How Ringier’s EqualVoice promotes gender equality in news coverage
By Hannah Chia
“We are all biased, and we have to be aware that we are all biased,” said Lea Eberle, Head of Finance Projects at Ringier, during our Asian Media Leaders Summit in Singapore.
“Conscious or unconscious bias, it is just how we are as humans,” she said.
While bias manifests in many forms across journalism, Eberle highlighted gender bias as a persistent blind spot in media coverage.
For example, despite women making up 40% of all athletes worldwide, only 4% of all sports media articles include women.
Although 35% of all executives in Switzerland are women, the country’s largest business magazine, Bilanz, has only rarely featured a woman on its covers in the past 25 years.
WAN-IFRA Members can replay Lea Eberle’s presentation and see the related slides on our Knowledge Hub.
Turns out AI results are often biased, too
She also pointed out that misrepresentation in the media can be amplified by technology.
For example, Ringier collaborated with Microsoft and Switzerland’s IMD Business School to conduct a study on diversity-specific AI bias.
When AI was given “CEO” as a prompt, nearly 100% of the results were white males over 40 years old.
When given “business woman” as a prompt, almost all results showed a white, young lady with an open shirt, leaning against furniture in a sexy pose.
When AI was prompted to show a firefighter saving Arnold Schwarzenegger from a burning car, the results never show Arnold Schwarzenegger as the victim, but rather, the hero. Eberle also points out how some of the generated pictures show Arnold Schwarzenegger saving a female firefighter.
“AI cannot imagine him being a victim … and that’s exactly the problem,” Eberle said, “If you don’t have the data which goes into AI, the output of AI will be biased as well.”
Rather than denying these biases, Eberle encouraged media organisations to focus on recognising such biases to ensure fairer reporting.
Using technology to reduce biases
“It’s so important that we as media companies take our responsibility to society and change the narrative within our publications, but also for the future within AI,” Eberle said.
She pointed to EqualVoice, a project initiated by Ringier in Switzerland in 2019, as a potential tool to help newsrooms confront their own blind spots. The initiative was founded on the belief that “seeing is believing.”
Eberle, who serves as the chief operating officer of the initiative, said the project aims to focus on gender equality and diversity from a media perspective.
The data-based initiative analyses the media’s “EqualVoice-Factor,” a measure of the visibility of women in media by analysing text, images, video and audio.
EqualVoice-Factor is an algorithm developed in-house at Ringier, but also fact-proof and checked by ETH Zurich, a University in Zurich, Switzerland.
It analyses how women are represented in media, including the types of questions directed at women, the images selected to accompany the coverage, and the adjectives used to describe them.
“If we can see women in every role in our everyday lives, then little girls… believe they can become this role and change the narrative about stereotypes as well.” Eberle said.
Moving beyond stereotypes, and making a measurable difference
EqualVoice also employs the same analytical approach used to represent men, moving beyond narrow stereotypes, such as the dominant, strong roles, to include a wider spectrum of identities and characteristics.
Eberle explained how, in Europe, men are not portrayed as fathers, caregivers, or kindergarten teachers in the media. She emphasised the importance of children seeing men in such roles as part of efforts to change the narrative around masculinity, alongside the shift in narrative about women in media, to ultimately change the future of society.
“We call it EqualVoice because it’s equally important to address stereotypes for men,” Eberle added.
When the initiative began, business publications in Switzerland started with an EqualVoice-Factor of 17%, which was below the global average. Now, the figure has risen to over 34%, doubling the number of women shown in business publications in Switzerland.
EqualVoice also serves 32 newsrooms in seven countries, and reaches 50 million users with its tools.
Improving media representation with EqualVoice-Assistant
Building on the Equal Voice framework, the initiative expanded with the development of EqualVoice-Assistant. EqualVoice-Assistant is an AI system embedded within its content management system to flag bias and stereotypes before publication, and suggest an alternative solution instead.
Eberle noted that newsrooms can benefit from EqualVoice-Assistant as a tool to identify biases in articles before they are published.
“We want to educate our journalists to improve … so we don’t reproduce stereotypes anymore”, Eberle said.
Developed in the past year, EqualVoice-Assistant is used in four countries, serving more than 730 users and analysing over 32,000 articles a month in Ringier’s newsrooms.
A collective responsibility for change
Citing the World Economic Forum’s Global Gender Gap Report, Eberle noted that at the current pace, it would take 123 years to close the gender gap. She said Ringier does not intend to wait that long, emphasising the need for the media industry to collectively participate in reducing bias and gender risks being replicated unconsciously.
“Gender equality and diversity change needs time,” Eberle acknowledged, “It’s not a sprint, it’s a marathon.”
Click here to learn more about Ringier’s EqualVoice initiative.
About the author: Hannah Chia is a student at Nanyang Technological University’s Wee Kim Wee School of Communication and Information.
