An exhibition of generative AI application in South African design curricula
Curated by Kyle Rath (Information Design, University of Pretoria, South Africa)
AI is like a flea. For designers, at least, it is often intangible, but irritatingly always there. It leaves us with a tempting itch to generate neatly packaged, superficially engaging images that manage to provide a cheap, quick and then scrollable thrill. It automates processes, taking over tasks once done by hand, leaving designers chasing after their own work. What was once the domain of human-driven conceptualisation is now influenced by algorithms and data, pushing pixel reproduction to even colder, more calculated spaces. The leaps generative AI takes can feel erratic, disorienting, and even overwhelming - transforming design into something driven more by technology than by human agency.
Thanks to ChatGPT for generating the above paragraph.
There is likely a feeling of resentment, or at least mild irritation that most audiences will feel, at the above paragraph (Dwork & Minnow 2022). At the very least, there may be the feeling that we have been cheated, or duped into believing, momentarily, that the content that we engaged with, took something more than the effort of a single prompt (compare AI in design to a flea). That in the context of an academic teaching environment, it is somewhat thoughtfully produced. Instead we are seduced by the allure of content as a kind of hyperreal simulacra; we read artificial signs not as a simulation of the real but as the real itself (Baudrillard 1988:94, 1990:149). It is a ‘reality’, however, that is generated by little more than binary pattern sequencing and term frequency (Marivate 2023).
There are of course far more immediately devastating consequences of ‘generative’ AI: extant anxieties ranging from the fear of job loss, to social implication of racial and gender biases, to faulty or intentionally manipulative data sources, and of course risks commensurate with global disease outbreak and AI assisted warfare.
These are all very real consequences, and for designers, who are trained to coerce the way humans interpret data, engaging in generating AI assisted design has very real ramifications. AI is undeniably reshaping the design landscape. Moreover, it is not a case of if we should integrate AI into our courses, as it is very much here already (and, I might add, it has been here for the past few decades). The question is thus not if, but how we, as educators, are integrating AI into design education, in a meaningful way. On the one hand, it can potentially equip students with cutting-edge skills, preparing them for an evolving industry, while on the other, it can foster irresponsible and ethical design practices (by way of intentionally deceptive design practices, or worse, lethargy).
The following is an exhibition of a generative AI design project, whereby 4th year Information Design students were ‘forced’ to develop a corporate identity system by means of generative AI software alone (ChatGPT for copy generation and a host of image generators for the visual design). The focus, however, was not on the extent to which these AI generated identity systems actually served as viable branding tools. In fact, the choice of developing a corporate identity system versus, perhaps, a wayfinding system or animation for instance, was more or less arbitrary (or at best, a choice based on convenience of time). Instead, the aim of the project was to test the limits of current AI systems, to see where and to what extent AI could think on behalf of the designer. More importantly, the idea is that students had to document the ‘thinking process’ that they engaged in, ‘alongside’ various generative AI systems. Instead of fighting against the use of AI in an instructor/learner or teacher/student dichotomy, the intention is to put the student in a directorial position. That is, students often turn to AI as a way of trying to meet a deadline or lessen apparently extraneous hours of labour and so it typically becomes a game of mirrors; will the lecturer spot my use of AI or ‘will I get away with it.’
As the second part of the project, students were required to map their process by way of a self-generated infographic, documenting their frustrations and successes with the generative AI platforms. As is made clear in the exhibition, the students were given a space to make extensive use of and then reflect upon the use of generative AI software, and thus more broadly, to reflect on their role as information processors in an era of ever increasing, rapid generative sign production. As design student, Hendrik Smith notes:
In a way, every project facilitates a conversation between the designer and design. The will and vision of the designer engages in a vicious tug of war on the infinite possibilities a design presents until an uneasy equilibrium is reached. In the case of this project, a third party joined the battle, but it's unclear whether it pulled for the designer or sided with chaos. AI shines in streamlining time consuming tasks, providing the spark for creative ideas and even works as a buddy for talking through an idea. There are some shortcomings though. At the time of the project, AI’s creativity (if it can be called that) was severely limited, and its responses bland. So I decided to push the absurdity of my ideas to see if ChatGPT and its AI friends could keep up with human whimsy.
Interestingly, it appears that a common theme amongst several of the students tends to be AI’s ineptitude for humour and absurdity. Bronwyn Sinclair’s Artificial Invasion for instance, is a tongue and cheek critique on the late (2023) adoption of Artificial Intelligence into the creative and artistic industries. In particular, Sinclair examines whether machines (as a physical metaphor for generative AI) can adapt themselves to the realm of the ridiculous, but at the same time be constrained by way of careful prompting.
Similarly, Johan Steyn implements humour in The Atrium, a brand intended to mimic a crime investigator's table, detailing the story of a fictitious company involved in money laundering, while visibly showcasing the company's genesis through generative AI. The focus here was on how the AI was utilised to establish this ‘fake brand’ as a real company. Steyn notes that during the process of generating images of the fictitious locations, alongside actual place names, he felt a sense of eliciting a kind of forgery, especially considering the potential traceability to a unique IP address.
Rachel van Zyl’s Designer in the Dark centres on Shadow, a brand created for a restaurant that offers a dark dining experience, serving gourmet desserts to blindfolded guests. The intention is to investigate the degree to which AI could navigate the ethical and cultural sensitivities that would inevitably arise when branding for such a specific target group. As Van Zyl points out, with ChatGPT suggested names like Moonshadow Munch and trypophobia-inducing images of strawberries, AI’s involvement in the creative process highlighted the need for the empathetic designer.
Along with that of Taeya Duke, Marcholette Minnarr, Eshile Zungu and Lean van den Berg, the projects demonstrate a variety of different approaches to investigating the limits and affordances of AI, and as generative AI continues develop the project continues to adapt and expand as part of the Information Design course at the University of Pretoria.
In its restless movement, AI forces design to keep up, for better or worse. As this project develops over the next few years, the hope is that it provides something of a ballast against which design students conscientiously reflect on the pitfalls and successes of generative AI, their roles as directors of information in an era that appears to preference generative visual candy over thoughtful design, as well as the interdisciplinary nature of human creativity and machine ‘thinking’.
References
- Baudrillard, J. 1988. On seduction, in Jean Baudrillard: Selected writings, edited by M Poster. Translated by J Mourrain. California: Stanford University Press: 149-165. (Original text published in 1979).
- Baudrillard, J. 1990. Seduction. Translated by B Singer. Montréal: New World Perspectives. (Original text published in 1979).
- Dwork, C & Minnow, M. 2022. Distrust of Artificial Intelligence: Sources & Responses from Computer Science & Law. Dædalus, the Journal of the American Academy of Arts & Sciences 151(2): 309-321.
- Marivate, V. 2023. AI and language: a mirror to ourselves. Paper presented at the Generative AI and the Humanities: Addressing the Challenges for Teaching and Learning conference, 12 May, University of Pretoria, South Africa.
- Poster, M (ed). 1988. Jean Baudrillard: Selected writings. Second edition. California: Stanford University Press.
