Between invocation and agency: AI deep research as a scholarly problem

This probe explores the experience of writing alongside ChatGPT’s Deep Research feature and asks what becomes of authorship, agency and ethical responsibility when scholarly work is performed through negotiations between humans and AI systems. Drawing on media ecology and actor-network theory (ANT), the probe theorizes advanced large language models as invocational actors : non-human entities summoned through prompts that appear to undertake autonomous intellectual labour. The argument traces the concept of translation across three domains – machine translation in AI research, McLuhan’s media-ecological understanding of translation as perceptual transformation and ANT’s account of translation as the enrolment and transformation of actors within networks. Using Deep Research as a case study, the probe examines how translation gives way to the appearance of agency, unsettling established norms of scholarly authorship and intellectual labour. Rather than offering a definitive position for or against AI-assisted research, the probe concludes by proposing provisional heuristics for ethical knowledge production that resist full delegation while acknowledging the long history of non-human participation in scholarly work.