There are any number of good reasons as to why the tech industry needs female reinforcement. There’s a shortage of skilled workers. Homogeneous teams and uniform ways of thinking represent a clear obstacle to innovation. The digital industry is booming, new digital business models are being created each and every day, and lucrative jobs are being created. But all too often, women are still missing out. The eco Association wants to change that. As part of its “Women in Tech” topic field, eco invites inspiring female specialists and executives from the tech industry to take the floor in a series of interviews. Here we deal with the important topics: from development perspectives, career tips and hopes for the future, to the challenges in a male-dominated working environment – and ultimately, to highlighting why working in the Internet industry is fun. In this issue of dotmagazine, you’ll find an interview with Claudia Pohlink, Head of Artificial Intelligence at T-Labs, which was first published in German as part of the interview series on eco.de on 31 March 2021.
Hanna von der Au: As Head of Artificial Intelligence and Machine Learning at T-Labs, Deutsche Telekom’s unit for research and development, you and your team work on topics such as the use of AI in cybersecurity, quantum tech, and network automation. What does your daily work look like and what do you particularly like about your job?
Claudia Pohlink: At T-Labs, we are essentially bridge-builders between the academic and business worlds. We act as tech scouts. This includes identifying trends, with a current example being that of zero-shot learning. With this, I can enable an AI which doesn’t have the image of a zebra to recognize a zebra image in the future by training the AI with an image of a horse and adding the missing information via text formulations (e.g. + white, + stripes). In addition, we identify further topics that are discussed in universities and the academic realm, which will become important for businesses in 3 or 5 years.
On the other hand, from these topics, we derive actions and business fields for ourselves as a company. In doing so, we cooperate with research institutions, start-ups, and universities, and are guided by the question: Which topics will be relevant for us as a telecommunications company in the next five to ten years? The most exciting thing about my job is that I’m always working on the latest trends, and for me that creates no end of fun and learning.
Von der Au: What do you see as the tech trends for 2021? What will be the next disruptive innovation or the next big thing in the field of AI and machine learning?
Pohlink: The next big thing is most definitely quantum. Think of how rapidly hardware is developing in this field. IBM, for example, is very strong in this sphere. D-Wave (from Canada) and Google are also pioneers in this field and are developing competing hardware. Amazon now also wants to move in the direction of hardware development. Other major players such as Microsoft and Alibaba have a lot going on in this area. In addition, the development of applications is of course also relevant.
Quantum computing is presenting us with a completely new era, because it is replacing the classical “zeros and ones” programming learned in computer science. Here the effects and insights of quantum physics are being fed into computer science and later even refined with AI and machine learning. This is truly a disruptive, completely new way of processing data – huge amounts of data can be processed at an incredible speed.
Von der Au: What does that mean for business in concrete terms?
Pohlink: Let’s take traffic flows and navigation as an example. If we use the navigation system in our car and are heading towards a traffic jam, we are shown an alternative route. However, other drivers will be presented with the identical detour to avoid the congestion, with the result that a jam also ends up on the alternative path.
Quantum computing vastly increases the options and combinations for alternative routes. Quantum makes it possible to manage traffic in real time without incurring any disadvantages, for example, by sending car A to the right and car B to the left. Quantum computing doesn’t need to only take the shortest or the fastest route into account as a variable, but can include many more variables and for example optimize the overall traffic flow, also for emissions. Currently, no conventional computer can do this in real time because there are simply too many possible combinations. Quantum physics and quantum computing make such complex calculations possible at enormous speed and with great intricacy. This will contribute significantly to solving global, complex problems in the future.
Von der Au: What does that mean for our back-end digital infrastructures? Because, after all, the data also has to be processed. When do you think we will see the first use cases based on quantum?
Pohlink: That’s bound to be a question currently in everybody’s minds. I believe that concrete use cases will definitely be implemented within the next five years. Primarily, this will involve use cases in the fields of chemistry, molecular biology, and in the finance industry. However, it’s not just about speed, it’s about tackling complex computational problems that could never be solved before. For example, a research team at the Free University of Berlin recently succeeded in solving the Schrödinger equation with the help of AI. I am therefore very optimistic that we will be able to master complex global challenges with quantum technologies in the future.
Von der Au: AI can have a very positive effect, but on the other hand, there are also instances where AI-based solutions are contributing to the reinforcement of bias, for example. Are fears about AI justified in your view?
