Alan Liu

“Is Critical Infrastructure Studies of Big AI Possible? — The Critique of AI (II),” HASS Research Infrastructure in an Age of Artificial Intelligence workshop, HASS Digital Research Hub, Australian National University, 20 August 2026.

• Abstract: In this short keynote, Alan Liu revisits two key premises of critical infrastructure studies (CIS) that he has previously asserted. One is that “infrastructure,” as approached in CIS, carries on the edgy, critical function of “culture” in “cultural studies” from a half century ago. The second (in the introduction to Critical Infrastructure Studies and Digital Humanities by Liu, Pawlicka-Deger, and Smithies) is that CIS extends the lineage of interpretive methods whose “objects of study” were ideas, language, culture, media, and identity—except that CIS’s object is a literal one, inviting both critical thinking and critical making. However, Liu asks: does today’s big AI—because of its scale, complexity, cost, intellectual-property issues, and other challenges—now stop CIS in its tracks? To do AI in order to critique AI, will CIS have to be satisfied with just “cultivating its garden” (in Voltaire’s phrase from Candide) by working with non-mainstream local, open-source, institutionally “walled,” or toy AIs? Or are there less proper, fastidious, and clean—but still acceptable—ways of doing/critiquing big AI?

“Modernity, Critical Infrastructure, and the Limits of Critique — The Critique of AI (I),” Digital Modernities, Alternative Futures symposium, HASS Digital Research Hub, Australian National University, 17 August 2026.

• Abstract: What now is the right balance in the humanities between critiquing AI in the manner of the “hermeneutics of suspicion” and embracing the promise of AI? The “now” framing this question is not just the present moment but the longer history of modernity constituted as critique, and, recursively, of the modern and postmodern humanities (and adjacent social sciences) as the critique of modernity itself. Alan Liu proposes as a thought-object for considering this question something that is difficult to square with the problem-space of humanities critique: “critical infrastructure” in the sense of cannot-fail power grids, transportation systems, water and sewage systems, financial exchange networks, communication and information platforms, and, increasingly AI. Such infrastructures embody a notion of criticality that is orthogonal to humanities critique. Using a digital “kit” of word embedding models, topic models, and other models of a corpus of 537 recent articles on AI, Liu illustrates how the humanities might explore the larger critical problem-space of the “AI modern” as a case study of the critique of the modern.