CHIP: Mrs Nemat, what does Telekom have to do with artificial intelligence (AI)?

Claudia Nemat: A lot. We have been using artificial intelligence for several years. A practical example is automated route planning for the 2.5 million fiber optic connections we lay each year. This wouldn't work at this scale without algorithms and artificial intelligence.

You need to explain this in more detail.

Nemato: We have high-tech vehicles, the so-called T-Cars. They film, measure and analyze the expansion paths of the optical fiber with the help of numerous cameras and sensors. An AI evaluates the images and calculates the best routes. This saves us many hours of on-site inspection. In Germany, but also at Telekom as a whole, there are not as many fiber optic planners as would be necessary if we did not have this AI.

Is this the only area where Telekom uses AI?

Nemato: No. There are many more. For example, we also use AI to defend against digital threats more effectively. Every minute there are 19,000 attacks on the Internet. Artificial intelligence helps us recognize harmful patterns and react in real time.

Have you established guidelines for dealing with AI?

Nemato: Yes. Already in 2018 we gave ourselves ethical guidelines to address AI. My favorite guideline is: the ultimate responsibility lies with the people.

“AI is a powerful technology, but it also has its disadvantages”

That sounds contradictory. Shouldn't AI replace humans?

Nemat : No. AI is a powerful technology. This makes it easier to solve many problems. But there are also dark sides. At Telekom we take a “people-centric” approach. On the one hand, this means that AI-based technologies should benefit people and not harm them.

And on the other hand?

Nemato: On the other hand, we want to raise awareness that the ultimate responsibility lies with the people. For example, our customers should always know if they are talking to an AI or a real person.

In addition, we regularly check the results of the AI ​​we use. For example, an algorithm at one of our national companies was no longer suggesting certain rates to seniors. We find that to be ageist. Then we correct the algorithm.

They are working with South Korean company SK Telecom on their own AI.

Nemato: Here I have to specify: we use and test a wide variety of large language models. For example, we use GPT 3.5 and 4 from OpenAI, but we also test open source models like Llama de Meta. Another chatbot we are testing is Claude from Anthropic.

And what is the model in question in relation to South Korea's SK Telekom?

Nemato: Exactly. We have been working with SK Telecom for many years on various technological issues. Together we are now sending the Anthropic model to a kind of summer school to “fine-tune” it. We want to know if Telekom can use it to answer questions better than with other models. The goal is not to invent our own AI, but to perfect the best existing technologies before incorporating them.

What does that concrete look like?

Nemat : Years ago we started creating an AI-based chatbot that our customers can use when they need help. For several months we have been using large language models to answer non-standard questions.

For example: I have booked rate X in Germany, what costs will I have to pay for a stay in Turkey? There is also a bot for employees. I like to call him the “employee concierge.” For example, anyone working as a fiber optic planner can ask questions to the chatbot and will no longer have to laboriously search through 9,000 pages of PDF documents.

“AI is a bicycle for the brain”

This means that Telekom relies on AI for customers and employees.

Nemato: Exactly. I believe AI will greatly change the way we live, learn and work. In the telecommunications sector, but also in other areas. Perhaps we cannot yet imagine what will be possible.

Do you understand that this scares a lot of people? Automation, the integration of artificial intelligence in more and more everyday areas?

Nemato: Absolutely. But that's nothing new. Even when horse-drawn carriages were replaced by cars, many people were afraid. Because the second intelligence, that of the horse, disappeared. Artificial intelligence challenges us in particular.

Steve Jobs once called the computer a “bike for the brain.” The same applies to AI. We could “ride” it faster, but we have to learn to ride a bike again. There is also the fear of losing control. I think we should “embrace” AI. Of course, without being naive.

How does it work?

Nemato: Preferably playful. At Telekom, for example, there is a project called LEX, Learning from Experts. This is an international network of employees. Members train each other. Telekom has also set a goal of providing 70,000 employees across Europe with basic AI skills. Our first Promptathon took place last summer. People from all over Europe came and were able to test the instructions in real day-to-day tasks.

What exactly is the increased use of AI at Telekom aiming to improve?

Nemat : On the customer side, things go faster: fiber optic connections are installed faster and problems are resolved more efficiently. You can also more accurately predict where interruptions or errors might occur. AI also helps provide our customers with personalized offers. Our employees also get answers faster thanks to AI, for example in fiber optic planning or marketing.

“AI helps us recognize harmful patterns”

Doesn't artificial intelligence make everything more impersonal? It seems like at some point you can no longer talk to anyone when there are problems.

Nemat : It always depends on the combination of people and AI that is supposed to support people. Therefore, we continue to consciously use people. For example, we have service teams that specialize in helping seniors with technical questions over the phone.

It is important to us to give all clients the feeling that they are really being helped. People are always the center of attention. Furthermore, there is no concept of “a customer.” My mother has different service requirements than a Gen Z representative who would like her questions answered through the Magenta app.

AI also carries risks. Cyberattacks are becoming more sophisticated and AI applications may also have vulnerabilities – keyword: fast hacking. How do you want to prevent AI damage at Telekom?

Nemat : Both with AI and with humans. AI helps us recognize and respond to harmful patterns. But people are still important. We have several of our own Red Teams. These are employees who investigate how computer systems can be circumvented. To do this, they simulate attacks. We also sometimes hire external teams to find and close security gaps.

Many AI systems still have teething problems. Isn't it dangerous to trust artificial intelligence too quickly?

Nemat : Of course, it is not advisable to trust AI too quickly. But I think it's important to engage with technology and try it. We always do this initially on a smaller, more controlled scale. Our motto is: “Security by design”. This means that when we develop hardware and software, we aim to design them from the beginning with as few vulnerabilities as possible and to be resistant to attacks. We tried a lot. But there is no such thing as 100 percent security anywhere.

They repeatedly emphasize: “Let us become a republic of digital AI opportunities.” Where exactly should Germany rely on AI?

Nemat : I am convinced that AI can be an opportunity, especially in view of the demographic change. In many industries there are not enough apprentices and not enough people who can do the work that needs to be done. Keyword: shortage of qualified workers.

AI models can compensate for this, at least partially, in the future. Artificial intelligence and machines can also help older people be independent for longer. Robots can be used, for example, for housework or gardening.

“AI offers many opportunities, but it is not a magic pill”

Are there other areas where you think we should use AI?

Nemato: Absolutely. It is essential to respond to the consequences of the climate crisis. In the United States, we use AI to analyze high-resolution videos from 5G still cameras. The AI ​​detects the appearance of smoke in forest areas, that is, signs of fire.

This way, firefighters can arrive and react more quickly before wildfires get out of control. This is just a small example. AI can also be useful in healthcare, for example to detect diseases at an early stage. In short, AI offers many opportunities. But one thing is also certain: it is not a miracle pill for all the problems we have.

What would you say to people who fear losing their jobs due to AI?

Nemat : Unfortunately, today there are many things that are cause for concern. But change as such should not be part of it. It is important for us as human beings to shape the changes and overcome the fear of losing control. Every new technology has changed the world of work. In retrospect, this was a positive for most people.

For example, 150 years ago there were many more blacksmiths or weavers, but not taxi drivers or controllers. What I mean: The nature of work will change in all professional groups. We do well as a society and as a business when we make sure people are ready for and shape change.