One of the most under-scrutinised aspects of the rise of LLMs and foundation models is their transformative impact on work done through digital interfaces. The graphical user interface (GUI) in particular, whose established ways of interacting had seemingly stabilised over the last three decades across both desktop and mobile, is undergoing rapid change. Generative models, no longer confined to questions-and-answers chat interfaces (Grosser and Pold 2025), are being amalgamated more explicitly as co-actants - not only in dedicated new spaces, such as OpenAI’s Codex or Anthropic’s Cowork, but also through various forms of integration into a growing range of existing software, such as IDEs, design tools, email clients, spreadsheet editors, and word processors.

In this reading group, we look at the rapidly unfolding changes in where and how users meet AI. We read for systems rather than models: what users encounter is never a model as such, but an assembly: weights bundled with (hidden) system prompts, file retrieval, tool access, data pathways, normative assumptions and defaults, guardrails, and pricing tiers, held together by interfaces that connect the underlying technologies to actual sensory arrangements, mental models, felt user practices, and social ramifications. Reading AI this way means thinking the backend from the interface and back again. We ask: how can thinking “interfacially” help us understand the work GenAI does in the world? What forms of engagement are presupposed and afforded by the emergent user interface (UI) and experience (UX) designs? And what (updated) forms of media studies and interface critique does this approach make way for?

The group builds on repertoires of foundational interfacial scholarship (Drucker 2011; Galloway 2012; Farman 2012; Hookway 2014; Ash 2015; Dieter 2024); critical exploration of foundation models (Luitse and Denkena 2021; Burkhardt and Rieder 2024); and political economies of cloud and AI technologies (Narayan 2022; Van der Vlist et al. 2024; Gekker et al. 2026). We encourage literature and case study suggestions from members. The goal is a broader understanding of the conditions of mediation of computer work in the generative AI era: one that takes seriously the interplay between designed intentionality and lived user practices.

We meet around once a month. Our next meeting will take place on November 3, 2026, 10:00 – 12:00 at BG1 (room announced per email). We will be discussing excerpts from Galloway’s (2012) The Interface Effect and Pold & Grosser’s 2025 paper, “Reading the Praise/Prompt Machine: An Interface Criticism Approach to ChatGPT”. To join us and receive updates, please email a.gekker@uva.nl or d.vrabicdezman@uva.nl.