AI Guidelines for Communicators
INTRODUCTION
Members of GVSU’s University Communications, Information Technology and the University Communicators Network have compiled a series of guidelines and tools for faculty and staff to implement generative artificial intelligence purposefully, efficiently and ethically.
These guidelines are intended to provide a framework of best practices for the Grand Valley community as it considers the use of generative AI across social media and web-based platforms as well as communication channels/networks. Generative AI is capable of creating text, images, videos or other data using generative models, often in response to prompts by the user.
The GVSU academic policy, meant for use by students, faculty and staff, can be found on the GVSU IT website, along with other AI resources and initiatives on campus. Before considering the use of AI, read the university’s Acceptable Use Policy for Public AI Solutions.
AI is a powerful tool that can be used for a variety of methods and must be approached with thoughtful consideration to its purposes and applications.
WHAT IS GENERATIVE AI?
Generative AI is a form of artificial intelligence that learns from large datasets, such as publicly available internet content, books, periodicals, research papers and other data to design and create original content, including text, images, music or graphic design.
Generative AI primarily seeks user input and prompts before initiating its tasks. Some examples include ChatGPT, which can assist with many text-based functions (see below), DALL-E and its updates that can create image content (illustrations, sketches and photorealistic images), and MusicLM, which can create music.
GUIDELINES
- Artificial intelligence is a tool that can enhance productivity and creativity. It should not be used to replace human-centered work. Respect individual data and privacy.
- AI can be a useful tool for concept development or editing human-created content, but avoid using it as a tool to replace user-created content.
- Because AI inputs are driven by humans, errors and misinformation do occur. Make sure to cross-check AI-derived content with reputable sources. AI-generated material should be carefully reviewed, edited and overseen by a human author.
- Establish clear and defined goals for when artificial intelligence can and will be implemented. Transparency is key to establishing trust with your audiences. Consider whether the use of AI in one’s work warrants disclosure of the use of AI tools in their workflow. When in doubt, talk to a supervisor and refer to the university’s policy.
- Given the rapid pace of AI innovation, stay informed of advancements in the technology to better understand and utilize its deployment.
As AI technology advances, new platforms and tools emerge. There are a variety of publicly available tools, which are listed below, and feature AI capabilities available to GVSU communicators. (These are only a sampling of the rapidly evolving AI tools available).
EXAMPLES
Firefly integrates with other Adobe products to help users develop audio, images and video. Firefly can help create B-roll, background images or visualize an idea for a presentation. Other controls allow users to adjust camera settings, such as camera angle or zoom. Firefly can also translate audio and video to help you deliver content in your audience’s native language. Purchasing upgrades to the app allows unlimited access to images and ability to create more videos.
Apple Intelligence can assist users with tasks like composing, proofreading and editing text across email, documents and presentations. In Notes, Voice Memos and during phone calls, users can get a transcript of a recording in real time and share that transcript via email or Notes. In Photos, Apple Intelligence can identify and remove selected images from the background.
In the similar way that ChatGPT can create original text based on a user’s instructions, DALL-E 3 can produce images based on a user’s text descriptions and requests. DALL-E 3 is integrated within ChatGPT and provides users an enhanced experience that responds with more depth in crafting images. There is a limit to the amount of images created per day though (usually 3-5 images per 24-hour period). An upgrade to ChatGPT Plus is available, if you’re looking to create an unlimited number of images.
There are other limitations as well. Due to copyright and trademark restrictions, DALL-E 3 cannot incorporate Grand Valley branding (ex. slogans, taglines, Louie the Laker, logos) into its designs. Instead, the AI will give the option to create stylized or artistic versions of mascots and logos (see examples below).
This AI image was generated with DALL-E 3 and presents a stylized version of the GVSU mascot, Louie the Laker. Trademarks and copyrights limit an AI's capabilities when creating images.
This stylized image of GVSU's logo was created with DALL-E 3, an example of generative AI. Trademarks and copyrights limit an AI's capabilities when creating images.
