Kenza Ait Si Abbou, Deutsche Telekom: “Think twice about who you work for”
Kenza Ait Si Abbou works at Deutsche Telekom as a Senior Manager Robotics and Artificial Intelligence. In her job, she and her team develop solutions for all sorts of use cases ranging from HR processes to Q&A pages for customer support. She has also dedicated herself to the mission of making AI more understandable to consumers.
On this episode of REWRITE TECH, we talked to Kenza Ait Si Abbou about her passion for math and artificial intelligence and discussed solutions to make algorithms less biased.
Why is AI so biased?
There are several public examples that demonstrate that AI is not neutral. Like the Australian app Giggle that wanted to create a safe space for girls, but excluded trans-girls. Or even worse and potentially dangerous: facial recognition techniques that weren’t trained enough to differentiate people of colour but were used by law enforcement.
Right at the beginning, Kenza makes clear that such cases are not the mistakes made by artificial intelligence: “The biases don’t come from AI, they come from humans. The problem is that we are the problem.” Specifically, our unconscious biases — stereotypes we have without being aware of them — are a big problem because we don’t see them. “We teach these biases to the machines and the machines then learn from this data”, Kenza explains.
Kenza Ait Si Abbou: “We need responsible consumers”
As a first step, Kenza believes, developers and companies must admit that they have an unconscious bias and consider those biases already in the ideation phase of a product or feature. “When the software is finished, it’s too late”, explains Kenza, who has more than ten years’ experience in this field. In the best case, checking software for bias is an obligatory requirement in each development process.
But even with more ethical practices on the developer site, it doesn’t release the user to make better and thoughtful decisions. “We need responsible consumers”, says Kenza. We can’t just rely on the things that AI-powered software presents to us, we need to “use our mind” as Kenza puts it.
“What we should keep in mind is that AI is just supporting activity in all sectors. It shouldn’t be the decision-maker. It should help to make a decision but not make the decisions.”
How the Deutsche Telekom uses artificial intelligence
Kenza spent the last ten years working at Deutsche Telekom in Berlin. As a Senior Manager Robotics and Artificial Intelligence, she and her team develop solutions powered by AI. This includes internal tools for human resources, but also consumer-faced technology like the Q&A section that helps clients to find the right answer to their problem automatically.
One thing that makes Kenza especially happy is that Deutsche Telekom was one of the first corporations that developed guidelines for the use of AI. To do that, Deutsche Telekom held several workshops with people from different professions and age groups to understand what they expect. “Our guidelines represent the wish of our customers,” concludes Kenza.
Teaching the art of AI
Apart from her role at Deutsche Telekom, Kenza is a very vocal member of the tech scene in Germany. Her goal is to educate people on AI and explain how deeply integrated it is already in our lives. Recently she published her first book: “Keine Panik, ist nur Technik” (Eng: “Don’t panic, it’s just technology”). She wants kids and adults to have fun exploring new technologies and to understand that machines are only as smart as the people who develop them.
At the end of our conversation, we asked Kenza about a piece of career advice she would like to pass on. After thinking about it for a moment she says: “Think twice about who you work for” and continues: “The impact you have is much higher than you might think.”
Listen to REWRITE TECH with Kenza Ait Si Abbou
You can listen to the whole conversation with Kenza Ait Si Abbou on the REWRITE TECH podcast. Our podcast is available on your favorite streaming service including Spotify, Apple Podcasts and Google Podcast.
