- Words
- Lucy Bourton
- Identity
- Suzy Chan
- —
- Date
- 15 November 2023
- Tags
Will AI mean that creative directors get stuck in an “echo chamber” of design?
Collaborating with machine learning might speed up parts of the design process, but this could change the responsibilities of current leadership.
Share
Share
In our final article of Shades of Intelligence, It’s Nice That’s new series investigating artificial intelligence and the creative industry, we look at the ways machine learning might impact the production phase of a project and influence decision makers.
Whether you’re curious or cautious about machine learning’s effect on your practice, learn where it can be implemented – and the positives and negatives – via creative case studies and use cases shared by those in the know. This article looks at the final stages of a project: creative direction, campaign roll out and communicating AI’s involvement. To look at the earlier stages of a project, read all Shades of Intelligence articles here.
In the lifecycle of a creative project, once the concept is distilled and agreed upon, there is always that hope that the remaining elements will handily fall into place. Yet the final stages of any project are often the hardest. In reality, concepts might not live up to the vision you’d imagined, decisions are needed from dozens of creative stakeholders, and then there are countless assets to create and put out into the public domain.
You’d think that automation and the speedy execution of machine learning would be particularly helpful here. But the reality, according to those in positions of leadership in the design world, is often quite different.
Step One: Creative Direction
Throughout our investigation into the ways in which AI might alter creative practices, one thing that has become increasingly clear is that the role and responsibilities of the creative director could change dramatically.
If a team decides to use machine learning tools in, for example, the idea generation stage of a project, the weight on the creative director’s shoulders may increase if there are even more avenues to choose from. Their references will need to be more vast than ever, and their inputs more impactful than the countless other creatives using these tools for the same reasons. It may also become their responsibility to ensure learning opportunities are facilitated, if the use of AI tools becomes widely adopted and expected from clients embracing such technologies.
As the founder and creative director of London-based agency Templo, Pali Palavathanan currently describes his role as encompassing “a bird’s-eye view of all the projects on the go at any one time, making sure the creative really answers the brief and captures the essence of what our clients are about.” To shape that view, and in turn lead decision making, Pali is constantly in “absorb” mode. This could be the news, human behaviour or, in his words, the broad scope of the “ever-changing environment around us, to ensure our projects feel spiritually on point, politically accurate and culturally nuanced.”
We’re sure this is a similar job description for many creative directors, but such a vast perspective is particularly necessary at an agency such as Templo. This is also why Pali is sceptical towards AI. As a “cause-led branding and communications agency committed to using the positive power of design for social change,” the concerns around the bias built into an AI’s outputs are front of mind for Templo. The possibility of referencing factually inaccurate information provided by an AI could be catastrophic to a project, and the general lack of transparency around where this data is sourced could lead to mistakes.
For this reason, Pali has steered clear of heavily adopting such technologies in his own role. At present, the agency has explored tools like Midjourney and ChatGPT for creative coding, and experimented with Generative Fill’s capabilities in Photoshop. However, there is a wariness towards employing it further, given the context of “working on areas such as human rights, climate change and anti-corruption,” explains Pali. “All of our work is anchored in truth and accuracy. There’s so much misinformation and a lack of transparency. For credibility, we cannot be generating inauthentic imagery or responses.”
Interestingly enough, the only instance in which Templo has employed AI’s data-mining capabilities is in a style that diverges from the conventional approach. “We worked with an AI human rights organisation called Syrian Archive to launch a new platform, MNEMONICS, using AI programming to trawl the internet and preserve photographic and video war evidence being deleted or censored by tech companies,” says Pali. “So, it’s great to see it working for a better, more transparent world.”
Considering this experience, Pali doesn’t envisage AI appealing to directors in his position when it comes to decision making – arguably, their key responsibility. “The critical and key decision-making moments in a project tend to be instinctive, gut feelings that are rooted in lived experiences or awareness of cultural nuances,” he says. “Also, if everyone uses AI in the same way, at the same stage in the creative process, it’s likely we’ll get the same outcome.”
