“Prejudices are not generated by AI, they are already there in the data used.“

Since 2017, Artificial Intelligence is part of the three core subjects of the ‘Telekom Innovation Laboratories’ (T-Labs) owned by Deutsche Telekom and since then the research- and development unit has been considered to be one of the most active AI-players in Berlin. #KI_Berlin spoke with Claudia Pohlink, Head of AI T-Labs about how to maintain a balance between specific solutions of problems as well as about social responsibility and established why the triumvirate of research, companies and start-ups is necessary for innovations to be successful.

Since September 2018 you have been responsible for the KI-Innovationen in the T-Labs of the Deutsche Telekom. What exactly are you working at?

In the T-Labs we check which of the technological trends and fields of application may be relevant for us. Currently this concerns three subjects: Blockchain, Experience and Artificial Intelligence. We are concentrating on three strategic areas: Network-Automation, Quantum AI and Intuitive UI. For the first area we have specific examples, especially where the optimisation of communication networks is concerned. To this end, we have made transparent how the network load is distributed in cooperation with the Telekom start-up Benocs. By processing historic data and thanks to Machine Learning, we can predict, which of the networks - such as for instance Netflix or YouTube - are the most important drivers, where the load appears and how it could be better distributed in future; whereby the latter two mentioned are explorative fields. In case of Quantum AI, it is all about improving speed or quality of the machine learning processes. In order to achieve this, CPU-intensive routines are carried out on a quantum computer. Intuitive UI (User Interface) on the contrary, adapts the user interface individually to a person by means of AI such that this person can operate it all intuitively.

You mentioned collaborating with a start-up…

Correct. T-Labs have entered into co-operations with start-ups for quite some time by now, have a tech-incubator with Hubraum and operate an own start-up programme with TechBoost. Numerous Deutsche Telekom spin-offs have been on the market for some years. Cooperating with start-ups is very attractive for us as they are fast, flexible, innovative units that have specialised in a specific area and contribute their knowledge which we can subsequently make use of. In turn, start-ups benefit from this cooperation as they can fall back on the infrastructure and contacts of a concern.

How does this cooperation work in detail?

I have recently read about a study on the innovation force of small and large teams on LinkedIn. It transpired that patents more often originate from smaller teams. The larger teams are the more suitable ones subsequently to carry out the implementation and innovations. This is interesting in as far as we, at Telekom, do exactly this: We do not research somewhere far away but do try to prepare specific solutions for customer problems together with start-ups and research institutes and incorporate this innovation force directly into the concern. Over the past ten to fifteen years we have learned a lot; it is really hard work. Working in a lab within a unit for Research & Development with its trend-scouting activities sounds rather cool and fancy; however, it is with a great effort that subjects are identified and transformed such that they fit into Telekom and are accepted positively by those units concerned. This is why the internal teams must be involved right from the start to be able to co-control the developments. It is a tight-rope walk.

What part does the location of Berlin play in all this?

It is definitely very helpful to be close to research institutes such as the TU Berlin, who are a strong partner of the T-Lab, or the Fraunhofer Institute as well as the start-up community. This triumvirate is represented very well in Berlin. In addition, we are also close to the talents, followed by the industry and other branches that are also located in Berlin with their AI-research institutes. We also have established research co-operations with Budapest and Vienna; Asia, too, or the U.S. are also interesting partners for us. But we should not forget Israel that is very strong in the AI - and cyber security section. Currently we are cooperating with the Ben Gurion University and our T-Lab colleagues in Israel on the possibility to recognise cyber-attacks by the pattern in network data before they actually attack and it is one of our objectives to achieve this in 2019.

But the T-Labs do not only want to find solutions for Telekom problems. They are part of the most active players in the IT-scene in Berlin...

Yes, it is really important to be active in the Community to gain credibility. This is why we started the 'Long Night of Start-ups‘, take part in Meet-ups such as  „AI Monday“, and also organise these ourselves. The ‘Meet-up‘ culture is very active in Berlin. You could attend events on Blockchain, AI or Experience every evening.

Apart from that you are involved in the Non-Profit-Organisation “Women in AI“. What is this all about?

(Takes a deep breath) Where shall I start...? Through my network I met Moojan Asghari, one of the founders of Women in AI. Together with two colleagues, Kenza Ait Si Abbou Lyadini and Vidya Munde-Müller –  herself a WAI ambassador in the meantime - we organised an event on ‘Women in AI’. The objective is to bring together women of the AI section on an international basis, exchange knowledge and to support them to appear on stages, in events and media.

Why is this important?

We need more women and more diversity in the AI sector. When working with Machine-Learning methods, the quality of the results depends on the data used. Prejudices are not generated by AI itself they already exist in the data used. I will give you an example: Amazon used AI during its application processes and found out that it was always male applicants that were suggested. Not women. This originates from historic data, as more men applied and therefore more men were employed. This is quite normal in a technological concern. We must, however, find a way to remove such prejudices and make recommendations independent from the gender. This does not only apply to women but to any other group - what we need is an increased diversity. Being a woman, I find it is my responsibility to commit to this much more.

In this connection you organised a ‘Hackathon’ for the first time with the T-Labs in 2018, especially for women. How did this come about?

We noticed that more men take part in Hackathons - irrespective of the subjects such as Blockchain, AI, or Data Science. We tried to find out the reason for this and noticed that women do not appear to be confident enough. They believe that they are not qualified sufficiently, although they are at least as good an expert as their male colleagues. The dynamics in a team is generally strongly dominated by men, women keep back. During the Hackathon for women, however, another dynamics was apparent. It was not about ‘who contributes most‘, nobody took the lead straight away, but they worked more team- and target-oriented compared to other Hackathons. We also received a fantastic feedback. Many of the women said: “Please let us have more of this - and not only in Berlin.” We have envisaged organising another one in 2019.

What else have you planned?

I generally think it is important to maintain a balance. On the one hand we envisage two specific projects each year that will have an impact on the concern - Cyber Security and Network Automation are the subjects for 2019. On the other hand we are committed to education. We want people to recognise the use and also see what Deutsche Telekom is working on already. When talking about AI we have experienced many times that everything turns around the question of ethics very quickly; and this is why we need transparency. Therefore we at Telekom have established our own binding principles in respect of dealing with and using AI. We also want to demonstrate the current innovations, to remove anxieties but also to meet expectations. We find that this is particularly important also for young people. In January 2019 we offered an event ‘AI for Children’, where children were able to test different robots and AI at six different stations. They could ask Alexa questions or interact with Nao or Lego robots. It was most interesting to see how curious children really are. In their world robots and machines are already part of their daily life. We received very good feedbacks and enquiries from numerous cities. There are many very different events we are thinking about.

Mrs. Pohlink, thank you very much for the interview.