A global premier as a Belgium based A.I. predicts the winner of IAB Mixx Awards 2017

In a remarkable experiment by the MIXX Awards in Belgium, Artificial Intelligence proves to be able to judge creativity and has the potential to replace human juries.

For the first time in history, an Artificial Intelligence algorithm was responsible for picking the winner of an award. Named Pearl, this AI took three months of studying creativity to find which properties of campaigns were important in winning prizes. Having watched and scrutinized more than 30.000 campaigns, Pearl is definitely the most experienced jury member in the world, as she chose successfully the same winner as the human jury.

Out of 101 campaigns, Pearl chose “Her street view” by Touche Pas A Ma Pote, a campaign that draws attention to catcalling. “Pearl made her decision based on scoring the campaigns on multiple criteria. What stood out most is the very strong presence of emotions and the wide reach of the campaign”, her makers said. What began as an experiment from the organizers took a surprising turn by proving that judging an award show is yet another thing computer algorithms can take over.

Gold for “Her Street View”

The winning case “Her street view” by Touche Pas A Ma Pote and the Brussels marketing agency JWT is a campaign to draw attention to catcalling. The website lets you walk through the streets of Brussels using Google Street View. While walking, random slurs and hurtful comments are displayed prominently on the screen, much like unwanted comments women receive in real life.

Blackbox

Pearl is a computer algorithm that supposedly has no soul, sense of creativity or way of interpreting context, so how did it end up picking the exact same winner as the human jury? “That is hard to say”, says Jos Polfliet from Faction XYZ, the company from Antwerp that is specialized in deep learning and machine learning.

Pearl taught herself how to judge cases, we just fed it with a large amount of data. Pearl analyzed about 10 years of submission, calculating things like sentimentality, content, format, soundtrack, … all that data is fed into a black box algorithm along with how well these campaigns did. We can’t fully understand the reasoning, only that it works - spectacularly.

Jos Polfliet, Head of Machine Learning, Faction XYZ

Just a prediction was not enough, so the development team behind Pearl tried to reverse-engineer the predictions: “Turns out three things stood out in the winning campaign. There is a large amount of positive, but also negative sentiment. By analyzing the case description, it picked up on the fact that this was an issue with high societal impact. Finally, Pearl analyzed the real world reach of this campaign, which was quite successful.”

Is the end of human judges near?

Pearl assigned her awards 5 days before the human judges got together. And therefore replaced the human judges all together.

Machine learning is not about replacing humans, but empowering them. Humans still decide what they like and what not. AI cannot decide that for you.

Joeri Van Steen, Co-Founder, Faction XYZ

AI algorithms like Pearl can, however, support the decision making process, for example by checking if a good campaign didn’t fall through the cracks because judges were distracted, tired or biased. Pearl is extremely fast and does not get tired. She is also perfectly objective and not influenced by personal relationships or prejudice.

About Faction XYZ

Faction XYZ is a European A.I. technology provider that enables organizations with a vision and budget to build competitive cognitive capabilities for a future where data science and machine learning fulfill a pivotal role in the rapidly transforming economic reality of our clients. We build machine and deep learning algorithms in the space of computer vision, natural language and sensor data for the telco, automotive, retail and financial industry.