With this in mind, before adopting such tools in a director capacity, Pali calls for regulation: “But who is going to do this? Google? Microsoft? We’ve already seen tech companies manipulating algorithms, shadow banning accounts that do not adhere to the mainstream agenda and collusion with governments.” In the meantime, it appears that Pali’s view is to in fact hold on to creativity as an act a machine can’t replicate. “I think creativity is the very thing that uniquely defines human beings. To have the desire to create from nothing – whether it’s music, beauty or art – that’s able to then resonate with the community,” he adds. “On top of that, I believe in the human spirit and I’d like to think ancestral intelligence – the ability to contemplate life and death and experience altered states of consciousness – will separate us from AI.”
Step Two: Campaign Roll Out
Even if your creative ethics puts you on the same side as Pali, in a team leader position it could be tempting to adopt AI tools for more logistical uses – for example, in the final stages of a project, such as asset creation for a campaign roll-out. For Eddie Opara, the London-born, now New York-based Pentagram partner, using machine learning in this way doesn’t seem worthwhile.
Largely, in Eddie’s view, this is due to what automating such processes might say about design as a craft. For example, he recalls hearing a talk by a representative of Space10 (the recently closed research and design lab funded by Ikea), who stated a belief that AI will be used for production reasons. “We’ll see more art directors going through more options. Where they used to go through three to eight options, now they’ll go through 50. It’s a production slave, as such,” recalls Eddie. “But the fact of the matter is that a designer doesn’t say they’re a ‘production slave’. I think a lot of designers have to get that into their mindset.”
The answer, as Eddie sees it, might rest in becoming more conceptual, less homogenous in the work we create as an industry, “and thus [we] can have a greater understanding of our needs for ourselves, and move away from the status quo,” he says. In the face of AI, “I think that’s one of the important factors that has come home to roost.”
In his experience at Pentagram New York, AI tools are mostly being employed for logistical, administrative reasons, as opposed to making the task of designing that bit faster. In our interview, Eddie has Read.Ai installed to make a summary of our call and create an action plan. He also admits to using AI to generate emails – “It’s got great prompts!” At times it has also been used to mine data. A current project recently called for information related to the different populations of New York boroughs, “and even though that’s all on the internet, the way it was distributed wasn’t correct,” says Eddie. So, his team prompted an AI to break down the populations of the city to devise a statistical method to then use. “It’s helping us in that way, more than with the graphics themselves.”
In fact, the prospect of machine learning becoming embedded in Eddie’s team is unlikely. “I don’t see the need,” he says. “I want to have a credible conversation with my designers. To sit there, shoot the shit, come up with a crazy idea. Maybe an AI is there to double as a counter to what is being produced, because it’s looking at the status quo. It’s looking at everything else, and we don’t want to be everything else.”
Eddie’s point of view is also mirrored in the findings of our AI survey. Respondents were generally less enthusiastic about AI’s capabilities in the outcome phase of a project, with just 6% believing it would be helpful in production stages.
Spurred on by an NBC report detailing how graphic design is an industry likely to be hurt by AI (“Bollocks!”), Eddie references the options of completing your taxes in the States. “There’s a specific day and everyone is scrambling. People go to H&R Block, which is like the Walmart of taxes. You either do it yourself, go to H&R Block, or some equivalent. But if you’re a larger company you go to another account agency who will do it the right way,” he says. “I think AI is a little bit like that, and graphic design is a little bit like that now. You’ve got options like Fiverr, and then the likes of Pentagram, Wolf Ollins… AI is really at the bottom of the barrel right now. Everyone can use it to make graphics, fantastic, but I wouldn’t be that person and I’m not going to lose my flipping job and neither is my team.”
