Responsible engagement with GenAI in 9 rules

GenAI literacy means that you are aware of the risks of GenAI and that you also have the knowledge, skills and attitude to engage with GenAI applications effectively, critically and ethically in different contexts. This also means that you make conscious choices about when to use GenAI and when not to.  

• Follow the e-learning course to learn more about how GenAI works, what it is good at, and what its limitations and risks are. 
  • For teachers: follow the e-learning in Brightspace.
  • For researchers: follow the e-learning in gROW “AI literacy for researchers”.
  • For employees: follow the e-learning course in gROW.
  • For students: follow the e-learning in Brightspace.
• Keep your knowledge up to date by continuing to learn and stay informed about new developments 

• Use GenAI in an ethical and legally responsible manner: ask yourself whether the use of a GenAI application is really necessary and whether it is legally permitted. 
• Never use GenAI for prohibited or harmful tasks (e.g. to create malware or deceive others).  
• Consider whether your question can be answered just as well or better by a regular search engine.  
• If you use a search engine, do not rely on an AI-generated answer that is displayed first.  

• Only use authorised GenAI applications to improve, support and/or enrich your work. Currently, Microsoft 365 Copilot Chat is the only GenAI application permitted within Radboud University for texts, provided you are logged in with your Radboud account and not with your private account or account from another organisation.  
• Work safely:  
  • According to the Radboud University Information Provision Rules of Conduct.
  • Use a strong password and multi-factor authentication (MFA).
  • Store your files securely.

The use of personal data is subject to legal conditions that GenAI cannot yet guarantee, which makes its use unlawful. 

• Do not enter personal data such as names, addresses, dates of birth, email addresses or student tests.  
• Pay extra attention to special personal data, such as health data. 

The companies behind GenAI applications use input data to train their models. If you enter confidential information into a GenAI application, this information effectively becomes public, whether or not it is broken down. This also makes it easier to unlawfully reproduce letters or invoices, for example. 

• Do not enter any financial data or confidential business information. 
• Do not enter licensed scientific articles.
• Disable training or memory functions, if available.  
• Unfortunately, it has become apparent that several AI companies do not comply with (copyright) legislation, so never share confidential information. 

GenAI always predicts the next word in a sentence. As a result, an answer may seem convincing, but still contain errors. In addition, the output may contain biases or violate copyrights (plagiarism). GenAI also does not understand emotions, intentions or moral nuances. 

• Be critical of the answers provided by a GenAI application: always treat the results as concepts, not as definitive answers. 
• Check every important piece of GenAI-generated content for facts, grammar, and possible plagiarism using reliable sources. 
• Use your own expertise and intuition: if something sounds implausible, look it up. 
• Check the references in GenAI's output: they may not exist, may be unreliable, or may not contain the evidence suggested by the output.  
• Be aware of your own cognitive biases when assessing the output. These include 'conformation bias' (the tendency to pay attention to and value information that confirms one's own ideas and beliefs) and 'automation bias' (the tendency for people to rely on automated systems). 

From a legal perspective, a GenAI application should not be allowed to make independent decisions, because it cannot interpret context and does not have human insight or cognitive functions. People must therefore remain in charge. For example, you should not ask who the best candidate is based on a set of application letters.  

• Use GenAI as a tool: it should not replace human judgement.  
• GenAI contains biases. Using GenAI, even when used by and with a human, can lead to discrimination. Make sure you can always justify your work, so always assess and check the results of a GenAI application.  

Under legal frameworks we have to inform where we use (Gen)AI. People have the right to know when (Gen)AI systems are used and how they influence decisions.  

• Clearly mark GenAI-generated content (e.g., "This summary was generated by GenAI and checked and edited by me.") 
• Cite sources and give credit to original authors. 
• Explain which GenAI application and steps you used. 

Generating or manipulating images and sound with GenAI is undesirable. The copyright on image, video and audio material   is a legal grey area. 

• Be aware that the material used to train GenAI is almost certainly protected by copyright, for which the author has often not given permission. You may therefore unknowingly infringe on someone else's copyright.