Be aware that you are responsible for knowing the rules regarding the use of generative AI during exams. It is important that you understand the rules for each individual exam to avoid engaging in misconduct, such as using unauthorized aids or seeking improper assistance during the exam.

If you use generative AI in an exam where it is not allowed, it is a violation of AAU's disciplinary rules. You can find AAU's disciplinary rules here.

In exams where generative AI is an allowed aid, you should also be mindful of plagiarism rules. If you utilize information generated by generative AI, make sure to provide proper attribution and references. This applies as well if you use notes generated by generative AI during the exam.

If unauthorized use of generative AI as an aid or plagiarism is suspected, AAU will initiate disciplinary proceedings against you.

If you are unsure about the rules regarding the use of generative AI for a specific exam, seek advice and guidance from your study board.

Unless stated otherwise by your study board, it is allowed to some extent to use generative AI services in project work, such as for idea generation or information retrieval. In this context, it is important to ensure proper source referencing (see more under 'Good (scientific) practices regarding generative AI').

It is not permitted, for example, to generate text that is directly inserted into the project report, according to AAU's disciplinary rules. 

You are not allowed to use generative AI for the oral examination in project exams unless it is explicitly stated in the course description that generative AI is an allowed aid.

Your study board may have established additional guidelines for the use of generative AI in project work and exams. In that case, you should follow the guidelines provided by your study board.

Generative AI can be encountered on various platforms, and it is not always explicitly labeled as such. You can find this technology in:

1. Services that clearly identify themselves as generative AI, such as ChatGPT, MidJourney, DALL.E, and Google Bard.
2. Elements within applications/services/software, such as the Bing platform, where generative AI-based components are presented alongside search results, or services that automatically generate summaries of scientific material. This also applies to browser plugins and similar tools that utilize these services.
3. Embedded within applications/services/software as a feature for generating text, images, or code, where the generative aspect is embedded as "type-ahead" or suggestions, such as GitHub CoPilot.

Please pay special attention to numbers 1 and 2 regarding the services you use. If the use of aids is allowed during exams, such as for literature searches, it does not automatically exclude the use of services that potentially embed generative AI elements, such as a search engine. However, the use of these generated elements from the services are regulated by plagiarism rules.

Be particularly mindful of number 3 regarding the software you use to compose your submission. It may be necessary to disable certain features in the software you use to avoid unintentionally violating exam rules or producing material that is considered plagiarism.

Note that, as a student, you may be subject to special examination conditions resulting from individual assessment and dispensation, which may allow the use of selected aids even if they employ generative AI.

Spelling and formulation skills are evaluated in written exams at AAU. This means that generative AI or similar technologies may not be used for text rewriting unless explicitly stated in the module description as an allowed aid. This is because the use of this type of generative AI prevents examiners from making an accurate assessment of the student's spelling and formulation skills.

Unless explicitly stated as an allowed aid in the module description, methods built on generative AI may not be used to rewrite, interpret, or create software code during exams.

Be particularly aware that code released under open-source software licenses may be included in the output from tools like GitHub CoPilot. When a significant amount of code is reused from generative AI, it is good practice to search for elements of the code to determine if the original code repository (and its associated license) can be included in your work.

Regarding generative AI that utilizes large language models, such as ChatGPT, you should be mindful of the validity of the responses provided by the service. Therefore, you should be aware of the following:

Unknown sources
You do not know the source or extent of the material the service uses. Typically, it is based on text from news media, publicly available research, social media, discussion forums, etc. Similarly, there is often a cutoff date for when the model was last trained.

AI services hallucinate
AI models can hallucinate, meaning they may invent sources or facts that are not accurate. This is due to the way these models function, and in simple terms, the algorithm will attempt to guess an answer. Language models will go to great lengths to provide you with a seemingly convincing answer. Always use other tools to verify statements and sources, including referencing the original source rather than the generative AI. However, describe in, for example, a methodology section that you have used generative AI as part of your methodological foundation.

Ethics
The service you use will have a code of conduct to assess what is right and wrong. This shapes the responses, and therefore, perspectives existing in the source material may be suppressed or omitted in the answer. This is largely an ethical consideration regarding, for example, the inclusion of viewpoints.

Methodology
If you use generative AI, for example, to gain an overview of a scientific field, and this influences your further analysis or assumptions about a topic, it is considered proper, good scientific practice to declare this as part of the methodology for your scientific work.

In all examinations, the rules for plagiarism defined in the disciplinary regulations at AAU apply. It states that plagiarism: "(...) contains the use of the wordings or ideas of other authors’ without such other authors being duly credited." (AAU's disciplinary regulations). 