And like Pali, it’s the wider implications of artificial intelligence on a societal level that remains Eddie’s primary concern. “There were initial concerns about social media, nobody listened, and look where we are today,” he says. “You can’t put it back in the box. Even though we’re investigating how designers will be able to use this technology, we also have to be concerned on a larger level. If we’re not, and we just want to play with this more and more, we have to make sure that we don’t lose ourselves.”
Step Three: Communication
As outlined in a previous Shades of Intelligence piece investigating AI’s usefulness in generating imagery, it’s also crucial for creatives – individuals, studios and brands – to be transparent about its use. One studio with a long history of embedding creative tools into its process is London and Hong Kong-based studio Hato, who have maintained an openness about how such tools are made and how viewers might use them, too.
Since 2009, Hato’s production processes have always influenced the direction of the studio’s output. At first, this was through its printing press where its designers gained a reputation for pushing the boundaries of format. Then developing a sister digital studio, Hato expanded the concept phase of design in relation to how a website might look. Now its focus is directed towards apps, and imagining what they can facilitate.
With this in mind, it’s no surprise that Hato has been experimenting with machine learning longer than most. For example, before the pandemic, the studio designed an interactive installation using Pose JS Library, which would track the movement of visitors’ poses to create a typeface. In 2023, it used the tool again, this time in its annual visual direction and identity for New Contemporaries, a showcase of new artists. Whatever the context, “We just see it as another tool to play with and as a means to express creativity,” says creative director and co-founder Kenjiro Kirton.
When releasing such projects, Hato is open about the tools it has used. Through simple steps – for instance, listing the technologies applied alongside the client name and sector on its website – the studio’s work is more transparent. This openness is also facilitated at a team level in the studio. Research and development days are scheduled every other Friday, when the team opens up tools – from a cloth simulator to machine-learning experiments exploring space and light – for the wider industry. “People will just take it, use it for their own creative output, evolve it,” says Kenjiro. “The more we feed into the industry, the more transparent we are, the more it can evolve. On our end, we’re playing with it as coding libraries. There’s already a culture that’s been instilled for decades, in which everyone shares, and open source is natural. We try as much as we can to feed into infinite play.”
Given this, Kenjiro is also relatively optimistic about AI’s potential relationship with creativity. Outlining how “It makes me feel uncomfortable, but I don’t necessarily believe it’s wrong,” we have to keep patterns of previous developments in mind. “In terms of looking at spaces where technology has had a huge jump, of course there is going to be a shift but I think it’s neither for the better nor the worse,” he says. “Think about the telephone. We can’t live without it now. It lives in our pockets. It’s a super tool for us, and we don’t really remember the switchboard people who lost their jobs in its invention. A lot of people sacrificed their jobs for the telephone, the calculator and the computer. Everyone has adapted and we will need to evolve. If anything, we need to be curious and aware.”
That’s not to say Ken is without concern. Hato uses AI as a collaborative tool, and the creative director is suspicious of those adopting it for idea generation to fit a brief. “I don’t think we would ever see it as a replacement,” he says. “At the end of the day, if you do that – and maybe this is contradicting myself – then what value are you bringing to the project?” This tallies with Eddie’s point around the dangers of homogenous creativity. “I also believe that every project you pitch for is in some ways pre-decided,” says Kenjiro. “If you tackle it in a way that’s truest to what you offer, and everyone does the same, it’s just a matter of what fits the dynamic on the other side. We’d always see it as an addition, as opposed to a replacement. When you turn it into a replacement, you get into the dangers of an echo chamber, and output the same thing.”
Looking to the future, Kenjiro believes we need a societal shift in how we view AI. For instance, “In the west, we’ve been brought up with robots being enemies – in films like iRobot where it’s rogue and causing havoc. In Japan, robots are the friendliest things you grew up with,” he says. “You even see this in production lines and in businesses – the robot isn’t seen as something that would replace a job but as a partner, a collaborator.” The rhetoric around AI feels a long way away from this narrative at present. “Our view is automation is going to kill jobs, and you never see a human in these visualisations. From my end, if AI were to grow bigger, and the West is constantly fearing it, that’s an ethical issue. From my perspective, I’d like to see that shift in perception. I find it ethically wrong to portray it as an enemy when we’re also creating it.”
