AI created faces: what does this mean for the creative industry?

Thu, 03/01/2019 - 10:50

Artificial intelligence, or AI as it is commonly known, is nothing new. We use it on a regular basis simply by looking at our cell phones and apps such as Facebook and Instagram (and getting those annoying “Recommended for you” notifications from each one). It is no longer the wild ideas of science-fiction authors but a part of our reality. And it is growing in strength and ability almost as regularly as we use it. 

One recent example of this is Nvidia’s AI generated faces. It has been a headline in the world of both tech and everyday news, and for good reason. Nvidia has been using artificial intelligence, machine learning and algorithms to create extremely realistic faces. It has been a four-year-long journey, but the amount of progress that has been made in these four short years is impressive. But, what does this mean for the creative industry?

A bit of a backstory

In 2014, Nvidia began work on creating life-like human faces using AI. And, while the results then were crude, it was a unique and exciting project for the tech industry. The tool they used is known as GAN, or generative adversarial network. Nvidia has modified and improved the model in order to create their incredibly realistic faces. 

These generated “faces” can also be customised, meaning that you can create a specific face if necessary. The method that Nvidia’s programmers used is known as “style transfer”, which you might have heard of from photo manipulation apps such as Prisma or VSCO. While their version did not make the faces look like a Cubist masterpiece, it did allow them to personalise the faces significantly. 

Their source material was a photograph of two to three real people, whose facial features and structure were then blended together to create an entirely new “person” or face. The faces are so realistic that the untrained eye wouldn’t know the difference. And this can have some troubling implications for the world. 

Not only could it be used for misleading information from brands as well as eroding the trust of the public in the media, but photographers and designers might be without jobs if the AI keeps improving. However, this is all future projection. But, what do these AI-generated faces mean for the creative industry right now?

The technology is not perfect

Many people are wary of losing their work to artificial intelligence. There might be good reasoning for this, but for the moment, Nvidia’s face-generating tool is nowhere near perfect. There are ways to easily tell when an image has been created by an algorithm or using machine learning, because AI does not quite understand the symmetry of human faces. 

One of the most difficult aspects of a face to replicate is the hair, and this can easily be seen by the naked eye. An AI-generated face will have hair that looks “painted on”, or that is blended in with the rest of the face. Ears, eyes and noses are often placed in different levels to one another. And this is all good news for stock photographers, as they can provide the real thing to advertising companies who need stock images of people smiling while using their products. 

It can help designers get started on a project

As we all know, you cannot create something from nothing, which is where AI tools can come in handy. By using machine learning, they can detect patterns in data where humans cannot, which designers and creative teams can use to create better image designs for advertisements and branding. 

Ideas usually come from creative impulses that are external, which means that designers can look at the machine learning data for their ideas. For example, they can create an image or logo for a brand using certain colours which consumers react positively to based on information from their AI tool. Or, an AI-generated face can be touched up and made more natural to be used as part of a campaign for a makeup company. 

Originality is still the job of humans

AI and humans have a mostly collaborative relationship. And this is because originality is still the job of humans. Sure, Nvidia’s fake faces look extremely realistic and could easily be mistaken for real people, but they all have similarities to one another and problems that are due to machines not understanding the uniqueness of the human face. 

While AI might be able to create near-flawless images, these are based on patterns and can often look all too similar to one another. Human designers, photographers and creative teams inject their own originality into the creation process, meaning that everything they produce is unique and one-of-a-kind. Many designers do collaborate with tools to produce high-quality content but the ideas are their own and the tools are used to enhance it. 

All around me are familiar faces…

Nvidia’s new AI-generated images might be new and exciting, but they are a long way off from being perfect or even properly “human” looking. But, we should not dismiss them too quickly. Creative industries can use this tool to enhance the branding they already have or the images they use for their advertising campaigns. 

The technology is nowhere near perfect, but it can still be used to help designers come up with ideas and to create designs. So, the machines are not rising yet but it is a good idea to keep an eye on their progress. Creative web design is just one of the services we offer. Contact us today to find out more.

Need Assistance with Digital Strategy?

Rogerwilco’s team of strategists, business analysts and data scientists is here to help.