To set up a structure for the campaign, 50 scripts were written based on location, behavioral insights, and occupation data that explained the car’s features. The scripts were then used to train Watson so it could whip up thousands of pieces of copy that sounded as if they were written by humans. To give Watson guidelines yet allow it to have creative freedom to come up with some of the thoughts on its own, the AI scraped the internet, including sites like YouTube and Wikipedia, to create a neural network of words and phrases about what it means to be a scientist and put the thoughts together into sentences.

50 scripts were written based on location, behavioral insights, and occupation data that explained the car’s features. The scripts were then used to train Watson so it could whip up thousands of pieces of copy that sounded as if they were written by humans. The experiment phase was essentially about finding deep, weird insights to mine out and put into sentence structures. Through meticulous training sessions with IBM Watson, the team focused on word meaning and cohesively using words together. The AI responses were then ranked in order to train the technology on whether something was “good” or “bad.”

To pull off the complex targeting and find intriguing behavioral data, the campaign ran solely on Facebook. While each video looked the same, the copy was tailored to more than 100 different demographics, resulting in “thousands” of pieces of content.