Guidelines for the use of generative AI
AI (Artificial Intelligence) refers to a machine's ability to exhibit skills - such as reasoning, learning, planning, and creativity - that are inherent to humans. Within this domain, generative AI shines as the master of creation. It empowers machines to generate new content based on existing data, ranging from images to text and sound.
On this page, you will find the current guidelines for the use of generative AI at Utrecht University (UU). These guidelines enable us to harness the opportunities presented by generative AI while simultaneously mitigating the risks associated with its use.
What can generative AI be used for?
Generative AI can be applied in many different areas, making it a potentially valuable tool for various kinds of use.
Risks
To safely harness the possibilities of generative AI, it is essential to recognize the risks associated with its use and proactively take measures to reduce them. Below, you will find a list of the main risks associated with the use of Generative AI.
Privacy
The main privacy concerns related to AI involve the risk of data breaches and unauthorized access to personal information. Given the volume of data collected and processed, there is a risk that it could end up in the wrong hands, for instance, through hacking.
Copyright
Generative AI can generate content that infringes on copyright by creating material based on existing copyrighted works without the permission of the rights holders. This can lead to legal issues and claims of infringement. If AI generates content closely resembling existing works, it may result in accusations of plagiarism, where the original copyright holders may contend that their work has been unlawfully copied.
Lack of responsibility and accountability
Because generative AI can autonomously generate content, it is sometimes challenging to determine who is responsible for the generated content. Various parties, including the developer of the AI tool, the user, or even the AI itself, may be involved.
Incorrect or discriminatory information
Generative AI relies on training data and may contain biases that can manifest in the generated output. This can lead to discrimination or inequality, for example, in the case of job applications or decision-making. Additionally, the technology can generate output of lower quality or incorrect results.
Dependency on technology
Dependency on generative AI can result in employees losing certain skills and expertise. Additionally, the use of AI may lead to job loss in areas that can be automated. It is important to find a balance between the use of generative AI and the preservation of human knowledge and insights.
Safety
Generative AI can be exploited by malicious actors, for instance, to forge documents or manipulate image and audio content (deepfakes). This can pose security risks, such as identity fraud.
Guidelines for the use of generative AI
When using generative AI applications as an employee of Utrecht University, you are required to adhere to the following guidelines.
a. Privacy en data protection
Do not share sensitive personal and/or company data when using generative AI applications.
Utrecht University handles personal data carefully and operates within the boundaries of the law, specifically the General Data Protection Regulation (GDPR).
Personal data is not rented, sold, or otherwise shared with or disclosed to third parties.
b. Intellectual property
Every AI-generated expression must be thoroughly checked to ensure that the content complies with applicable copyright laws and that any required permissions or licenses have been obtained correctly.
Online plagiarism detection tools are available to help identify potential duplicates of existing content. Popular plagiarism detection tools include Turnitin, Copyscape, and Grammarly.
When using specific information, quotes, or ideas from a copyrighted work, the source and author must be acknowledged.
c. Accountability and Transparency
All AI-generated output must be thoroughly checked to prevent the dissemination of incorrect information and/or discriminatory content.
Under no circumstances should generative AI be used for illegal, harmful, and/or discriminatory activities. Utrecht University has established specific behavioral guidelines. The Code of Conduct (pdf) provides the framework for behaviour and reflection for employees and students of Utrecht University. The Code describes the values that govern the way people work and study at the University.
When using generative AI in your duties as an employee of Utrecht University, you are deemed personally responsible for the generated content. This means that you are liable in case of damage or unwanted consequences resulting from the use of AI.
It is crucial for employees to acknowledge the use of AI in their duties. If any form of AI is used, the tools employed must be disclosed.