.
  ► An example of this could be: "Create an essay on the topic of combating unemployment. Include the areas of long-term and youth unemployment. The aspect of a possible reduction in working hours and its influence on unemployment should not be addressed."

Other helpful tips:

• Ask the AI tool to generate a summary of previous chats to track the progress of your conversation. How did you arrive at this point?

• Ask the AI how it could improve its own output.

• Give the AI counterarguments in order to receive an improved text suggestion.

• Provide an outline of the desired text to help the AI tool organize itself.

If you are considering artificial intelligence for your studies, you should think carefully about how you want to use it to support and improve your learning. We also advise you to take a close look at the AI tools and their possibilities, as not every tool offers services that are suitable for your purposes. For further information visit the WU Library website.

In the following, we would like to introduce you to some potential areas of application of AI in the context of your WU studies.

The use of AI tools in your academic work, especially in theses, changes the requirements for authors. Good scientific practice includes clearly stating which AI technologies you used and for what purpose.

Transparency and documentation

It is recommended that you clarify with your supervisor whether and for what purposes AI tools may be used before starting the writing process. AI-generated content must be transparently identified. In addition, we recommend that you keep the prompts you have created, including the generated answers, and document them accordingly.
Comprehensive information on citing AI-generated texts can be found in the WU Library.

Please note that the development of tools for recognizing AI-generated texts is progressing rapidly. WU already uses AI verification software in suspicious cases.

• Information may be wrong: AI tools create their answers by combining data sets that are largely taken from the internet. But not everything that circulates on the internet is factually correct, even if it reads that way. It may appear that AI tools can "think", but they are literally putting together language that presents completely or partially fabricated facts.  The tendency of AI tools to invent information is sometimes referred to as "hallucination".

• Quotes may be made up: AI responses may contain quotes (especially when prompted), but use caution here. Even if the quotes appear reasonable and are formatted correctly, they may not actually exist or may be incorrect. For example, the tools may invent an author or give a book title that does not exist. Teachers are well versed in their areas of expertise and can easily spot these errors. Generative AI tools are not reliable sources.

• Answers may contain bias: Bias is a distorted perception or judgment. AI tools cannot distinguish which data sets reproduce biases or distortions of facts. Please be aware of where bias may be present.

• Risk of violating the guidelines of lecturers: When using AI in an academic paper, it is possible that the guidelines for permitted aids (to be found in the syllabus) may be violated. Please work carefully here to ensure that you adhere to your lecturer's guidelines and the Code of Conduct in general.

• Ideas could be changed unintentionally: When reviewing AI-generated answers, work critically and thoroughly. This ensures that the output matches your own understanding and expectations.

► Consequences according to Study Regulations: According to WU's Study Regulations, the unauthorized use of AI in courses is not permitted and will have consequences.

The data you put in an AI software is generally used and stored to improve it. It is therefore essential to avoid entering personal data or other sensitive data (e.g. health or financial data). This applies to your own data, but above all to data from other people. Before using an AI tool, it is advisable to carefully read the respective data protection regulations.