It’s impossible to miss the phenomenon of generative Artificial Intelligence (AI), which is developing at lightning speed, marking an essential turning point in the history of the artistic professions. Its beginnings in the 2010s revolutionised graphic design by making it easier to retouch images, recognise patterns and automate repetitive tasks. AI has evolved to offer advanced capabilities such as automatic generation of visual content and personalisation to meet the needs and preferences of a wide audience. Its impact on graphic design is profound, overturning methods of creation, collaboration, the notion of creative time and even copyright. This major advance opens up new horizons for innovation and visual creativity. If AI offers unprecedented possibilities, what relationship can it have with the creative professions? Is it a valuable ally for creative people, or a formidable competitor?
Time savings and productivity: advances in graphic design thanks to AI
Two entities, Dall-E and Midjourney, have emerged as essential pillars for creatives, alongside Chat-GPT, in the evolution of AI. It first entered open beta on 12 July 2022, and The Economist published the first magazine cover using it in June 2022.1
Cover of The Economist, June 2022
Since then, AI has continued to evolve, offering considerable time savings by automating repetitive tasks, creating personalised content and improving overall productivity. In a society where everything has to be more and more immediate, faster and faster, and where the attention available to our audience is shrinking considerably2 everyone has quickly seized on the democratisation of these platforms. Some have even gone so far as to create entire works based on prompts. Such is the case with Thierry Murat’s ‘initial_A.’, a self-published comic that tells the story of a lonely young girl in the future, on a planet similar to our own.3
Thierry Murat, ‘initial_A.’, produced with Midjourney.
Yet, as the artist points out, this work is still a work of research and definite creativity: several weeks of prompting, collecting, sorting and archiving images, followed by 5 months of developing the pages to bring his project to life.4 In parallel on the other side of the globe, the first Japanese manga generated entirely via Midjourney. ‘Cyberpunk: Peach John , was created in 6 weeks, where an established mangaka would have needed a whole year.Even more surprising, the author even admits ‘to having zero talent for drawing’.5
Cyperpunk : peach john , Rootport, 2022
This message spreads a certain unease in the profession: can you replace creators with a few words? Thierry Murat writes on the subject: “ I haven’t done anything illegal. I have not endangered the publishing industry. I don’t bear responsibility for the invention of this machine. I’m just a free artist, allowing myself to take a look at the world, by adding this tool to my creative toolkit.”4 With ambitious projects being completed in record time, this example hints at changes to come that are radically transforming the way we create, so much so that the manga’s author, Rootport, says in an interview that he can’t imagine a future where artists continue to work in the traditional way.
Control vs. random: mastering the unexpected
Paradoxically, the interest and problem of generative AI lies in the production of unforeseen and random results. A result that can be quite bluffing at first glance and a huge time-saver (getting an image in a precise style in a few clicks, for interfaces like Dall-e or Midjourney) but which proves frustrating when, despite a fleshed-out prompt, over the course of several attempts we fail to get the desired result, and don’t have the upper hand to adjust the design exactly as we’d like.6
The @ai.business.expert account on Tiktok, trying to remove the chopsticks from the image for several minutes to no avail
While some of the AI’s shortcomings have questioned -even mocked– the tool’s real impact through this random aspect7 others have seen it as a creative opportunity to be exploited to take advantage of its media coast.8.
At present, these bugs have been ironed out, and some highly polished campaigns9 hint at infinite potential.
So how can graphic design professionals reconcile this need for control, to precisely meet their clients’ creative needs, with the exploration of new possibilities offered by generative AI? Since May 2023, traditional tools such as the Adobe suite (Photoshop, Illustrator, etc.) have enabled image professionals to generate prompts within their applications (Firefly AI), and thus retain the possibility of intervening in this creation.
If the unpredictability of the tool is no longer a problem for aesthetics and realism, and human intervention is becoming increasingly refined on these tools, what about a visual identity or branding? In this particular context, the use of AI raises the question of consistency and uniqueness. Is it possible to maintain a strong, recognisable visual identity while exploring the possibilities offered by these technologies?
There is a risk of standardisation: these uses could lead to a standardisation of graphic styles and trends. Firstly, AI models are often trained on predominant data sets, which influences the styles they produce. In addition, biases in the training data can reinforce specific trends. The widespread adoption of the same AI technologies and the influence of platforms and tools can also contribute to a homogenisation of creations. Finally, easy access to pre-trained models may encourage the use of predominant styles, to the detriment of creative diversity. In this way, AI could potentially create a funnel effect and limit the variety of styles and graphic approaches.
