Karen Lerbech (media and diversity consultant)
Since 2017, DR has had a formulated diversity strategy focusing on content, with a diversity team implementing the strategy on both a strategic and editorial level.
Data has been an important component of the strategy since day one. As Heidi Sivebæk, Editor of Diversity at DR, says: “You can’t make a change if you don’t know your challenges.
Data plays an important role in unfolding the blind spots and unconscious biases of editorial teams across DR, driving the change towards more diverse content.”
Up until 2022, diversity data was gathered solely by the diversity team, including student assistants, through manual tracking. And while it has been a sound method, it has also been very time-consuming. It involves a lot of voluminous analyses, looking across whole genres or channels and generating a huge amount of data. But it’s a race against time to finish the analyses before the data becomes outdated.
Furthermore, the diversity team at DR has realized that gathering big data samples also limits the number of analyses it’s possible to conduct, which again limits the impact a diversity team of a limited size can have in a big organization like DR.
AI-based tracker
To increase the potential of tracking, gathering more data at a faster pace, DR last year decided to put an AI-based diversity tracker into use, Heidi Sivebæk says: “It allows us to use our resources differently and more efficiently.”
The diversity tracker is delivered by MediaCatch, a Danish tech company.
MediaCatch builds on AI technologies that audit, for instance, live channels 24/7, using both facial and speech recognition. Where manual tracking normally counts heads, MediaCatch counts minutes on screen.
According to Lars Damgaard, CEO of MediaCatch, the diversity tracker offers a tool for editorial teams to get a deeper and more nuanced understanding of their content.
“We hope that we can help public broadcasters with insights that make them act on their diversity strategy. When we couple knowledge of gender and ethnicity on screen with who’s doing the talking, how long they have been talking, and what
they actually talked about, we get facts that can help you take actions,” he says.
“Our insights are delivered live so editorial teams can take immediate action, because a dashboard gives them all the data they need immediately after a finished broadcast. In this way, the teams have all the insights to evaluate every segment of a programme and discuss it on a daily or weekly basis.”
Data dashboards
At DR, MediaCatch has delivered both programme- and channel- specific dashboards to ensure the information provided is as relevant as possible for the editorial teams who have access to it. These dashboards, combined with a feature that allows the diversity team to extract data from specific time slots, mean that it’s now possible for them to get an overview of the gender balance on screen in, say, the evening news throughout a full year with just a few clicks.
“AI has proven very useful to us. Of course, it also has its limits as there are some categories it cannot track, for instance disabilities, just as it cannot monitor more qualitative insights like stereotypical portrayals. However, we truly believe that AI has the potential to form a solid base in our work. A potential that we’re just starting to see the full scale of”, says Heidi Sivebæk.
In 2023, DR will continue using the diversity tracker, bringing new features into use. For example, in the coming months the team will start testing how the tracker works on ethnicity.
This article was first published in issue 55 of tech-i magazine.