Since its release in January 2025, DeepSeek has disrupted the AI landscape. Developed by a Chinese firm, DeepSeek performs many of the same functions as ChatGPT, and in some cases, exceeds ChatGPT (such as providing in-depth, technical answers or producing simplified responses from complex data input). However, some AI experts have raised concerns over DeepSeek’s privacy and security with its data.
Utilized across Google’s products, Gemini has the functionality of ChatGPT. Unlike ChatGPT, Gemini can accept multiple input sets (text, graphs, images) from a user at the same time to create its content.
A writing assistant that can detect spelling errors, grammar issues, style issues and sentence construction errors in real time.
Facebook’s version of a chatbot and virtual assistant. Capable of performing many of the same functions as ChatGPT, but connected to Facebook and Instagram accounts.
Similar to Google Gemini, Copilot is integrated across Microsoft 365 products and can initiate a variety of tasks.
Developed by Google, NotebookLM can parse and analyze up to 50 documents and files (ex. audio recordings, transcripts, PDFs, course readings, research notes) and construct a variety of summaries for the user.A chat window provides the functionality of ChatGPT whereby the user can ask the AI questions or perform functions. In a separate window, called the studio window, the AI can construct a study guide, FAQs, a briefing doc or timeline. The AI can also summarize the data into a podcast.
Virtual assistant which can develop transcripts and summations from audio files of lectures, meetings, presentations.
Text-to-speech that converts written text into audible speech. Ideal for audiobooks, textbooks, websites, PDFs, presentations.
A social media management application that can help users plan and schedule posts across an array of platforms, like Facebook, Twitter/X, LinkedIn, Instagram, and others. Integrated tools, such as social listening, can help users develop strategies and content based on engagement and reactions and see which posts are resonating with audiences.
An AI assistant within Zoom that can summarize video calls in real-time and share a summary through email and Zoom chat; create a list of action items derived from the meeting; or if a portion of the call has been missed, to catch up and get acclimated without interrupting the call. The Zoom’s host can access the AI Companion at the start of the call, while attendees must request access from the host.
SOCIAL MEDIA AND AI
AI can be extremely useful as a guide or tool to reach audiences more effectively by recommending ideas, strategies, posts and hashtags or offering edits and suggestions to user-created content.
AI resources like ChatGPT can be beneficial for communicators looking to develop ideas for content and social media messaging, while management tools like Sprout can determine optimal scheduling of social media posts to better engage with audiences while tracking its impressions and reactions.
Some platforms, like Sprout, engage social listening tools which are driven by AI and can monitor impressions, reactions and comments of social media posts and provide a deeper understanding of audience engagement.
SECURITY AND ETHICAL CONCERNS
Everyday usage of generative AI applications can pose serious security risks and legal issues. University proprietary data, intellectual property (trademarked university or athletics department logos, branding or websites; scholarly works, research, data sets; confidential data), or personal data are prohibited from being implemented with AI usage.
GVSU’s Acceptable Use Policy for Public AI Solutions requires AI users to uphold all privacy laws and safeguard data, and comply with all GVSU policies and procedures. Because personal and private data may be collected by the developer to train, build and increase the AI’s functionality, FERPA data, PCI data, or any data protected by state or federal regulations should not be utilized.
BIAS WITH AI
As per GVSU’s Acceptable Use Policy for Public AI Solutions, “any users of AI or AI generative solutions should ensure such use does not introduce or amplify unjust bias.”
Generative AI’s ability to scan and learn from a variety of human-driven datasets like websites, periodicals and research papers, is transformative. But, it also leaves AI open to the possibility of biases.
For example, researchers at UCLA found that ChatGPT showed bias in the process of drafting letters of recommendation. ChatGPT used words like “expert” and “integrity” when drafting letters for male job applicants while it used words such as “beauty” and “delight” for female candidates’ letters.
Generative AI content should be shared with fellow communicators and supervisors to identify and mitigate potential biases and misrepresentations while representing a diverse population.