Nikolay Ironov: The Graphic Designer Everyone Thought Was Human

AI has a lot of potential in the world of graphic design, but despite its ability to use human-style reasoning to create content, experts have always been aware of its limitations. That is until fairly recently, when a Russian graphic designer named Nikolay Ironov was revealed not to be all it seemed.

AI as a graphic design partner

The capabilities of AI graphic design have evolved in recent years and what was only dreamt of a few years ago is now a reality. Software applications can translate human requirements to make unique logos, machine learning facilitates the creation of visual assets and intuitive search helps in locating the right content.

While all of this is impressive, AI is still seen as an assistant to human designers in producing creative work. After all, the technology has its limitations, with the inability to understand nuances, not being able to produce work without large data sets and no capability to filter biases amongst them.

The Nikolay Ironov experiment

The perception of these limitations has been completely blown apart, however, after Art Lebedev Studio in Russia used an AI system masquerading as a graphic designer for a year. By the time Nikolay Ironov was revealed not as a person but a machine, ‘he’ had already completed 20 commercial projects, ranging from startup logos to beer bottle labels.

Unveiling the true identity of the designer was shocking, but what was more surprising was the inability to tell the difference from work carried out by a human. The clients and end brand consumers were all pleased with the work when it was completed and continued to have no issues with it when they discovered the truth.

It was a conscious decision by Art Lebedev Studio to end the pretense and not as a result of anyone becoming suspicious of the fictional employee. Since then the studio has declared that synthetic design can be used and loved by a client without limitations.

Harnessing the power of AI to create Ironov

In order for Art Lebedev Studio’s audacious plan to work, it needed to ensure that Ironov could complete the tasks successfully at every stage of the design process. This meant giving the system the capabilities to understand the client’s brief and context, as well as delivering on creativity and exporting the files to be used by the client.

The process began with training Ironov on a custom dataset of Scalable Vector Graphics (SVG) icons. They covered a number of different themes, giving the AI system a strong foundation to build upon. This allowed Ironov to then analyze a company based on input text, taking its name and a description of its operations into account.

Primary images based on this analysis then went through a series of modifications dictated by carefully prepared algorithms. Not only did they simplify the design, smooth its hard edges and scale it, they also generated multiple versions with differing color schemes and fonts to offer a number of options for the client to choose from.

As well as being able to work longer hours than human graphic designers and being immune to sickness and creative blocks, Ironov can surpass its counterparts in other ways too. It’s able to push boundaries more successfully because it hasn’t been educated to fear being bold will lead to rejection.

Human versus AI graphic designers

Even before Ironov, the idea of pitting human graphic designers against AI was intriguing and creative planner Shun Matsuzaka of McCann Japan decided to stage a competition and put the two processes to the test. He tasked creative director Mitsuru Kuramoto to make a commercial for Clorets that would be judged against an AI-created commercial for the same brand.

Based on a database of award-winning ads from the preceding decade, which had been marked up with tags to identify their elements, the two approaches were put to the test. Unlike Ironov, the AI-enabled system wasn’t autonomous end-to-end, as the technology hadn’t advanced enough yet, but it created the core idea that a human team edited into a final clip.

The two completed ads were then broadcast to the public and a nationwide poll undertaken to decide the favorite. The human-created commercial won 54% of the popular vote, but it swung the other way when presented to 200 ad executives at the Incorporated Society of British Advertisers conference, who preferred the AI commercial.