Archive

Archive

Blog.Telekom

Klaus vom Hofe

1 Comment

Artificial intelligence: shifting to a lower gear

Robots and artificial intelligence rank high in official journals. This is associated with fantastic future scenarios as well as uncertainty and dystopias. It's nice to see instances where the topic is grounded. Like now at Telekom's "AI Summit."

I had been a day part of the AI community. I found it exciting.

I had been a day part of the AI community. I found it exciting.

Trust of Telekom colleagues and customers is decisive for the success of AI projects – this trust has to be achieved and strengthened: Jan Hofmann opened the "2019 AI Summit."
Almost every chair is occupied, the hall is full of people and decorated with hearts carrying the message "We love AI." Colleagues involved in artificial intelligence – the "AI Community" of Telekom – came together in Darmstadt. I am looking around: professionals from a wide range of business units as well as from other companies. They want to network even more and exchange information on a topic in which more and more people are getting involved: artificial intelligence (AI). Since 2011, it has really picked up speed. Because from that point onwards, three crucial things were available: algorithms, big data, and computing power.

This means that systems are becoming increasingly more able to recognize patterns in large data. As early as 2015, automated face recognition overtook human face recognition – it is faster and more reliable. Nevertheless, it is still easier for humans not to misidentify women with short hair and glasses as men. Chatbots and digital assistants, which are shooting up like mushrooms, are also becoming increasingly capable. Not to mention virtual assistants, which are conquering millions and millions of households. Properly fed and trained and linked to databases and the internet, they help in many everyday situations. Around the clock and immediately. Whether it is in the cell phone, the household, or the processes of companies: AI is moving into ever more areas of life.

"Enormous disruptive possibilities of weak AI"

So it's a powerful subject that's evolving rapidly. I realize how important it is for the Telekom experts to exchange ideas, learn from each other, and pull together. The speed and the unknown also unsettle: I look around and see confirming nodding when Jan Hofmann, Top Program Lead AI at Telekom, points out the perspectives on AI. Some people therefore see it as a savior, others don't understand what it is all about, and still others are frightened: -What will become of it? Will this be controllable? ... And: What happens to my data? However, according to Hoffmann, the trust of customers and employees is decisive for the success of AI projects. Telekom's self-binding ethics guidelines for AI help to strengthen and even regain trust.

AI between hype and everyday life – Dr. Christian Schulmeyer from Empolis is one of those who further ground this topic. He explains that there is strong and weak AI. The latter can solve individual tasks faster than a human being. This includes all current AI technologies. There is no strong AI yet: a system that optimizes itself, deduces quickly, and learns.  Schulmeyer calls for the focus of the discussion to be removed from strong AI and focused on weak AI: "Let us use the enormous disruptive possibilities of weak AI in all industries, in public authorities, and in society."

Listening to customers

It's about an unimaginable number of applications. One of them is the chatbot of Telekom Deutschland – presented by Laura Knochenwefel and Stefanie Wallau. It should become the main point of contact for customers, and always transfer them to a real Telekom consultant when things get too tricky for the machine. People will focus less on recurring requests and more on the more demanding matters. The Digital Service Assistant masters over 450 "intents," matters in which it helps customers. 

The AI Summit also makes it clear that the systems behind this and other applications are only as good as the training they receive. Ivonne Engemann, responsible for chatbot development in the Lufthansa Group, knows that only those who listen carefully to their customers help their chatbot "get fit" by feeding it with the right keywords. It's all about the right topics, as well as the right synonyms. For the "Premium Economy Class" fare alone, airline customers have at least four other names, which need to be taught to the bot.

"We're talking about a completely new way of doing things," says guest speaker Chris Boos. "Everything revolves around automation." There seem to be no limits to the possibilities. The AI expert focuses attention on data, semantics, and the meaning of context and causality. Sounds convincing to me. After all, excellent AI should understand that as a father I will no longer have to order Pampers at some stage because the child has grown older.

In short, looking at AIs from different perspectives is just as important as looking beyond the Telekom Group's own horizons in this area. And it's good to hear that the SciFi scenario of strong AI doesn't have to frighten you today and that you should dedicate yourself to something beneficial instead: to the feeding and learning of systems for an infinite number of applications that enrich life. I think it's now clear to everyone that this can only be achieved together with customers and users and by applying ethical guidelines. Hopefully.
 

More Informationwelove.ai
Transparency is a must for artificial intelligence

Digital Responsibility

AI Guidelines

Deutsche Telekom defines its own policy for the use of artificial intelligence (AI).

FAQ