Artificial Intelligence

Generative AI tools, such as ChatGPT, are becoming increasingly commonplace and have tremendous potential to change the way university communicators—writers, editors, photographers, videographers, graphic designers, etc.—approach their day-to-day work. When given a prompt by a human user, this technology can generate content such as images, text, music, or video. These tools can increase efficiency, assist with mundane tasks, and serve as a useful springboard in creative brainstorming.  

However, we must also acknowledge the inherent risks of using generative AI, such as errors, misinformation, bias, copyright infringement, and breach of privacy. 

While we encourage exploration and openness to generative AI, in order to uphold Johns Hopkins’ reputation as one of the most trusted brands in the world, it is essential that humans remain central to the research, creation, editing, and review of content.  

• Guiding Principles

Guiding Principles

AI should be assistive, not autonomous. 

Open Close

Artificial intelligence can be used as a tool to assist university communicators, a starting point for content generation and knowledge gathering; it should not replace human effort.

AI does not possess the ability to reason or empathize, nor can it replace human creativity and imagination. Furthermore, generative AI is known to “hallucinate,” creating nonsensical, false, or misleading results. There is also a high risk of plagiarism, bias, and copyright infringement.

We recommend publishing imagery, written content, music, audio, or video that is created by humans.

• Written content: We do not recommend using generative AI to create wholly new pieces of written content.
• Imagery: Imagery created by humans is preferable. However, assets created using trusted, licensed third-party platforms, such as AI-generated images available via Getty, are acceptable with proper citation/credit when human-generated assets are not available. AI should not be used to create photorealistic images of people, or to create any art that is deceitful or misleading.

Generative AI tools are most useful in ways that augment human ability, such as brainstorming copy, creating art direction mood boards, and speeding up routine tasks (e.g., drafting metadata).

All AI-generated material should be reviewed, edited, fact-checked, and overseen by a human. The person using the AI tool should take accountability for all work created using AI.

Transparency is key. 

Open Close

Disclosing the use of AI in published work is paramount to maintaining the trust of our audiences and colleagues.

Employees and contractors should discuss any use of AI tools with supervisors prior to beginning work.

Public-facing AI-generated content should include a citation, such as “AI Image generated by: [source].”

AI should not be used in a way that intentionally misleads or deceives the audience.

Privacy must be protected. 

Open Close

Assume that AI tools are not encrypted or private. Protect confidential information when using them. Any information you enter into an AI platform can be used to train that tool, and that information can then be repeated back to other users.

For this reason, do not enter any personally identifiable information (names, addresses, salaries, etc.) or proprietary information (institutional data, logos, etc.) into generative AI tools unless approved by Johns Hopkins IT.

Exploration is encouraged. 

Open Close

A spirit of curiosity, exploration and experimentation—along with a healthy dose of skepticism—is essential when using artificial intelligence.

Artificial intelligence technology is rapidly advancing, and we anticipate that it will continue to become more and more commonplace in professional settings.

It is recommended to keep up with the latest trends and best practices in generative AI by seeking out opportunities for education and training, reading reputable news sources, and regularly checking back for updates to this page. We also encourage teams to invest in ongoing training, whether internal or external.

Implementing AI Across the Enterprise

Interdisciplinary teams of stakeholders from Johns Hopkins Medicine and Johns Hopkins University are meticulously curating guidelines for responsible integration of AI across the enterprise. With a deliberate and thoughtful approach, these groups are navigating the complexities of this evolving technological landscape. Please review these general guidelines before engaging with or promoting the use of a new AI tool. 

IT at Johns Hopkins

USE CASES

The AI landscape is constantly changing. Therefore, we do not recommend specific tools but rather provide direction on how best to approach the use of generative AI. The following lists offer just a few examples of how AI can assist in your role, while cautioning against overreliance on any AI tool. All content generated by AI should be followed by human review and fact-checking. 

Acceptable Use  

• Brainstorming story ideas, headlines, or art direction 
• Creating mood boards for new projects and campaigns 
• Serving as a thesaurus 
• Drafting alternative text, metadata, and video descriptions 
• Speeding up routine tasks such as transcript generation or closed captioning 
• Designing internal-only proofs of concept and mock-ups  
• Making minor adjustments to photos and illustrations, such as sharpening the image, extending a nondescript background, or removing a temporary blemish from a person’s face 
• Creating templates or outlines, such as converting a Word Document into a PowerPoint presentation 
• Capturing meeting notes, with sensitivity to participant awareness and meeting topic  

Proceed with Caution

• Answering grammatical questions 
• Conducting preliminary research  
• Taking a first pass at copy editing  
• Coming up with a list of SEO keywords 
• Creating new imagery with a licensed generative AI tool 
• Translating foreign text 

Not Recommended 

• Drafting public-facing editorial or marketing copy, such as news articles or brochures  
• Fact-checking  
• Creating artwork in the style of a certain artist or photographer 
• Creating on-brand designs for communications and marketing materials
• Recreating or altering official university trademarks, mascot, or fonts
• Creating a photorealistic image that could be misunderstood as representing reality 
• Fundamentally altering the integrity/truth of a photograph  
• Working on sensitive or private projects 

---

ABOUT THIS PAGE 

These guiding principles were developed by a task force of communicators from the Johns Hopkins University Office of Communications, with input from communications professionals across the institution. It was also informed by the guidelines of more than a dozen peer institutions. These guidelines do not represent official university policy. 

This page was created as a resource for communicators from across Johns Hopkins University who are curious about the ethics of using generative AI in their professional roles. It is not intended to serve as guidance for students or faculty members, nor does it provide IT-related guidance on safe/recommended tools. We recommend reaching out to Johns Hopkins IT with questions and concerns about the safety of a particular generative AI tool. 

AI and the laws/standards surrounding its use are constantly changing. This page will be periodically revised, with updates posted to the JHU Brand Guidelines Blog and sent out via the Brand Guidelines Mailing List. 

If you have any questions about these guidelines or suggestions for future iterations, email [email protected].