In examinations or other written work where, generative AI is allowed as an aid, you still need to be aware of the rules for plagiarism and that it is not allowed to obtain improper assistance in your answers. Even though the result from generative AI is often unique and novel (as it has not been formulated before), you must adhere to the same rules for the use of material produced by generative AI as with other sources, including citations and references, and follow the rules for paraphrasing. This also applies if you use notes generated by generative AI during the exam.

Similar disciplinary measures apply if generative AI is used to falsify or fabricate data, as this can be considered a violation of good scientific practice.

Note that even if you provide clear referencing when using generative AI, the statements you reproduce may be factually incorrect due to the way generative AI models function.

If you are unsure whether you can reproduce the result of generative AI (e.g., text generated with it), it is good practice to save as much information as possible about the timestamp, prompt, output, and information about the version of the service/model. For example, use screenshots to ensure this.

When using generative AI to generate images, video, and audio, you need to pay special attention to the licensing conditions regarding usage and proper inclusion of information about the same.

For example, material created on a service and used under a Creative Commons license should have the license mentioned when the material is used. Similarly, the terms of service of the service apply, where these conditions may be further regulated.

You should also remember to properly attribute images.

Images, video, and audio based on generative AI are built on large amounts of data collected from others. Therefore, there may be a possibility that the data it relies on is protected by copyright. Hence, it cannot be assured that the material produced by AI can be used commercially without permission from the copyright holder.

Similarly, you need to be aware of whether the terms of a given service will assert the right to use the material you use as input. This can be a violation of copyright if you use material that is protected by copyright or made available under a specific usage right.

As a student at AAU, you are not obligated to use generative AI as part of your education. Your educator may recommend that you test generative AI, but educators can only require you to use generative AI services that are GDPR-compliant, which is approved under the General Data Protection Regulation. For example, at the current time, ChatGPT is not GDPR-compliant.

If you use generative AI and other online services, the service provider will process your personal information. This can include the information you provide during the process of creating a user, but it can also include more hidden information, such as your location, the website you came from, and the device (smartphone, tablet, etc.) and browser you use when accessing the service. The provider processes this information to deliver the service to you but also to develop and optimize their service.

Many providers of free services also use your information for their own and their partners' commercial purposes. This can involve creating a profile of you that can be sold and/or used to tailor marketing specifically to you.

When using generative AI, you should also be aware that the provider of generative AI often transfers your personal information to countries outside the EU, such as the USA, which may have weaker personal data protection than what is required under the GDPR in the EU.

Some generative AI services incorporate the information you share with the AI service to build knowledge within the service. In other words, you contribute to the creation of AI, and what you have shared may be passed on to the next user of the service. Therefore, you should only share information that is either already publicly available or that you are willing to make public.

You should never share other people's personal information (such as name, gender, age, political beliefs, email, etc.) in generative AI services, as you would be responsible for the processing of this personal information, and you would need to comply with GDPR. You should also avoid sharing confidential information in generative AI services. Information can be considered confidential, for example, if you received it from a company with which you have a collaboration.

”As a student I have utilized generative AI to assist me I various ways a lot. Sometimes I use it as a research assistance to provide summaries of academic articles or as a study aid to better understand complex concepts providing me with explanations, almost like a personal virtual tutor. I have used it for brainstorming, coding, writing support for feedback, on my drafts, generating pictures for presentations and possibly many other things.

From my experience, generative AI can be a very helpful resource, but I have to approach it critically. It takes some time to master and understand to use it in a meaningful manner.”

Student, Techno-Anthropology, 8th semester.  

"I have used ChatGPT extensively in my studies. After starting my studies in digitization and application development at the department of Computer Science, I also had to start coding. ChatGPT has truly helped me in coding courses by explaining the code in understandable language, and it remembers the conversation and builds on previous "commands". For example, I have used it when a piece of code is difficult to understand, and ChatGPT can break down the code and explain what each line does. So, for humanities students pursuing this technical master's degree, it is indeed a very useful assistant. It is a place where I gain easy understanding and draw inspiration from - not a place where I seek more in-depth knowledge. Additionally, I have used it for literature searches - it can provide peer-reviewed texts on a specific topic. Lastly, I have also used it to summarize what larger books are about.

ChatGPT has great potential because it assists and streamlines some of my study practices. However, there will always be a pitfall we need to be aware of regarding exam cheating, etc., so as students, we should also know exactly what we can and cannot use it for within the university context."

Student, Digitalization and application development, 8th semester