Pohlink: In their current usage, AI-based solutions consist of mindless, repetitive tasks that are sequenced and set in motion: such as robots that dance to the beat, climb steps, or do somersaults. That may be cool, but it’s not really intelligent. These days, an AI can usually do one thing particularly well: play chess, dance, or recognize faces, for example. But there are hardly any applications so far that are in a position to solve multiple tasks. Which means that fears of a truly intelligent AI rebelling against humans are definitively unwarranted.
But on a smaller scale, every technology has its risks, and in technology development, we naturally always have to factor in the societal impact. That means we need to be mindful of potential dangers at all stages. For example, if training data for AI models shows a certain bias, then the outputs of this model will be biased. This is not a problem of the technology as such, but how we add safeguards in development, training, and operations.
Von der Au: How does this work? In the field of cybersecurity, you are committed to identifying and minimizing unconscious or conscious risks of AI algorithms and have developed a technical solution that makes bias technically measurable. How does that work exactly? And to what extent does it help to eliminate or rectify bias?
Pohlink: In the first instance, we distinguish between conscious and unconscious hazards, with the former leading to distortions in the system through targeted manipulation or attacks. Some readers may already be familiar with this from the field of autonomous driving, where a stop sign tagged with a sticker can manipulate the car’s AI. Applied to our telecommunications business, this would be conceivable with a voice assistant, for example. We make our systems robust against such attacks so that nobody can suddenly trigger a command not intended by us by laying on disruptive noise.
In addition, we are dedicated to preventing unconscious distortions, otherwise known as bias. As a telecommunications company, for example, it is important to us that our existing mobile and fixed-network customers do not switch to competitors. To retain customers, we naturally also engage in targeted marketing campaigns.
At one national company, we discovered that a bias in the system meant that customers over the age of 50 were not being included in certain marketing campaigns. Simply because the AI had learned that people over 50 don’t tend to change their contracts. But people who are 50 today are certainly very different than that age-group was ten years ago. Bias can thus be quite damaging to business, which is why we focus on technical solutions that detect bias in AI-based systems.
Our third key topic is privacy. We make sure that the test and training data used in machine learning models are also completely protected.
Von der Au: Which societal problem would you personally like to see resolved by AI?
Pohlink: I’m firmly convinced that AI can be helpful in tackling certain societal problems. What I am personally particularly excited about is the link between medical research and AI. A combination of these two fields for beating cancer would be something I would have the greatest respect for. In general, I’d like to see Europe become a greater player in AI developments – after all, it has worked very well in vaccine development.
Von der Au: You are the daughter of a mathematician father and a mother who worked in the field of IT, so you had early contact with STEM and IT. You are an advocate for digital education for children and also a speaker on topics such as diversity and digital education. From your perspective as an AI/ML and robotics expert and as a mother, what do you see as the challenges in digital education, and what can politics, business, and society do to spark interest among students for the tech industry?
Pohlink: In the field of digital education, there’s no doubt but there’s some catching up to do. Especially in home-schooling times, we’re all experiencing things which make us feel like wringing our hands in dismay. However, my motto is: Instead of moaning, take action. Everyone can make their contribution to digital education. To give you a concrete example: I always took time off for the science days at my daughter’s elementary school. I packed tech paraphernalia like VR goggles and robots, and taught the kids about tech topics. A colleague of mine started a tech club at a school. As a corporation, we are also involved in and support digital education, for example by inviting school classes to our T-Labs. For me, having no time is not an argument. My day only has 24 hours. If you don’t have time, you take it, and I prioritize things differently and say that this is my responsibility and I’ll put it into practice.
Von der Au: The fact that that you are personally so committed to digital education is terrific. You also host visits by school classes to your T-Labs. What can students expect there?
Pohlink: We explain AI to the children using a child-friendly approach. We also ask them what they already know. Straight away, we get answers about vacuum cleaner robots, or they name voice assistants like Alexa or Siri. I’m impressed by the knowledge the children already have. In the T-Labs, we have four stations with robots where the children can try them out and actively participate. For example, they can learn to program with a Calliope mini single-board computer. NAO, the little robot brother of the famous Pepper, also stirs up great enthusiasm among the kids. The best thing for me personally is to see how enthused the children are about the subject – and that even the biggest rowdies hang on my every word.
Von der Au: So, do children not have any fear of contact with AI?
Pohlink: Children are very open and like to try things out. That’s evident from the creative questions they ask Alexa in our workshops, for example. Of course, we have to teach our children a certain level of media literacy. MIT scientist Stefania Druga has been researching how children interact with AI for years, and her research findings confirm that children are open to AI. If there is any fear of contact, it is because parents have imparted it to their children.
Von der Au: In your team, there are two women, yourself included. Where do you think we stand on the issue of women in tech?
