“Creativity is not a talent. It is a way of operating.”
– John Cleese –
Never has this statement seemed more relevant.
While the discussion about the age of automation and future of work rages on prominently amidst a cloud of confusion and doubt, there are some whose roles defining and creating the future have led them to confront a more abstract question: how will augmentative technology impact the creative process?
The answer – in stark contrast to cries of “a robot stole my job!” – is highly optimistic, and one of the most exciting elements of digital intelligence.
Rather than replace the need for human creativity, such tech looks to amplify it as a creative collaborator; a muse in the machine.
This relationship has given rise to a process called generative design, pairing human intuition with technology’s ability to explore potential in a logical manner. The result emulates biological evolution, only it is humanity that defines what needs this evolution is required to meet.
We’re already seeing simple, yet stunning examples of generative design at play. Researchers Neill Campbell of University College London and Jan Kautz of NVIDIA Research looked to democratise the process of creating fonts – a role performed by highly specialised designers who generally demonstrate high levels of artistry after years of training – in a project entitled Learning a Manifold of Fonts.
By injecting 1000 fonts into a generative manifold they called the Gaussian Project Latent Variable Model (GP-LVM), Campbell and Kautz produced an interactive 2D tool that allows users to explore a font family’s variations with a drag of the mouse.
“This unsupervised learning process requires no input from either an end user or a professional typographer and yet is capable of generating new, high quality fonts,” their research paper attests. It’s all thanks to the machine’s ability to use what already exists to hypothesise what could exist, but doesn’t, ultimately leaving the human to decide what works best.
Here’s another example, this time from Creative.ai co-founder Samim A. Winger’s presentation on generative design. This tool has been created to assist in the stylising of a comic book by applying elements of reference images to the artist’s completed panels.
Instantaneously, the tool can present a range of possibilities to the artist, and automated possibility is what generative design is all about.
The end result goes a little something like this, according to Maurice Conti in his TEDx talk below: you sit down in front of your computer, and tell your artificial intelligence partner to build you a car. It gives you an example. You tell it to change the shape – perhaps to make it more Japanese rather than European in design – and that you want an automatic transmission. You keep going, making minor improvements by voice rather than hand, and seeing your instructions carried out instantaneously on the screen. When you’re done, you have a car that you created not with a mouse and keyboard, but with your mind.
Of course, there’s potential to go further. You might simply tell the system to create the optimal car. In turn, it will design a vehicle that no human could ever make; one free from such artificial restraints as aesthetics, or traditional geometric make up.
The key will be treating augmentative systems as peers rather than tools. We must be willing to admit that machines will be able to design in ways that we can never imagine, and that this is a positive step to utilising all of our creative ability.
For more on generative design, check out Conti’s speech, The Incredible Inventions of Intuitive AI below. He discusses creation as a method through which people can be given exactly what they want (a definition that, I believe, does some disservice to the exploratory element of the creative process), offering some fascinating insight into design on a much larger scale.