Zoek in de site...

Guidelines GenAI

The developments in the field of generative artificial intelligence (GenAI) are progressing rapidly, and naturally, there are many questions about its impact on education. GenAI has the potential to offer much to education but also has intrinsic shortcomings, can hinder the learning of skills, and raises questions regarding ethics, privacy, safety, and academic integrity. To provide guidance to students and teachers, the FFTR faculty, in alignment with other faculties at RU, has established the following guidelines regarding the use of GenAI in education.

What Do We Mean by GenAI?

By GenAI, we mean applications that can generate text, images, music, and other content (output) based on a written or spoken instruction (prompt). GenAI applications use models that have learned to recognize patterns in large amounts of training data. With that information, content can then be generated based on statistical probability.

Points of Attention

  • GenAI applications do not understand the world; they merely perform probability calculations. While it is possible for them to create accurate content, they frequently produce content that does not align with reality, a phenomenon known as "hallucination."
  • For very specific, specialized, recent, or local topics, GenAI applications are likely unable to generate accurate output as these topics are expected to be underrepresented in the training data.
  • Often, it is unknown which data has been used to train the models that GenAI applications are based on. This is problematic because the output of GenAI applications is only as reliable, representative, and fair as the training set. The information in the training set may be incorrect or outdated and could contain biases against population groups, gender, and religious beliefs.
  • Chatbots present privacy and security risks, as well as broader ethical, legal, and strategic concerns. It is not always clear what developers of GenAI applications do with the texts entered as prompts. For most (free) GenAI models, all input becomes the property of the developer. The training sets may also contain copyrighted material without the author's permission.
  • The training and use of GenAI systems can lead to violations of privacy and data protection rights, as laid out in the General Data Protection Regulation (GDPR). Training data of GenAI models may contain personal information about someone or share personal information about someone as output. The GDPR stipulates the conditions under which personal data may be processed.
  • The recommendation is not to share personal data, research data, sensitive information, or intellectual property. This is only advisable – as with other new tools and applications – once a license has been purchased and a processing agreement has been concluded between RU and the GenAI provider to ensure the protection of personal data and security measures.
  • Developing and training GenAI models consumes large amounts of energy, having a significant impact on the environment and climate. Additionally, developing this technology requires a substantial financial investment, which can almost exclusively be borne by large commercial tech companies.

Guidelines for Students

Basic Principles:

  1. You are fully responsible for the work you submit.
  2. Ensure that your work unambiguously allows for the evaluation of the competencies you have acquired.

Misuse of GenAI

Clear misuse of GenAI – where (a significant portion of) your work is largely generated by GenAI and you were not transparent about using the tool – can be considered fraud according to Article 3 of the Fraud Regulations, with all the associated consequences.

How Can You Use GenAI?

A GenAI model can be used in various ways. Certain uses are similar to what we already do with other tools and do not require additional transparency. Other uses may obscure your own contributions and are therefore not allowed, except in specific exceptions.

GenAI: What Is Permitted?

  • You may use GenAI as a language assistant to improve self-written texts, provided the model does not add new content. In such cases, its use is equivalent to spelling and grammar checkers, and you do not need to explicitly mention this use. However, always ensure that the use of GenAI does not hinder the evaluation of the learning objectives you have achieved.
  • You may use GenAI as a search engine to orient yourself on a topic or to identify relevant literature. This form of information gathering is similar to using a regular search engine. Then, seek out the source documents yourself and conduct your own analysis of them. Interpret, analyze, and process the obtained information yourself; do not simply copy the information. If you subsequently write a text yourself, you do not need to mention the use of GenAI.

Note: When you do take specific parts of GenAI output verbatim (e.g., due to the nature of the assignment), you should cite and include a source reference.

Here is how you can do it:

Reference List

Author of the GenAI model used. (Year of the GenAI model used). Name of the GenAI model (Version of the GenAI model used) [Type or description of the GenAI model]. Web address of the GenAI model.

Example: OpenAI. (2023). ChatGPT (April 20 version) [Large language model]. https://chat.openai.com/

Note: The output generated by ChatGPT and similar tools is not always correct and can vary with each attempt. For factual information, ChatGPT is not a reliable tool; verifying factual information through other sources is always recommended.

GenAI: What Is Not Allowed?

  • Any form of verbatim copying without complete source attribution (citing, referencing) of material generated by GenAI. Academic work requires that original sources can be traced. Use of GenAI with source attribution is only allowed in exceptional situations.
  • Any use of GenAI that prevents the teacher from tracing and/or evaluating your own capabilities.
  • Any use of GenAI during on-campus exams or other assessments where the use of GenAI has been explicitly prohibited.

How to Be Transparent About GenAI?

Being transparent about your use of GenAI is essential. Only then can an accurate assessment of your knowledge, understanding, and/or skills be made. How you achieve transparency, however, is difficult to outline in general guidelines. It usually depends on the purpose of the (written) product you are creating and can be derived from the assignment description.

A good practice for maintaining transparency is to keep a record of how and why you used GenAI. This can take various forms, for example:

  • Save the entire exchange with GenAI by storing your chat log. Highlight relevant sections if necessary.
  • Provide an explanation of how GenAI was used (e.g., for generating ideas, text fragments, longer pieces of text, arguments, illustrations of concepts, etc.).
  • Note why GenAI was used: to save time, to overcome writer’s block, to stimulate thinking, to manage increasing stress, to better understand a concept, to translate, to experiment with GenAI, etc.

How you report this can vary. Check if this is clarified in the assignment description of the (written) product you are creating. Besides source attribution and inclusion in a reference list (for verbatim citation), it may be desirable to explain the above elements in a methods section, footnote, or appendix.

If it is unclear how best to report the use of GenAI, consult with your teacher/supervisor.