Marking the end of the Human Futures research program after nearly six years of collaborative projects dealing with technologization from a Humanities perspective, scholars from Aarhus and abroad were gathered for the closing conference “AI and the Humanities: Emergent fields, critical perspectives, ethical implications” this week in Aarhus (and online). Six different panels provided the framework for various discussions of current issues and questions at stake within the Humanities as a result of the fast-moving development of AI and machine learning.

Creativity and Visual Culture

A panel on “AI and Creativity” opened the conference on September 6 with presentations dealing with the impact of AI in fields of creativity and artmaking, e.g., how AI alters notions of creativity and influences the art industry. Starting off with Pei-Sze Chow’s (Assistant Professor, University of Amsterdam) talk on the role of algorithms in the film industry, Nicola Bozzi (King’s College London) continued with a discussion of the emergent format of AR face filters as a form of digital performativity and AI-driven art on social media. Imen el Bedoui (Assistant Professor, University of Kairoun) followed with a discussion of revisions of artificial visual perception through contemporary artworks, whereas Carlo Forlivesi (composer, organist, and professor) questioned how much AI can be said to emulate the “world of humans” and vice versa. Jan Løhmann Stephensen (Associate Professor, Aarhus University) concluded the panel by discussing the case of Marcel Duchamp, arguing that artificial creativity and artmaking implicates what can be determined as a set of “Duchampian” questions.

With a total of seven speakers, the “AI and Visual Culture” panel was divided across the two conference days with a wide range of talks exploring the interrelations between AI and contemporary visual culture. Mette-Marie Zacher Sørensen (Associate Professor, Aarhus University) initiated the panel focusing on deepfake animations of dead people as an example of how bodies in a moving image not only affect but are, in fact, affected themselves. Amanda Wasielewski(Docent, Stockholm University) continued with a discussion of whether AI-generated photorealistic images of people can be considered photographs and how they affect the field of photography. This was followed by a talk on AI-driven art with artist-researcher Asker Bryld Staunæs discussing how theory and practice, science, and artwork can lead to a more consistent, imaginary, and investigative art for AI. Ending this first part of the panel on visual culture, Daniel Chavez Heras (King’s College London) introduced the notion of ‘computational ekphrasis’ as a way of understanding the current development of text-to-image models.

Perle Møhl (Aarhus University) continued the panel the following day, presenting her research on human and AI vision in border control and radiology. Next, Lotte Philipsen (Associate Professor, Aarhus University) discussed the aesthetic implications of Google Arts & Culture’s AI image methods, followed by Jamie Wallace (Associate Professor, Aarhus University) who rounded off by considering the visual culture of facial expression analysis and the implications and predispositions of the human-machine gaze.

Knowledge & Learning

Day two of the conference was opened with a panel on “Knowledge and AI” and an introductory discussion by Peter Danholt (Associate Professor, Aarhus University) on how to think about and with AI and the digital in a more-than-human ontology. Antonio Picone (Ph.D. candidate, University of Catania) continued by presenting a sentiment analysis project[1] based on a large dataset containing tweets about the war in Ukraine, using this as an example of the risk of human-like biases in the implementation of AI technologies. The second part of the panel revolved around the concept of learning with Robin Auer (Research Assistant, TU Braunschweig) stressing the notion of embodiment as a fundamental part of learning for both humans and embodied AI (robots). Cathrine Hasse (Professor, Aarhus University) continued by discussing how AI and machine learning can bring new perspectives to our understanding of human learning, emphasizing the point that humans are ultra-social learners.

Text & Literature

Both conference days were further marked by a recurrent focus on text and literature. Initiating the panel “Automated Text Generation”, Henrik Køhler Simonsen (Director of Learning, Danish Insurance Academy) focused on the necessity to study AI-text generators as they increasingly impact how students and professionals approach text-writing. Ulf Dalvad Berthelsen (Associate Professor, Aarhus University) followed by presenting a study demonstrating that a recently released GPT-3 model is, despite a rather simple vocabulary, in fact capable of generating high-quality academic writing. Finally, Şule Akdoğan (Assistant Professor, TED University) presented her and Özsel Kılınç’s (Honorary Research Fellow, University of Warwick) exploration of bias reproduction in GPT-3 generated texts based on prompts from Daniel Defoe’s Robinson Crusoe (1719) and John Ruskin’s Of Queen’s Gardens (1865).

The conference also hosted a distinct panel presenting a data analysis project based on the 2021 virtual ELO (Electronic Literature Organization) conference: “Platform (Post?) Pandemic.” The group - Christian Ulrik Andersen (Associate Professor, Aarhus University), Malthe Stavning Erslev (Ph.D. student, Aarhus University), Pablo Rodrigo Velasco González (Associate Professor, Aarhus University), and Søren Bro Pold (Associate Professor, Aarhus University) - presented their project as twofold – firstly, as a way of mapping and analysing the platform conference through data analysis, secondly as a way of viewing the data analysis as a form of electronic literature in itself, suggesting that the platform conference could indeed be “seen” or “read” as a “data-orama.”

Finally, the panel “AI and Literature” rounded off the conference with a series of talks on Kazuo Ishiguro’s Klara and the Sun (2021): Shoshannah Ganz (Associate Professor, Memorial University of Newfoundland) focused on the cyborg-human reciprocity represented in the novel as it moves far beyond the question of what the cyborg can do for the human, whereas Sheng-mei Ma (Professor, Michigan State University) offered a reading of Ishiguro’s representation of the more-than-human as linked to a cultural mourning and melancholia resulting from expatriate experience. Eckart Voigt (Professor, TU Braunschweig) also touched upon the novel as he juxtaposed the current aesthetic practice of ‘Robot Literature’ with the cultural imaginary. In line with this, Tom Halford (ESL instructor, Memorial University of Newfoundland) discussed how distance is invoked as an enabling rather than repressive notion in Larissa Lai’s Automoton Biographies (2009) and Sybil Unrest (2008), co-authored with Rita Wong.

All in all, the conference highlighted an important intersection of humanities and technology through AI and machine learning, which was not as high on the agenda just a few years ago but which is now having a pronounced influence on a number of domains in the humanities. Human Futures is happy that Shape will continue to explore this field at Aarhus University and with our many cherished colleagues and partners.