Pohlink: As I see it, there are many fantastic women in tech. The challenge at the moment is the lack of visibility. However, I am confident that this will be different for the next generation of women. In my view, things are changing – in small steps, but they are changing.
Von der Au: You’ve had an impressive career yourself, working for corporations such as MTV, Jamba/Rocket Internet, and Deutsche Telekom. In retrospect, what qualities or skills were particularly helpful for your career? And what is your career tip for women?
Pohlink: I would like women to simply be more courageous, to speak up, to just say things and stand up for themselves and their projects. I think a lot of women are raised to be reticent. In meetings, it’s often men who speak up first. Women tend to hold back, perhaps because they are afraid of saying the wrong thing. Women absolutely have to get rid of this sense of reservation.
Von der Au: If we gave you another job and made you the editor-in-chief of a leading media publication, what would your headline be on the topic of women in tech and what would your article say?
Pohlink: That’s a tough one. Nowadays, when a woman becomes a CEO, this can prompt massive headlines in leading media publications. I wish it were quite a normal thing that there are female CEOs or board members. My leading media publication would therefore only mention the competent person who is the new CEO because they have achieved great content, without explicitly mentioning whether it is a woman or a man.
Von der Au: As part of eco’s #LiT Ladies in Tech interview series, we recently interviewed Dr. Julia Freudenberg, CEO of the Hacker School. Now we’d like to put a question to you that Dr. Freudenberg offered for our next interview partner: There are a whole host of negative beliefs that hold women and girls back and put them at a disadvantage. How can we as women concretely contribute to dispelling these beliefs, reinterpreting them, and replacing them with new, positive ones?
Pohlink: I think we all need to work on our own prejudices and stereotypical images. Far too often, we simply have stereotypes before our eyes: the man as a hunter, the woman as a gatherer – that’s simply not the case. An article from the German Stern magazine comes to mind: “End of a patriarchal myth – women went hunting in the Stone Age”. Researchers found bones of a hunter in the Andes. They were surprised that the bones were so small and light. The startling realization: they were the bones of a female hunter and thus proof that, nine thousand years ago, there were also female hunters.
Von der Au: In the context of diversity and gender, what question might you yourself put to an interview partner?
Pohlink: What really perturbs me is the question: Why do we in Germany have such a low proportion of women in top management in the industry compared to other countries? We have had a female chancellor for 16 years, Ursula von der Leyen is President of the European Commission. We have a high number of female business administration students. Why is this not reflected in the top management of German corporations? And what are other countries doing better than we are?
Claudia Pohlink is Head of Artificial Intelligence and Machine Learning at T-Labs, Telekom’s unit for research and development. With a background in data science, data management, and innovation management, Claudia Pohlink seamlessly connects business and data science aspects of analytics and artificial Intelligence (AI). Projects from Claudia’s previous position in the Deutsche Telekom’s Chief Data Office include ‘Data Cockpit’ (data transparency and data control for end customers) and a ‘Portal for Intelligence & Analytics’ (internal community for data and AI use cases).
Claudia’s team at T-Labs, the research unit of Deutsche Telekom, drives the adaptation of AI methods such as machine learning (ML) across relevant business areas. The team’s main research focus is on the application of quantum computing in the field of cybersecurity as well as on the future of networks and sustainability use cases.
Claudia is a member of the Bitkom Board for Artificial Intelligence. Moreover, she plays an active role in Berlin’s AI/ML and start-up communities and regularly shares her knowledge as guest speaker at kids´ events, schools, industry events, and reviews at Berlin’s universities. In 2019, she was honored as one of the Global Women Leaders in AI.
Hanna von der AU is PR Manager at eco – Association of the Internet Industry. She is responsible for the development of communication strategies and concepts, content marketing activities, social media channels, and press releases. The topics closest to Hanna’s heart are diversity and Women in Tech. She leads the activities around eco’s German #LiT – Ladies in Tech initiative and loves getting in touch with more women in the industry. Her aim is to make the Internet industry more colorful and diverse. Before joining eco in 2019, Hanna worked as a digital campaign consultant with a focus on content generation at a digital consultation agency (for customers like NRW.INVEST, GS1 Germany, Peek & Cloppenburg, Coop, Lufthansa).
Further information on the topic of diversity can be found on eco's Diversity focus page and in the eco Association 2020 study on Women in Tech Across the Globe: A Good Practice Guide for Companies.
Please note: The opinions expressed in Industry Insights published by dotmagazine are the author’s own and do not reflect the view of the publisher, eco – Association of the Internet Industry.