Could a Machine Win a D&AD Pencil?

AI looks to the free-thinking potential of the human mind

Image: Colin Anderson/Getty Images

There’s something inherently human about this award-winning ad by BBDO New York. To think of an artificial neural network (ANN) coming close to this level of sentiment seems far-fetched. But actually, when you dig a bit deeper into the traits of successful ad creatives, is it really that hard to imagine?

What is an idea, anyway?

According to the Oxford English Dictionary, it’s “a thought or suggestion as to a possible course of action” and “an opinion or belief.” Machines are more than capable of making a rational suggestion, as they process billions of inputs in no time at all—IBM’s Watson can scan 40 million documents in 15 seconds. So the question ultimately relates to whether they can form an opinion.

Andy Shield

Descartes noted, “I am certain that I can have no knowledge of what is outside me except by means of the ideas I have within me.”

Aren’t all of our opinions based on a set of subliminal preconceptions? Even though we feel completely unconstrained, we actually see the world through our own set of biases. It’s likely that when BBDO created the Ticket Twosdays ad, the team drew upon their own experiences.

In a similar way, an ANN constructs its opinions based on the information it’s aware of. But unlike ourselves, it has the potential to easily connect the dots without an underlying bias.

Built-in diversity

Google recently came under fire for the lack of diversity in its image search results. If you Google the word “face,” you’ll find only two non-white examples in the top 40 results—one of which is Barack Obama’s. Critics blame the lack of diversity within its own organization—only 19 percent of its tech staff are women and 1 percent are black (January 2016).

But the main culprit is the disproportionate use of white individuals within advertising, marketing and publicity. Google is drawing upon the references it has at its disposal, and if the input were more representative, the results would improve.

For example, the engineers at Lobster developed AI which presents a much broader interpretation of the same search—it even includes a meerkat. So it’s logical to conclude that, given the right content, an ad created by a machine would be gender and race agnostic. Olga Egorsheva, CEO of Lobster, explains, “Lack of diversity is discussed over and over again within small groups of industry experts, but brands are still using tired old stock photos of white, middle-aged men in suits. We gather content from all over the world to create a better reflection of what real life actually looks like.”

Knock, knock, it’s Alexa…

Ask Alexa for its favorite color and it’ll tell you that “infrared is super hot.” Ask Siri for the whereabouts of Elvis and it replies “I believe he has left the building.” This kind of punnery is becoming increasingly popular in personal assistants, but clearly it’s all pre-configured and relies on its creators.

The most promising direction for humor in AI is accidental, claustrophobic and kind of creepy. In June last year, an AI called Benjamin debuted its first screenplay at Sci-Fi London. Sunspring is a bizarre story about three colleagues on a space station who seemed trapped in a deeply troubling love triangle. The dialogue is captivatingly weird and at times hilarious.

The project is the result of a collaboration between AI researcher Ross Goodwin and director Oscar Sharp. Ross fed the AI with a stack of ’80s and ’90s sci-fi scripts, and over time it learned to imitate the format of a screenplay. It included stage directions, such as, “He is standing in the stars and sitting on the floor.”

Creative AI researcher Luba Elliott explains the complexity involved in creating a funny AI. “Being funny is tough,” she says. “It requires a combination of linguistic mastery and human qualities that machines currently struggle to reproduce, such as being spontaneous, understanding cultural references and showing empathy. It will take some time for AIs to master these. Until then, our best bet is to laugh at their attempts at getting there.”

AI creative direction

Last year, McCann tasked an AI to provide some creative direction for a Clorets Mint Tab TV ad. After analyzing a database containing ads from the All Japan Radio & Television Commercial Confederation’s annual CM Festival, it came up with “convey ‘wild’ with a song in an urban tone, leaving an image of refreshment with a feeling of liberation.” From there, it was over to the humans to bring that concept to life. It was the kind of creative director that hands over a page of notes and disappears for a three-hour lunch.

But it’s not impossible to imagine that a machine could take an initial idea and create the various components needed for an entire campaign. Jukedeck does that with music, to a certain extent, although it is quite formulaic.

Could a machine win a D&AD pencil? There’s no doubt it could write a script for an ad similar to Ticket Twosdays. However, under the surface, there’s an ocean of sentiment that a machine can’t fathom. It’s often the nuances of human nature and the intangible aspects of our relationships that provide the inspiration for an award winner.

As the advertising industry becomes more programmatic, to win any kind of creative award, machines need to become unmistakably more human.