Looking back on AI events in October 2017
Companies talked about their experiences deploying AI systems at the IBM Watson Summit in Frankfurt and the World Summit AI in Amsterdam. Experts from Deutsche Telekom attended both events to demonstrate what the company has achieved in AI-powered customer service.
IBM Watson Summit, Frankfurt
There was a time when only Sherlock Holmes could call upon him, now everybody can have his or her very own Watson. The IBM Watson Summit, of course, was not about detective stories, but the eponymous artificial intelligence built by IBM.
Representing Deutsche Telekom were, among others, Vidya Munde-Müller, Product Manager User Interfaces, and Joachim Stegmann, Head of Future Communication. Here’s how Munde-Müller summarized the opening remarks by Martina Koederitz, General Manager of IBM Germany, Austria and Switzerland (DACH): “She explained that cognitive learning needs the cloud to better distribute all the rich information. That’s right on the money because only one-third of all corporate decisions turn out to be correct. It’s important to save and secure data. She reported that IBM was one of the first companies to develop an EU-wide code of conduct for data transparency. All very interesting viewpoints.”
Rats use digital networks to communicate
Dr. Miriam Meckel’s keynote also grabbed the audience’s attention. According to her, research into controlling computers by thought shows that technology is already capable of recognizing letters. Going the other direction, i.e. to feed external data into the human brain via an electronic interface, is proving to be more difficult. It’s been successfully demonstrated in experiments with rats, even over a large distance. That’s how a rat in Washington was able to learn new tricks from a rat in Rio via a digital network.
Munde-Müller’s take: “Facebook has the vision to let people dictate 100 words per minute just by using our brain but no keyboard. Several startups are exploring how to feed information from a computer back into the brain. Applied to medicine, such technologies could open up new opportunities for paralyzed patients which aren’t able to speak but now could use written communication.”
Meckel also reported about a self-experiment after which he was unable to sleep for 36 hours. Her final question for the audience was to ponder what happens if brains were to develop a life of their own? “When humans start implanting chips into their brains, the future might belong to neuro-capitalism.” Munde-Müller agreed with much of her keynote: “It was a great talk.”
AI powers customer service at Deutsche Telekom
Shorter panel discussions rounded out the afternoon. Joachim Stegmann spoke about Architecting Disruption or how Telekom uses AI for customer service. Currently, the various service offerings are fragmented and not connected to each other, meaning that different touch points access different data silos. There’s static, canned content, while customers experience real or human, support and advice via self-service. The experience with such “self-care,” though, is still rather limited.
Stegmann promised that would change. “Our goal is to create just one single source for eCare which consists of our own and third-party offerings. We’ll have AI-based, context-aware and personalized content so the customer will hardly notice the difference from manually produced service offerings. There will be one consistent assistant across all product categories.”
Stegmann also talked about the experience of using digital assistants. T-Mobile Austria has for some time now been working with Tinka, a chatbot handling customer inquiries. Tinka even made a brief appearance as a 3D hologram at this year’s Mobile World Congress, but her commercial roll-out is not (yet) on the horizon. She’s mostly about demonstrating what’s technically possible based on language and gesture recognition as well as classifying intent.
Here’s a summary of Telekom’s successes, according to its innovation expert: “As a proof of concept, AI as already proven itself when it comes to chats or fielding customer requests via phone. She is available without a wait 24/7 and can already answer most questions. If necessary, the AI can seamlessly hand over to a live agent.”
AI has also demonstrated its value as a proof of concept to support its human colleagues. “Agents find the correct answers faster and gives us a guarantee that information is of consistent quality,” Stegmann said. Finally, he presented lessons learned. When it comes to data processing, NLU training and extracting information, all challenges have been solved. NLU performance, knowledge management and basic dialog management are on their way to be checked off, too, while advanced dialog management and dynamic learning still need work.
World Summit AI, Amsterdam
The World Summit AI opened just a few days later. It drew the brightest minds from the fields of AI to Amsterdam, totaling about 2,000 active AI researchers from around the world. Its two days of exchanging knowledge, networking and listening to keynote presentations brought together the three pillars of the AI ecosystem: companies, startups and investors, as well as the deep tech community.
Jan Morgenthal, Chief Product Owner at Deutsche Telekom and responsible for its eLIZA project, participated in a panel featuring an enterprise story where he could grill AI startups and experts.
Attendees of the presentation by the first Afghan all-girl robotics team were in for a treat. After the six teenagers had recently been denied entry into the US where they had hoped to participate in an international robotics competition, they presented their work in Amsterdam. Morgenthal was impressed: “Even though they live in a country that at times appears medieval, Afghan girls are actively engaged in AI and robotics. It was extremely interesting and a good sign for how technology can bridge the gap between modernity and tradition.”
How NASA uses AI
Morgenthal also met quite a few interesting AI experts and vendors who are important collaboration partners. “At the enterprise level, it became clear very quickly that we at Telekom are ahead of the market, at least in the EU,” he says with a smile. “I guess I helped other attendees from the enterprise space more than they could provide me with input.”
One of the highlights in Amsterdam was a speech by NASA scientist Steve Chien. “They’ve been using machine learning for decades. Chien gave a great overview what role this played in the case of the Curiosity Mars Rover,” says Morgenthal. “It’s light years apart from from my personal work and Telekom’s work, but NASA’s long operative deployment and the results they’ve achieved were even more impressive from that angle.”