For marketing and communications specialists as well as other employees involved in the content creation process.
AI usage in the communication materials of Radboud University
Today, it seems like Generative AI (GenAI) is everywhere. It is a powerful tool for working more efficiently, but it also bears various risks to be aware of. Do you use GenAI in your work? This regulation explains how to use GenAI more responsibly and beneficially.
Possibilities of AI within communication
GenAI can be an empowering tool in your work as a marketing and communications professional when it comes to:
- Assisting in generating ideas and discovering new perspectives
- Maximizing efficiency of work processes by, for example:
- Improving the grammar and style of texts
- Translating texts
- Editing images and adjusting an image format (e.g., from vertical to horizontal with background adjustments)
- Training AI to respond as different personas to your marketing output
- Creating better call to actions.
However, such benefits may come at a price. Incautious use of GenAI can also have negative consequences for individuals as well as for entire organizations.
Thus, it is crucial for everyone who uses GenAI, especially in their work duties, to be informed about the associated risks.
Risks and challenges of using AI in communication
The increasing use of GenAI in communication can bring significant risks to the authenticity and credibility of the Radboud University brand. Machine-generated content often lacks the unique tone of voice and emotional depth that we as human communications professionals can offer.
Moreover, machine-generated content can miss context-sensitive nuances, resulting in inappropriate messages that undermine Radboud University's values and image, negatively affecting our customer loyalty and perceived level of trust.
These challenges extend to ethical and creative areas, where a lack of originality, risks of plagiarism and privacy breaches can occur. For example, prompts in free AI tools (such as ChatGPT and Google Gemini) are saved and used to train AI models. The data you enter becomes accessible to the service provider (such as OpenAI or Google). Therefore, sensitive data, such as critical business information, texts under embargo and those containing private details shouldn’t be shared with free AI tools due to privacy and security risks.
In the digital age, consumers increasingly value human contact and brands they can relate to due to their human approach and communication style. If Radboud University's textual and visual communication is perceived as "too automated", we risk losing both potential and loyal members of the Radboud community (students/staff/partners, etc.) to competitors who offer a more human approach. Besides, being perceived as "automated" would take away from all the daily efforts our staff and students put in to make a significant impact.
This document therefore sets guidelines to ensure our communication will remain to be perceived as human.
Recommendations for responsible content creation
As a Radboud employee, you are responsible for the content you produce. Content created with AI assistance should always be checked and corrected for the right tone of voice (appropriate to the organization), accuracy (Is everything correct?) and completeness. If you are in doubt, don't hesitate to ask your supervisor or colleagues for their opinion.
Below you will find an overview of different ways to use AI, the associated risks and ways to avoid them.
1. Texts
Low risk
- Brainstorming ideas
- Improving the style and grammar of self-written texts without sensitive information via the paid versions of tools (like DeepL Pro)
- Translating self-written texts without sensitive information via the paid versions of tools (like DeepL Pro)
Moderate risk
- Generating a summary of a self-written text without sensitive information
High risk
- Completely machine-generated texts without revision
- Improving self-written texts which (might) contain sensitive information
Important! Due to unpredictable outcomes, usage is not recommended
2. Images
No risk (No AI tools involved)
- Images from BrandPortal
Moderate risk
- AI-modified images (e.g., adjusting size, improving quality, removing minor objects, editing imperfections)
High risk
- Completely machine-generated images
Important! Due to unpredictable outcomes, usage is not recommended.