This set of principles and guidelines was developed to align the use of AI in communications and outreach across Brown.
TagsGuideline
AI Guidelines for Brown Communicators
Community Resources
This set of principles and guidelines was developed to align specifically with the use of artificial intelligence in communications and outreach at Brown University. As a leading institution for education and research, Brown has an international reputation for advancing inquiry and new knowledge, cultivating original ideas and heightened understanding, and making an impact through innovation and discovery.
As communicators at a university committed to academic excellence, we aim to ensure that any early approaches to AI remain aligned with this scholarly ethos of Brown. However, this is not a policy document. Brown’s Associate Provost of Artificial Intelligence is expected to provide leadership on AI activities, development, use and policy across the academic and operational missions of the University.
Principles and Values
The principles for using AI in communications at Brown aim to support effective adoption of AI tools and resources while also reflecting the same values at the core to Brown’s mission of “discovering, communicating and preserving knowledge and understanding.”
This means recognizing the benefits of AI — including the many possible uses outlined in the “Applications and Uses for Communicators” section below — while also adhering to Brown’s business processes and departmental standards.
This includes recognizing that AI can improve time management, contribute to workflow efficiency, accelerate content ideation, support analytics and analysis, and improve our own original content, among other benefits, if approved for use.
This includes observing any existing standards for receiving approval from supervisors or unit heads for purchasing and adopting new tools, software and platforms.
This includes following relevant University policies related to privacy, copyright, compliance, data security and other standards that guide daily work. References to many of these policies appear throughout these guidelines.1
This includes pursuing any relevant contract review, per University policy.
This means contributing to available knowledge with original content developed by members of our academic community in areas of scholarship where we are considered authorities on providing the information.
This includes avoiding the use of AI for content generation if there is an expectation that the content would have been generated by [you/the department] because of your subject matter expertise.
This includes avoiding the appearance of “repurposing” content generated by AI datasets in our department’s own areas of expertise.
This means using our own personal voice to speak to members of our community in times of unrest, distress, crisis or tragedy.
This includes developing original content that speaks to specific situations using our own knowledge of our constituencies, shared values and the operating context of the difficult issue about which we are communicating.
This includes sharing support and information in ways that are authentic and distinct to our department, unit and/or the University.
This means ensuring that AI-generated content does not infringe on existing copyrights and is not presented as original work.
This includes a prohibition against inputting any proprietary information into any AI tool, including proprietary visual content such as logos or other trademarked material.
This may include copyright-eligible content that has not yet been published and may be under an embargo or have other sensitivities.
Note: The U.S. Copyright Office is still working on its guidelines, but any content that is purely AI generated is currently not eligible for copyright – and the degree of human involvement needed to make something copyrightable has yet to be determined. Therefore, purely AI-generated materials should not carry the Brown copyright in any form of communication.]
This means ensuring that human editors review any AI-generated content for accuracy (of fact and tone, as well as addressing potential bias).
This includes detailed research and review of any “fact” presented from AI-generated prompts.
This includes ensuring that any content aligns with established priorities (of the University, school or department) messaging strategies, and editorial style.
This includes avoiding use of any images fully generated by AI (because accuracy cannot be checked), or AI-generated data animations whose source data cannot be fact-checked by a human.
This means citing sources using proper attribution and securing rights or any permissions needed by law for the use of someone’s image or likeness, copyrighted material, or any intellectual property or content raising any proprietary or ethical concerns.
This includes taking great care to make sure that any AI output used in editorial content has proper citations and has been reviewed for plagiarism (request citations in all prompts, and check those citations for accuracy).
This includes citing the use of AI and how it was used if the use of human-generated content in the same circumstance would have demanded citation (this does not include use of support tools for efficiency, workflow, content enhancement, coding, editing, ideation, definitions, etc.).
This includes ensuring that any AI-generated content complies with accessibility requirements.
This means ensuring AI is used as a tool to enhance human creativity and decision-making, not to replace it.
This includes asking the following questions:
Was AI used only to enhance the content, or to more fully develop it?
Was AI used only to generate initial ideas, or to generate the finished content?
Is AI baked into the functionality of a multi-functional tool or platform, or is AI being used specifically to generate the content itself?
Is AI making a manual process more efficient, or taking the place of human creation?
This means protecting any Brown data or personal information of groups or individuals.
