What can Rhetorical Genre Theory tell us about AI and misinformation? 

**Quick link to article: https://journals.sagepub.com/doi/10.1177/00472816231226249**

Back in 2022, Brad Mehlenbacher and I were on sabbatical, and we spent some time looking at the then-recently launched ChatGPT, and back at GPT 3. We completed this work with the always outstanding and brilliant Ana Patricia Balbon.

All the talk about identifying what text was AI-generated or not made us think about what rhetoric and language scholars would call genre. In rhetoric, the idea of genre suggests that there are typified forms of texts (or visuals, videos, etc.) that we use to respond to recurring situations (Carolyn R. Miller theorized rhetorical genres in her foundational 1984 article Genre as Social Action). A common genre we can think about in research communities is the scientific research article.

Here Brad, Patricia, and I were in 2022 thinking about how generative Artificial Intelligence tools might replicate features of scientific research articles. Our exploration of genre features in genAI output wasn’t just expert-focused, however. What we really wanted to know is are there enough of those genre typifications we’d expect to make a text passable to non-specialists in some area. Specifically, could these tools be used to create scientific misinformation? We explore this in the article.

What can genre offer to the genAI conversation in addressing both its promise and perils? We suggest:
·     The study of scientific genres can help us explain the process of science and how scientific knowledge is collectively advanced,
·     Furthermore, the study of scientific genre-ing activities (all those compositional and rhetorical activities scientist engage in) helps us understand norms and values (for instance, that scientists hedge in their writing is important or that healthy skepticism is part of science),
·     That Miller’s insight about how genres mediate private intentions with social needs is a critical distinction in the work that humans undertake in their compositional and rhetorical activities,
·     With AI-generated texts being used, genre ecologies may be influenced by these typifications and that might, then, influence human-generated typifications.

Another key idea we begin to explore as we thought about these tools is the idea of “synthetic genres.” We are interested in continuing to explore this idea, asking what the implications for genres are as genAI outputs may influence the ecology of genres we all use.

Come for the genre theory perspectives and stay for the tennis metaphors. We draw on the brilliant Anne Freadman’s work, of course, and following that, Brad’s love of the game.

Check out the open access (CC BY-NC 4.0) version of the article “Synthetic Genres: Expert Genres, Non-Specialist Audiences, and Misinformation in the Artificial Intelligence Age” after the link! https://journals.sagepub.com/doi/10.1177/00472816231226249