Can art be distilled into math? Classical architecture and Renaissance art reveal elegant proportions at the core of their design, with some of world’s finest designers like Leonardo da Vinci deliberately incorporating computational elements such as the golden ratio and the Fibonacci series in their creative process. Harnessing recent advances in AI research, creative startups and established players like Adobe, Autodesk and Google push the mathematics of art further by infusing the design process with varying doses of artificial intelligence.
The rise of “creative” machines
Drawing from trillions of data assets, Adobe Sensei claims to know human faces so well that it can help you instantly turn a frown into a smile, an image-editing feat which normally requires hours of tedium. Google Brain’s Magenta project built on TensorFlow wants to both advance AI research in generative art and build a community of artists and engineers around open-source creative tools. Autodesk’s Dreamcatcher is a next-generation CAD system that accelerates industrial product design by leveraging AI to generate thousands of design permutations that meet project specifications set by engineers and designers.
Smaller but nimbler players co-lead the charge, often risking all on a bold bet that the potential business underlying AI will soon drive real and hefty profits. UIzard Technologies, a Copenhagen-based startup, built a solution called Pix2Code which uses enhanced computer vision, language processing, and deep learning to take screenshots of a graphic user interface (GUI) and convert the images to working code, simplifying the transition from design to engineering and dramatically easing the workload of front-end developers. Functioning across Android, iOS, and web, Pix2Code reportedly performs at 77% accuracy. UIzard’s performance target is to dramatically improve accuracy by sourcing the internet for unlimited training data.
Loosely named after Picasso, Pikazo is an app that aims to turn everyone into adept visual artists … and virtually anything into a work of art. Like their close competitor Prisma, consumers can use the app to turn their boring cat pictures into works in the style of Monet, Picasso, Van Gogh and other famous artists. Both Prisma and Pikazo leverage neural style transfer, a deep learning approach which is vastly more complex than regular mathematical image filters.
Like visual art, literature is another field currently being enhanced by AI. A Japanese AI system co-wrote a novel that nearly won a literary award, while bot journalists powered by Heliograf’s sophisticated storytelling AI write many well-recieved articles on The Washington Post. Uses of NLP have gone beyond functions like speech recognition and machine translation to make promising footholds in creative and editorial writing.
Narrative Science and Automated Insights, both key players in the NLP space have been expanding their respective markets. Narrative Science recently partnered with Deloitte, the world’s largest professional services network, to deploy Quill, an NLP-driven solution that produces compelling and insightful reportage from raw financial data. Meanwhile, Automated Insights provides a software called Wordsmith to organizations in more than 50 industries to generate data-driven narratives on a massive scale.
These mentions are hardly comprehensive. Many more tools exist to enhance the creative process for designers and artists of all kinds. For those of you interested in exploring algorithm-driven design further, designer Yury Vetrov has compiled many examples of AI-fueled creativity on his website.
Will AI artists displace human creatives?
As the capabilities of Pikazo, other robot artists, and NLP-driven automated authors rapidly mature, the possibility of an AI-dominated art world looms large. AI applications even go beyond creation to the realm of aesthetic judgment and literary criticism.
In 2016, McCann Japan “hired” an AI-powered creative director named AI-CD ß as part of its McCann Millennials project. For its first assignment, AI-CD ß helmed a video ad for Clorets Mint Tab, giving the production team the specific instruction to “convey ‘wild’ with a song in an urban tone, leaving an image of refreshment with a feeling of liberation.”
Meanwhile, Disney — the ultimate paragon of blockbuster narratives — partnered with the University of Massachusetts Boston to build neural networks that can evaluate the narrative quality of short stories. While still far from rivalling the rigorous training and analytical insight of professional literary critics, the AI software is reportedly refining its ability to predict which stories will be the most popular. Given Disney’s incredible record (the company produced nearly half of the 20 top grossing films of all time), their predictive AI is learning from the best human storytellers.
Should human artists feel threatened by the rapid progress of AI?
AI and robotics have historically been presented as solutions to eliminate physically tedious and non-mentally demanding jobs. Recently, AI systems have outperformed lawyers, doctors, financial advisers, and even managers in specific tasks require domain expertise. With AI capable of learning to create other AI, the realm of artistic and creative work – long thought of as being uniquely human – seems to be under threat.
Being creative in a world of designers
A world of AI doesn’t have to be as surreal and unnerving as a Dali painting. Many experts remain upbeat on the positive net effect of AI-driven design. Artefact Group’s Rob Girling believes that among many benefits, AI-driven design will:
- turn everyone into more or less decent designers; and,
- empower design superstars to remain ahead and achieve new breakthroughs
Girling compares the impact of AI to the content revolution ushered in by desktop publishing. While many professionals learned to use new tools such as PowerPoint, the most talented designers remain markedly differentiated even when using the same tool. Girling also predicts that instead of diluting the design occupation, AI will boost demand for designers, especially those willing to reimagine better user experiences and interactions with emerging AI systems. Software developer, inventor and author Patrick Hebron agrees. Hebron posits that human and machine can combine forces to drive superior results, that machine learning will simplify design tools without restricting human creativity.
“Tools are not meant to make our lives easier … They are meant to give us leverage so that we can push harder. Tools lift rocks. People build cathedrals.”