Standing Out from the Crowd: The AI Edition

Written by Lucie Robathan on June 29, 2017

Creativity and innovation are two of the most powerful buzzwords in contemporary culture. We push ourselves on a personal level to think differently and to imagine new landscapes in which to work, to invent, and to produce. We also strive at a company level to be new, to be singular, and to stand out – and in terms of our organizations, we want our marketing resources and brand messaging to be original and to demonstrate this individuality.

So what should we think of the creative and inventive potential of technology, or of the possibility of algorithmically predicting – and perhaps even constructing – creative success? Jodie Archer and Matthew L. Jockers’ “The Bestseller Code: Anatomy of a Blockbuster Novel” takes a fascinating scientific approach to analyzing bestselling fiction. The authors developed a text-mining algorithm, and collected data over a five year study of 20,000 novels, to present a new, algorithmically supported code for predicting successful creative writing. Every bestseller, they argue, displays similar patterns; from topics, to title structure, to the number of times specific key words are used throughout the book. Even the distinctive, individual style of popular authors can be treated like a “linguistic fingerprint” to be analyzed.

If it is possible to use an algorithm to unpack the formula for creativity and entertainment in this way, does this point to the potential of technologically creating entertainment itself? Last year, a Japanese AI program co-wrote a short novel that came close to winning a national literary prize, and although it didn’t succeed in beating its human counterparts, it certainly reveals that much of the creative process can be templated. In fact, another computer, IBM Watson, created the first AI-made trailer last year for 20th Century Fox – and did so in a fraction of the time it would take a human team.

Should this make us disheartened? Does it signal the demise of human creative pursuits, of personal expression, and of our attempt to make our individual stamp on the world? Or should we listen to Sony CSL, whose Flow Machines research project looks to build computational tools both for writing texts and for composing music, and who present their work as a way to boost creativity and to offer new ways for experimentation and stylistic development? Flow Machines’ findings have produced the first AI-composed pop song, Daddy’s Car, of which I’ll let you be the judge:

Sony CSL’s research also extends to the possibility of using such technology to enhance children’s learning. They also participated in the MIROR project, developing technology enhanced music systems, for music and dance composition and improvisation, as pedagogical tools to help people access new modes of creativity. The idea that AI can serve as a teaching tool, and actually support our own imaginative pursuits, is also shared by Archer and Jockers: they suggest that their algorithm can be useful for new, aspiring authors more effectively and confidently to find their own style, and to discover the best path towards popular success (if that is what they are looking for).

Analytics already have so much transformative potential for our organizations in terms of information-collecting and decision-making: will machines also be able to help us to be creative, to entertain, and to refine our style? Might it one day help us to expand our imaginative horizons, and to make sure that creatively speaking our marketing, our publications, and our media presence are as effective as possible? Will it be AI that shows us what there really is outside the box?