VISUAL SOCIAL MEDIA, FAKERY AND AI: AN EXPERIMENT AT THE INTERSECTION OF COMPUTER ART AND SOCIAL SCIENCE, USING DEEP LEARNING TO GENERATE AND IDENTIFY 'FAKE IMAGERY'
Fabrizio Poltronieri, Max Hänska
Visual social media is attracting ever greater attention, particularly as visuality intersects with artificial intelligence (AI). Recently concerns have arisen that artificially generated visuals could supercharge mis- and disinformation online. Sitting at the intersection of artistic practice and social scientific research, this experimental paper explores questions around computer generated visual artefacts, aesthetics, fakery, and wider socio-political concerns about the inexorable rise of visual online communication. Using deep learning techniques, the paper experiments with 'fake' visual content. We use Generative Adversarial Networks (GAN) to generate artificial images (e.g. selfies), and Convolutional Neural Networks (CNN) to try to distinguish real from fake. Furthermore, we use our CNN to scan 6 million images collected from public Twitter posts to try to determine the proportion of fake images.
The collaboration between artistic practice and social scientific research has proved invaluable in exploring the role of intelligent systems in shaping visual communication. It shows us that autonomous systems are able to generate photo-realistic fake images, but much less capable of determining which images are fake and with are real without systematic human guidance.
ONLINE MANIPULATION
Daniel Susser, Beate Roessler, Helen Nissenbaum
Privacy and surveillance scholars increasingly worry that data collectors can use the information they gather about our preferences, interests, incomes, and so on to manipulate us. Yet what it means, exactly, to manipulate someone, and how we might systematically identify cases of manipulation needs to be more thoroughly explored in light of the unprecedented capacities that information technologies and digital media enable. In this paper, we develop a definition of manipulation to address these enhanced capacities, investigate ways information technology can be used to facilitate manipulative practices, and describe the harms—to individuals and to social institutions—that flow from engaging in such practices. Specifically, we argue that manipulation undermines autonomy, both directly and indirectly. But since we value autonomy differently in different social contexts, we must carefully distinguish between the contexts in which manipulative practices operate.
FAKE OR REAL? HOW VIEWERS EVALUATE IMAGE CREDIBILITY ONLINE
Cuihua Shen, Mona Kasra, Grace Benefield, Wenjing Pan, Yining Zhou, James F. O'Brien
Due to the scope and speed of information dissemination across social media websites, visual misinformation is capable of manipulating crowds, propagating hysteria, confusion, distress, panic, violence, and escalating chaotic mass behavior at a fast pace and on a large scale. Yet we know distressingly little about how online viewers assess digital images and make judgments and decisions about their authenticity. This paper details a large-scale online experiment of image credibility on Amazon Mechanical Turk that probes how people react to, respond, and evaluate the credibility of images that accompany online stories in internet-enabled communication channels (social networking sites, blogs, email). We ran a series of six between-subjects experiments, each of which randomly assigned participants to one of 28 news-source mockups featuring a forged image, and asked participants to evaluate its credibility based on various features. The results were consistent across all six images tested, showing that that internet skills, photography and digital imaging experiences, social media use, and pro-issue attitude are significant predictors of credibility evaluation of online images. Our study is among the first to test the social and cognitive heuristics of information credibility and evaluation in the context of image authenticity online.
