Question. Answer. Repeat.

media release: This event is hosted by the UW-Madison Center for Humanistic Inquiry into AI and Uncertainty and supported by the RISE‑AI Collaboration HQ. Additional sponsors include: the Data Science Institute; the Institute for Research in the Humanities; the Digital Scholarship Hub; the Center for the Humanities; and the Division of the Arts.

Generative AI can take many forms and interactional logics, and more are emerging every day. But the prompt-response format, common to many GPTs and image generation tools, and has played a formative role in the public’s encounter with and understanding of these technologies. This particular structure of interaction invites the user into a specific form of sociality with AI and with the data corpus on which it is trained. The prompt-response form is intuitive, and yet was not inevitable; it is in one sense accidental, and in another a nearly foregone conclusion. Delivering generative AI to the public in the form of a chatbot may be the original sin of this latest AI revolution, or one of them.

In this talk, Tarleton Gillespie will trace its antecedents, technologies that use the prompt-response form to structure users’ relationship to a massive corpus of information. Expert systems, social media bots, and particularly the evolution of search engines reveal a longstanding commitment towards this specific modality for information retrieval. Then, turning to the study of media, he will consider what we know about artificial persona to which users may nevertheless feel connected: from puppetry to fictional characters to game avatars to celebrities. Media scholarship has grappled with how artificial persona can nevertheless be an object of parasocial engagement, while cloaking both the artifice and the corporations looking to profit from it.

These antecedents, both as intellectual orientations to information and as specific answers to political economic imperatives, prefigure generative AI, normalizing anthropomorphized chat as a way to relate – with software, with fictions, with data, and with the corporations that exploit it. Highlighting these persistent logics, set in motion decades before generative AI ever emerged, can expand our sociological lens about the newest examples.

Tarleton Gillespie is a senior principal researcher at Microsoft Research, and an affiliated professor in the Department of Communication and Department of Information Science at Cornell University. He is the author of Wired Shut: Copyright and the Shape of Digital Culture (MIT, 2007), co-editor of Media Technologies: Essays on Communication, Materiality, and Society (MIT, 2014), and author of Custodians of the Internet: Platforms, Content Moderation, and the Hidden Decisions that Shape Social Media (Yale, 2018). His current work interrogates the turn to generative AI for its links to histories of media and social media.