Artificial Creativity & Broad Data
In 2014, my friends at Admap asked me to write a piece about the impact of Big Data on creativity, to promote their annual essay contest on the same. I looked at Big Data and couldn’t find anything that anyone had done with it except retarget banners more invasively, in a way that I felt would drive greater rejection of digital advertising (which it did, as evinced by the meteoric rise of ad blocking over the past two years).
I went further, musing that since creativity is a combinatorial act -the process of combining inspiration in non-obvious ways to solve problems -then some day the creative department of an agency might look like IBM’s Watson.
Two years is an age in exponential times, where every step doubles. In March, McCann Japan announced it had ‘hired’ an artificial intelligence called AI-CD beta as creative director. The algorithm was created using ‘tagged and analyzed TV shows’ and data on the winners of the local advertising awards, which empowers it to give creative direction on spots. OK, so this is primarily a PR stunt, but the emergence of AI as a creative force is hard to ignore in 2016, already being hailed as the ‘year of the bot’.
In 1997, IBM’s Deep Blue beat Garry Kasparov at chess and the world looked on and wondered about our new robot overlords. Indeed, it was this spectacle that Watson’s appearance on Jeopardy! was specifically designed to supersede. Deep Blue is archaic by today’s standards; it played with ‘brute force’. It considered 200 million chess positions per second -twice as fast as the 1996 version that lost to Kasparov. (See? Exponential steps are double.)
This brute force attack works because computers can ‘think’ billions of times faster than humans and chess has a relatively small set of possible moves at any moment.
This year, Google’s AlphaGo beat the world’s best Go player with a different approach because brute-forcing Go isn’t possible, even with today’s enhanced computing power. There are too many possible moves. AlphaGo learns and makes experimental leaps.
What, I hear you asking, does this have to do with writing and designing ads? We aren’t playing games here. History is a guide, a map to how things progress along vectors, and how quickly. Deep Blue technology is so simple that today it can easily be deployed in bots, and the first advertising algorithms have already begun to appear.
Imagine bots as specialised copywriters. You brief them with a brand and a set of references, and they begin to write down words that might be relevant and rearrange them, looking for interestingness, probably in three-word exhortations with slightly broken grammar, as is so often the style. To brief the bot, you upload a word bank or tell it to absorb terms related to the topic by scanning Wikipedia. It then generates millions of taglines.
The system could then check for duplicates from history or across geographies and proceed to buy search-and-display ads, serve statistically significant amounts of inventory for each line, see how they perform, and optimise using evolutionary algorithms until you have a perfect slogan, tested for efficacy in the real world.
Take @taglin3r, add in some APIs, stir in some news feed and a handful of genetic algorithms and voilà, brute force creative! Now imagine what GoogleBooks’ AlphaAds will be able to do a decade from now.
Don’t freak out. AI is all intelligence and it’s within the messy heuristics of emotions that advertising operates. Beyond Big Data, the truly non-obvious connections of Broad Data™ are still the reserve of quirky associative human brains, like those powering the inspiration engine Seenapse.
Generating combinations is a matter of process and volume, ideally suited to software. Understanding whether they could have emotional impact — and how — is not.
The three words ‘Just Do It’ are often hailed as one of the Big Ideas in advertising — but are they? What kind of idea? Aren’t they just three words, derived from the final ones of murderer Gary Gilmore, tweaked for style, that Nike and W+K turned into something that echoes across ages?
A brand statement without actions and advertising to support it is meaningless.
Artificial emotions are not even on the horizon, so understanding how to use creative algorithms and gauging the emotional impact of their outputs will remain the reserve of agencies.
Better start playing with Deep Blue now, though.
THANKS FOR READING! FOR MORE CREATIVE INSPIRATION IN YOUR INBOX INSERT YOUR EMAIL ADDRESS BELOW!
Originally published in ADMAP Magazine