This includes a prohibition against inputting names, birthdates, addresses, grades, performance ratings, etc. into any AI tool in ways that identify individuals, even for the purpose of arriving at trend data or graphics.
This includes a prohibition against inputting proprietary information into any AI tool (patent information, embargoed findings, etc.), as well as confidential incident reports, or legal documents or filings.
This includes seeking permission before feeding CVs, resumes or other professional documents of other individuals into AI tools (even if the individual is considered publicly-facing or has themselves made the choice to post their previous CV/resume online).
This means remaining aware that there is potential bias in AI algorithms and datasets.
This includes assessing AI outputs to ensure any final applications reflect Brown’s strategic commitment to supporting diversity, equity and inclusion in all aspects of University life.
This includes ensuring that no outputs incorporate false claims, ad hominem attacks, or biased or prejudicial information.
This means including provisions about acceptable uses of AI in contracts when vendors or contractors are performing “work for hire” that will be considered the property of Brown (and for which Brown will be liable).
This includes making it mandatory for freelancers to disclose if they use AI tools in their work. They should specify the AI tools used and the extent of AI involvement (e.g., idea generation, drafting, editing, and specifying what percentage of the work is AI-generated).
This includes explaining why disclosure is important: to safeguard Brown’s reputation for excellence, ensure transparency, allow for proper attribution, and help manage potential copyright or ethical concerns.
This means regular evaluation to continually assess the effectiveness and impact of AI tools in our work.
This includes evaluating the quality of our work for any adverse effects and reviewing any feedback from stakeholders and audiences.
This includes exploring professional development and training in evolving AI applications and tools, as resources allow.
Applications and Uses for Communicators
We view AI like other forms of support tools and automation: as an opportunity to improve the ways we work, enhance our work product, and reallocate time to achieve University-prioritized goals. Using AI tools can create time for problem-solving, creativity, strategic thinking, future goal-setting and strategic planning, and assessment and analysis.
However, we recognize there is no one-size-fits-all approach for the suggested optional applications and uses below, and much will depend on the specific communications vehicle and approach (news/informational vs. feature/marketing/creative vs. event/outreach). Communicators and event managers should consider for themselves what aligns with the academic and/or administrative functions of their units and the appropriate approaches to maintain integrity in communicating and engaging with their audiences.
Improving Time Management and Workflow Efficiency
This means using AI tools to increase productivity by automating repetitive or mundane tasks. Common successful uses:
Resizing images, organizing digital assets, etc.
Generating taglines, social posts from original content
Drafting or conceptualizing illustrations and animations
Automating user testing for web development and UX/UI design
Making platforms more intuitive and responsive
Improving search engine optimization (SEO)
Drafting headlines, subheadings and captions for informational/news content (for instance, in cases where these elements are intended to synthesize content vs. being core elements of a creative story package, where original creativity is encouraged)
Summarizing original long-form informational content: social posts, newsletter blurbs, news abstracts, research briefs, etc. (for synthesis purposes, as opposed to creative content for marketing purposes, where original creativity is encouraged)
Editing and proofreading: grammar and spelling errors, inconsistencies, and checking for potential plagiarism
Expanding areas of inquiry: “brainstorming” topics, interview questions, etc.
Suggesting content structure
Fine-tuning tone for different audiences (e.g., adjusting language of social posts tailored to varying age or other demographics)
Drafting templated and highly formulaic copy that translates or factually summarizes long-form documents (research briefs, etc.).
Drafting event layouts and stage design
Proposing event schedules
Developing lists of event needs and/or recommended vendor services
Calculating/estimating costs of varying groupings of services
Improving Work Product
This means using AI-tools to enhance or refine original work, including generating ideas and suggestions for subsequent original content development.
Enhancing images and videos and optimizing image quality (e.g., via photo editing tools, including those embedded in design tools and other platforms; not generating images)
Audio Visual Support: Translation, captioning, etc. (important to consider accuracy implications)
Expanding areas of inquiry: “brainstorming” topics, interview questions, etc.
Suggesting content structure
Fine-tuning tone for different audiences (e.g., adding a “younger” feel to a social post)
Refining the user experience (assessing outcomes)
Surveying attendees (i.e., developing questions to capture attendee information)
Tracking user trends, helping to tailor content to audience interests
Comparing user trends to peers; ideating on how to improve user engagement
1 In the 2024-25 Academic Year, the Office of Information Technology expects to engage stakeholders across campus to explore uniform approaches to data security for AI.