Creative Machine

Publish date:

Unleashing a new wave of man-machine creativity by letting AI do the heavy lifting of producing it

What if AI writes your haikus? It seemed the final frontier; where technology would automate our repetitive, mind-numbing tasks, we would find our new forte as humans in creativity – an area where AI could never match us. Turns out that Generative Adversarial Networks (GANs) – in which AI systems collaborate in creating and testing results – can create spectacular results in areas as diverse as images, video, audio, text, art, products, medicines, games, and even entire business models. When done well, AI and humans can together unleash a new era of great creativity. But the boundaries of what is real and what is fake are stretched, and it takes more than a poetic mind to deal with that.


  • The basis of GANs is synthetic data – data generated by machine algorithms. Yet, this data is becoming as real as real data.
  • The trick lies with creators and discriminators. Part of the system generates the output, the other part tests it on how credible it is compared to the ‘real’ counterpart.
  • Many different GANs generate many different things. StyleGANscan generate certain ‘styles’ of human faces, creating an image of what you’ll look like in 20 years.
  • GAN technology is used to design certain attributes in games such streets, cars, houses and even people. It has a multitude of applications, from design of software to interiors of houses and fashion, even creation of music, books, paintings and art.
  • Over the last 12 months alone, there has been an explosion in the number of scientific papers on GAN technology, we are only scratching the surface on the technology’s future capability and function.
  • It can be used to design computer generated humans to be used in film and screen. Yet it has a darker side in the creation of ‘deep fakes’.


  • To test autonomous driving cars, synthesized cities are created – a digital representation of a non-existing city.
  • In the health industry, generated synthesized data improves the data quality of health screening such as MRIs and brain scans.
  • Swedish Mackmyra uses an AI system, coached and augmented by human experts, to created optimized recipes for whiskey.
  • The outdoor field game, Speedgateis entirely imagined and created by an AI system, verified by humans, created by analyzing hundreds of different existing games.
  • Researchers at University College London used human input with AI to recreate one of Picasso’s most famous (and currently lost) paintings from the Blue Period, The Old Guitarist. After the project, the researchers explained that the human-AI collaboration not only broadened their insight into an artists’ creative process, but gave an understanding as to the creative potential of AI as an artistic tool.
  • NVIDIA researchers in collaboration with University of California, Merced developed a deep learning-based, generative model that can automatically create a dance video with new, diverse and style-consistent dance moves that match the beat.


  • Automation of human creativity represents huge possible cost reductions.
  • A whole new human to machine interaction, where creative human processes have input from creative machines.
  • GAN technology enables an enhanced speed of working, including activities such as the invention of new medicines.
  • The use of synthesized data in testing leads to a substantiated improvement in quality of products and processes.


Featured Expert

Patrice Duboe

Expert in Digital, Innovation, IT Transformation, IOT