- Painting by numbers: Why AI can do creative too - 5. December 2018
- Customer Experience First - 18. July 2018
- 5 tenets of brand viability - 12. March 2018
- Strengthen your future viability with a platform strategy - 20. December 2017
- Artificial Intelligence: Who is responsible? - 26. October 2017
- Digital Transformation: Bye Bye Buzzword Bingo! - 1. June 2017
- Digital transformation – In the end it’s all down to the individual - 24. February 2017
There’s no longer any doubt that artificial intelligence (AI) can be creative. The question now is exactly what role AI is capable of playing in the creative process. Will this role be limited to that of any other tool, like a paintbrush or a camera? Or could AI become the muse, or even the independent originator of new creations? Could it even be responsible for the extinction of artistic directors as a species? If so, when?
For the time being at least, I can reassure my colleagues that their jobs are safe. Nonetheless, it might be wise for them to start getting on the right side of this new co-worker. Even though the beginnings of AI development go all the way back to the 1950s, it’s only today that exponential development of the three “ABC factors” is enabling it to really gather pace (for the uninitiated: A is for algorithms, B is for big data, and C is for computer chips). That’s why the time has now come for every sector and every company to ask itself how artificial intelligence should be transposed and integrated into its everyday activities.
Within marketing, applications for AI have so far been concentrated primarily in the areas of predictive analytics (for example, for providing recommendations in online shops), personalisation (for example, for individually-tailored newsletters), linguistic assistance, and automation (for example, in media planning). Another important area of marketing, which has so far been almost entirely ignored, is creativity. This is often entrusted only to human hands, and portrayed as an unassailable fortress. With sophisticated puns, poetry, sentimental melodies, magnificent graphics, and everything else that stirs our emotions, there’s surely no way that the processors of a cold machine could ever dream up creative content – is there?
Perhaps we shouldn’t be so sure. For there are already numerous examples today of how artificial intelligence can support, expand, or even imitate human creativity – and the numbers keep growing.
AI can write
How many journalists relish the prospect of laboriously scrolling through the same stock market updates, sporting results, and weather forecasts every day? No problem: the responsibility for texts like these that follow a fixed format can now be shouldered by AI – and without the reader being able to tell the difference. Who knows when robo-journalism will lead to the first advertising texts written by machines, or copy-CADs, as they’ve already been dubbed?
AI can speak
Adobe hasn’t only created the world’s most important program for image editing in the form of Photoshop, but has been hard at work on human speech as well: Adobe VoCo is Photoshop for voice files. After only 20 minutes of listening to a person talk, the program’s AI is capable of fully imitating their voice. VoCo doesn’t simply stitch together snippets of words already spoken by its human subject either, but is instead capable of pronouncing entirely new words as they are typed in.
KI can compose
A team from the University of Toronto has succeeded in programming AI to be able to compose and write catchy and memorable songs. The program, Neural Karaoke, was fed on more than 100 hours of music, based on which it produced an entire Christmas song complete with lyrics and cover graphics.
KI can construct images and grafics
So-called generative adversarial networks are capable of producing astonishingly realistic images from descriptions written by people. Simply put, they function by using a “generator” to randomly create pictures which are then evaluated by a “discriminator” that has learned to recognise objects with the help of real images. This process can turn the words “a small bird with a short, pointy, orange beak” into a photorealistic image.
KI can paint
AI program Vincent from product design specialists Cambridge Consultants, which is also based on generative adversarial networks, has extensively studied the style of the most important painters of the 19th and 20th centuries, and can now make any sketch drawn on a tablet resemble the work of a specific Renaissance artist.
KI can do product design
Intelligent CAD system Dreamcatcher from Autodesk can generate thousands of design options for metal, plastic, and other components, all of which provide the same specified functionality. The designs also have an astonishingly organic look which couldn’t be described as “mechanical” or “logical” at all.
KI can produce videos
Working together with MIT’s Computer Science and Artificial Intelligence Laboratory, Canadian company Nvidia has developed a technology that can synthetically produce entire high-resolution video sequences. The videos, which have a 2K resolution, are up to 30 seconds long and can be made to include complete street scenes with cars, houses, trees, and other features.
KI as the Art Director
Advertising agency McCann Japan has already been “employing” AI in the role of Creative Director for some time. AI-CD ß has been fed a diet of award-winning advertising for the last 10 years, and has already produced its own TV ad based on these data.
Big changes begin with small steps
What does all of this mean for us? Although we may still chuckle at the shortcomings of such AI applications today, development is now moving at an exponential rate – and the progress being made is impressive. This is why now is the time to start getting over the prejudices and fears, and to give proper thought to how we will construct creative processes in the future, together with the role that we want artificial intelligence to play in them. Big changes can’t be made at a single stroke, and are instead better implemented in many small steps. Barriers are best removed by being prepared to play around with new technologies in order to test them out and gather experience. True, doing this takes up a certain amount of a company’s time and resources. But those of us who begin with a small project and then slowly feel our way forward have much higher chances of achieving long-term success, and maybe even of helping to shape a new development in the AI world.
This page is available in Deutsch (German)