No matter the discipline or the tool being used by interviewees across Shades of Intelligence, a common thread has been the necessity of collaboration. And if you’ve finished the series and still remain unclear about where to start, experimenting with an AI collaborator – in whichever way you feel most comfortable – would be our recommendation.
Perhaps this could be an internal workshop or a personal brief with the goal of challenging your concepts, or indeed any work, against a machine-learning tool. There are also a whole host of individuals working in artificial intelligence with a creative lens. From Holly Herndon and Mat Dryhurst’s investigations into creative authorship in the context of AI to Stephanie Dinkins’ explorations of AI’s biases and shortcomings, or studios turning this emphasis on collaboration into a USP, like Comuzi.
Based in London, Comuzi is part of a growing group of studios that can create work using machine learning, and teach you how to use it yourself. Founded by creative technologist Lex Fefegha, strategy director Akil Benjamin and managing director Richard Fagbolagun, its projects include a Hip Hop Poetry Bot made with Google Arts and Culture, and an interactive digital platform on Black British art for Somerset House. But Comuzi can also guide organisations, as they put it, “to illuminate the vast possibilities of generative AI through a blend of play, exploration and hands-on experimentation.”
In creative director Lex’s view, there are multiple ways in which creativity can grow in the generative-AI era, built from a starting point of “always thinking about the future world we are creating and what technologies we will use to create them,” he says. When referencing this viewpoint, Lex notes an article by Steve Whapsott on “the different ways machine learning can help designers”; from “AI as a utility worker: to replace low-level tasks” or as “a co-pilot”, all the way through to “AI as a pioneer: to identify new ways to imagine and create”.
The varying roles of an AI collaborator have also become clear throughout the making of this series. As Lex puts it: “We’ve got to remind folks that AI isn’t a monolithic entity, it is really a collection of specialised algorithms and models designed for specific tasks.” Sharing this knowledge through workshops and projects is therefore a key step, as Comuzi aims “to foster an informed dialogue and responsible use of AI in the creative industries.”
As with any future-facing technology, AI presents a new set of heady promises for our industry, while at the same time presenting creatives with a host of new challenges. It’s still far from perfect, as we’ve seen, but at its best, it can be a boon for human creativity and can help free us up to be more creative. Or as Lex puts it: “These technologies are tools that can amplify our human potential and allow us to explore uncharted territories of the digital landscape.” And now is the perfect time to grasp it – ”We are still at the beginning.”
Glossary
Midjourney: Midjourney is an AI tool which can create text-to-image generations from inputted prompts. It is currently available via Discord where users will receive four images to a prompt, before choosing which they would like to upscale.
ChatGPT: Free to use, ChatGPT is a large, language model chatbot created by OpenAI. Users can use it to gather information by asking the model questions, or aid writing with its guidance.
Generative Fill: Featured as part of Photoshop, Generative Fill is a tool built to help edit images, either by extending the image at hand or removing content. As part of the Adobe suite, it’s arguably the most widespread visual AI tool currently available.
Read.AI: Read.AI is a tool users can download to appear at digitally held meetings across Zoom, Google Meet and Microsoft Teams. Following the end of a call, the tool will share a transcription of the meeting as well as automated AI summaries.
Pose JS Library: Pose is a machine learning model which allows for real-time human pose estimation in the browser. It can detect human figures in images and videos using both single-pose and multi-pose algorithms and can therefore be used to build installations, or in augmented reality and animation.
Share Article
About the Author
—
Lucy (she/her) is the senior editor at Insights, a research-driven department with It's Nice That. Get in contact with her for potential Insights collaborations or to discuss Insights' fortnightly column, POV. Lucy has been a part of the team at It's Nice That since 2016, first joining as a staff writer after graduating from Chelsea College of Art with a degree in Graphic Design Communication.