The Center for Public Interest Communications at the University of Florida College of Journalism and Communications is committed to science-based, ethical communication that can be scrutinized, replicated and trusted. We treat generative artificial intelligence tools as supporting instruments that help us refine and extend our own research insights, rather than as independent scholars, authors or decision-makers.
How we do not use generative artificial intelligence
- We do not use generative artificial intelligence tools to independently create deliverables or submissions for any sponsored product, including fee-for-service projects, consulting agreements or commissioned content unless specifically requested by our partners as part of a funded project at their direction to test an idea or concept.
- We do not use generative artificial intelligence tools to write articles or peer-reviewed research manuscripts that carry a human byline from Center staff, faculty, students or collaborators. Full human authorship is essential for these products.
- We do not use generative artificial intelligence tools to simulate the perspectives, reactions, or feedback of specific demographic or psychographic groups. (For example, by prompting a model to respond as a 30-year-old Latino man in Santa Rosa, California, or as a values-driven environmentalist skeptical of government intervention.) Research shows that this may actually amplify stereotyping by exaggerating the characteristics assigned to a group. When the Center conducts message testing, actor research, or any inquiry that requires understanding how people think, feel, or respond, we work with actual human participants, recruited through vetted research panel services.
How we may use generative artificial intelligence
- We may use generative artificial intelligence tools to help refine and extend insights from our prior work and research themes into more actionable formats for the field of public interest communications. These outputs will always be free to access by the public. The outputs will be reviewed and edited by a senior Center team member, and labeled in the publication information area as “prepared in collaboration with [LLM model name].”
- We may use generative artificial intelligence–enabled tools (for example, narrative analytics or social listening platforms) to help us make sense of large data sets, always with human review by Center researchers before any findings are shared. Where generative tools materially shaped findings or interpretation — such as when a tool was used to identify thematic patterns across data, assign sentiment classifications, or surface clusters in social listening data — we will disclose their role in a methodology section or footnote. That disclosure will describe which tool was used, what analytical function it performed, and how Center researchers reviewed or validated the output before it informed any finding.
- We may use generative artificial intelligence tools to help identify additional research, theory and evidence that our team then evaluates, verifies and, where appropriate, integrates, drawing on the Center’s established practice of grounding communication in peer-reviewed and empirically tested insights.
Standards for evidence, replicability and human oversight
- All Center content must be replicable, evidence-based and traceable to sources that a human researcher can locate, review and re-create using documented methods. This includes maintaining clear citations, methodologies and data sources that meet the expectations of a Research 1 institution.
- We will not publish content that relies on artificial intelligence–generated sources that cannot be inspected, traced or independently verified by humans.
- We do not use artificial intelligence–generated avatars, synthetic spokespersons, photographs, or artificial intelligence–generated “user” content in our materials. Our work is grounded in real people, real communities and empirical evidence.
The Center’s posture is one of cautious experimentation: when we use generative artificial intelligence, we do so in tightly defined, supervised contexts, with human review by a senior Center team member to identify and mitigate risks before any output is shared.
Our commitment to ongoing review
The Center will update this statement as tools, risks, evidence and professional standards evolve, while maintaining our core commitments to transparency, accountability, scientific integrity and the centering of human actors in public interest communications.
Updated: 3/9/2026
Prepared in collaboration with Perplexity, powered by GPT-5.1, and Anthropic’s Claude Sonnet 4.6