Activity: Talk or presentation › Invited talk › Academic
Description
The lecture Prompting as Research, delivered within the Digital Methods and Information Analytics course at the University of Amsterdam, introduced students to prompt design as a methodological approach within Critical AI Studies. Moving beyond “prompt engineering,” the session framed prompting as a way to interrogate generative AI systems, focusing on how techniques such as ambiguous, provocative, and reverse-engineered prompting can be used to study bias, content moderation, and underlying training data.
Through a detailed biodiversity case study, the presentation demonstrated how generative models produce stereotypical and uneven representations of the world, and how these outputs can be systematically analysed using digital methods. The session combined conceptual framing with practical guidance, equipping students with strategies to design prompts, generate datasets, and critically interpret AI outputs as part of research workflows.