Cannes Lions

I am Hunger in America

LEO BURNETT CHICAGO, Chicago / FEEDING AMERICA / 2020

Presentation Image
Case Film

Overview

Entries

Credits

Overview

Background

PROBLEM

?1 in 8 people struggle with hunger in the United States.

But most Americans have no idea that the problem is so extensive and so close to home.

The reason?

Most of those struggling with it don’t fit the usual stereotypes of hunger:

64% own a home, 50% of them are Caucasian, and 66% are employed.

?

To show that hunger can look like anyone, Feeding America, a non-profit fighting to end the issue, used Data and Artificial Intelligence to create an unbiased representation of it.

A single face that represents 40 million hungry Americans, and finally makes them visible.

Idea

1 in 8 people struggle with hunger in the United States.

But most Americans have no idea that the problem is so extensive and so close to home. They think of hunger using their bias and stereotypes that people are poor and unemployed.

To show that hunger can look like anyone, Feeding America a non-profit fighting to end the issue, used Data and Artificial Intelligence to create an unbiased representation of it.

A single face that represents 40 million Americans, and finally makes them visible.

We brought the face to life through a film that told real stories of people affected by hunger.

We developed a new compositing technique that allowed us –without using CGI– to make an AI-generated look human, by blending it with real actor's faces.

Strategy

Feeding America gave us tens of thousands of photos of their actual "clients" who depend on food banks. We then weighted the faces to align with detailed demographic data on hunger provided by the U.S. Department of Agriculture.

Then, we used a generative adversarial network (GAN), a form of machine learning, to study attributes of 28,000 of these real faces.

After over 1,400 hours of machine learning, the GAN was able to generate a face that didn’t exist before. The true Face of Hunger in America.

Execution

Feeding America gets its message out primarily through mass media donated by TV stations, magazines, radio stations, etc. They compete with other issues--racism, drug abuse, cancer, etc. for this free media. They needed a mass campaign that would get picked to run for free...and stand out when it did.

To challenge Americans’ idea of what hunger looks like in the U.S., we used AI to help create an unbiased portrait representing the 40 million Americans who go hungry every day. Our process used a Generative Adversarial Network (GAN) to use machine learning with a data set of 28,000 photos of people who visit Feeding America food banks. To be even more accurate, we weighted the photo dataset using the most recent USDA demographic data of hunger in America. With over 1,400 hours of machine learning, the GAN learned what hunger looked like, then created a new face that was a true representation of all 40 million Americans who struggle with it.

Compositing our newly created, unbiased face onto actors allowed us to tell actual stories from the people it represented. More stories were told in radio, which featured the "voice" of hunger that morphed into others. In print, we used the face itself, telling the stories in type. A website allowed people to dig deeper into the data behind hunger.

The campaign was picked up by multiple media companies leading to approximately 80,000 airings which is an equivalent donated media value of $50MM since launch.

Outcome

133MM Earned Media Impressions in the first week.

115MM Social Media Impressions.

141 News Outlets

Similar Campaigns

1 items

I Got You

COLLE McVOY, Minneapolis

I Got You

2021, FEEDING AMERICA

(opens in a new tab)