Part of the fun of judging the CR Annual (which will be published in our May issue) is in spotting this year’s trends. In the graphic design categories this time around, we saw a huge amount of Memphis-inspired work, alongside lots of geometric sans type and the continuing influence of the work of Design Studio, particularly its colour palettes.
In the face of the great automation debate (which we explore in depth here), creative people have taken refuge in a belief that machines could not possibly replicate what they do. But when you see so much work looking so similar, you do start to wonder.
The ability of digital systems to observe how people are using them and, therefore, what they want from them, has so far been portrayed in positive terms. Data, it is generally accepted, is key to the improvement of products and services: the more we learn, the better we are able to serve users.
Adobe for example, in line with many other software companies, already runs what it calls its Product Improvement Program. Designed “to understand and anticipate customer needs in order to deliver world-class products and solutions”, the scheme is voluntary. Users who sign up to it allow Adobe to collect data on how they use its software: how they carry out various tasks, what shortcuts or scripts they might use, what kind of images they are looking for and how they are using them and so on. Adobe says that “information collected will be used to develop new features and improve Adobe products”. Fair enough. But with machine learning, it becomes possible to take the next logical step in the process – observation becomes replication and then automation. The more such systems learn about how, for example, a designer employs certain steps and routines to get from A to B, the better able they are to do the work themselves.
No-one is saying that Adobe is planning to do that but there are already several prototypes of machine learning-based design systems which can interpret a range of styles or key words to produce collateral. By following the steps that designers and art directors take to make a piece of work, and associating images and typefaces with descriptive tags, these systems can teach themselves how to create, for example, a poster in an approximation of the ‘Memphis’ style. (You might throw your hands up at the idea of a machine regurgitating an intellectual and philosophical movement as empty style – but isn’t that what many graphic designers do anyway? Modernism, Constructivism – ditch the politics, keep the type.)
In 2010, a group of academics proposed “an automated graphic design expert system” to a conference on AI. The authors dispassionately described graphic design as “the process of creating graphics to meet specific commercial needs based on knowledge of layout principles and esthetic concepts. This is usually an iterative trial and error process which requires a lot of time even for expert designers. This expert knowledge can be modelled, represented and used by a computer to perform design activities,” they say. Their system, named Gaudii, promises to deliver design collateral, from brief to final artwork, in five seconds!
Last October (as reported here by Campaign) lingerie brand Cosabella replaced its digital ad agency with an AI platform. After spending some time getting up to speed, Albert (as the AI is known) took on the management of the firm’s social and digital marketing efforts, returning far better results than its human predecessors. Though it was using creative assets supplied by the brand (and presumably created by humans), Albert soon got to the point where it was effectively creating its own ads. “Albert tests different copy with different photos and spends the first couple of weeks optimising,” said Cosabella’s marketing director Courtney Connell in the piece. “Once he’s optimised your campaigns he’ll start to make his own.”
AI systems exist for writing copy, music and even film scripts (with as yet varying degrees of success). And they are constantly learning from everything that we do. As illustrator Michael Gillette says in a comment on our piece by Adrian Shaughnessy on this debate, “every post uploaded to social media, neatly catalogued with hashtags and given freely, expands the ever increasing research database that AI learns from. Whether it be our creativity, esoteric culture, or simply the way our world looks from breakfast to last orders, we have become the most diligent unpaid data gatherers for AI”.
How do we avoid the seemingly inevitable? What these systems are not good at is difference, idiosyncrasy or seemingly illogical departures from the norm. Creativity, in other words. The more designers and art directors spend their time endlessly recycling ‘inspiration’ from the same online sources, the more they become complicit in the possibility of their own extinction. Survival, if that’s not too dramatic a term, depends on thinking for ourselves.
Read all our coverage of AI and automation in this CR Special Report and in our April print issue. Cover illustration: Matt Murphy