In this test using Looka, an online tool for automatically generated logos, we realise that there is no concept and that the proposed graphics are rather poor / low quality.
The ease of creation reinforces this phenomenon of standardisation via a more insidious bias of technological dependency: creative people could potentially become too dependent on AI-based tools, which could ultimately compromise their creativity and expertise. ‘If a tool in a few minutes can do a job that would have taken me hundreds of hours of learning and experimentation, why invest my energy in such a process? This bias gives rise to a system that ‘nips creativity in the bud’, because it feeds on research and fruitless attempts, until it masters ways of thinking outside the box, and makes obvious gestures and skills that have taken time and determination.
Production machine or inventor of genius?
This controversy is on the lips of many creators: where does the creativity and originality of AIs come from? Can works generated by algorithms be considered authentic, and to what extent are humans still the creators?
The central debate revolves around the concerns of creative people who devote their lives to developing styles and skills, and who feel robbed of their skills and rewards by AIs. Indeed, the algorithms, through a system of inputs, integrate a set of references supplied by the programmer to the machine, enabling it to reinterpret all this information in a new form. The question of the intellectual property of works generated by AI is highly complex and still gives rise to lively debate.10
On the other hand, the legitimacy of the creator-promoter is also being called into question: How autonomous is the AI in its creative process? Can a well-chosen click legitimise the status of artist?
Defenders of AI point out that the prompts themselves are the result of intentional creative reflection guided by the human and that the machine is merely a powerful execution tool. The human controls the parameters and directives that influence the creation of the final work (be it the tool’s developer or its users) and there are an unlimited number of possibilities (by going beyond the known parameters of framing, light, style, subject and context, we can push the query towards very high precision). In this sense, the work can be seen as the product of a human artistic intention first and foremost, through a phase of learning and mastering this new language. This means that creative artists and designers need to acquire new skills to work effectively with AI technologies.
This new way of creating calls into question traditional notions of creativity and originality, but it also raises the question of adaptation: isn’t every new technological advance called into question at its inception before it is massively adopted and regulated? We can see this with ‘digital art’, which took a long time to establish its credentials, but is now being promoted by renowned artists such as David Hockney11, recognised as a great master of oil painting. He uses the iPad to propose works exhibited in leading museum institutions and helps to give these technological advances their full legitimacy. The challenges of the intersection between technology and graphic design are many, but our ability to imagine remains the key to perpetual renewal.
L’arrivée du printemps de David Hockney à Woldgate, 2011
By Alwine Morel, Artistic Director at Antidox
Sources :
1.https://fr.wikipedia.org/wiki/Midjourney#:~:text=4.2%20Liens%20externes,Historique,l’am%C3%A9lioration%20de%20ses%20algorithmes.
2. https://www.rts.ch/info/sciences-tech/10414540-notre-capacite-dattention-diminue-fortement-a-cause-des-smartphones.html
3.https://www.leparisien.fr/culture-loisirs/livres/je-nai-rien-fait-dillegal-une-premiere-bd-dessinee-entierement-par-intelligence-artificielle-autoeditee-03-10-2023-EJR7YQTAXNCKXPVDFQBVJJHVLE.php
En France, Jiri Benovsky (« Mathis et la Forêt des possibles ») ou Mehdi Touzani (« Le Voyage à Ravine »), et aux États-Unis Kris Kashtanova (« Zarya of the Dawn ») ou Steve Coulson (« The Bestiary Chronicles ») ont déjà eu recours à cette technique.
4. https://www.actuabd.com/Initial_A-la-mesaventure-pas-tres-virtuelle-de-Thierry-Murat-INTERVIEW
6. https://www.tiktok.com/@ai.business.expert/video/7319560051704843552
7 . https://twitter.com/PR0GRAMMERHUM0R/status/1637982384774696960
8. https://jai-un-pote-dans-la.com/burger-king-devoile-une-epouvantable-campagne-generee-par-intelligence-artificielle/
9. https://lareclame.fr/blitzworks-cocacola-masterpiece-277193
10. https://www.culture.gouv.fr/content/download/263325/file/Annexe%204-%20DGFLA.pdf
11. https://www.club-innovation-culture.fr/david-hockney-expose-pour-la-premiere-fois-ses-oeuvres-crees-sur-ipad/
