Artificial intelligence: on the path to deep learning

A Berlin story could not be any more typical: a small group of people meet at well-known hacker location c-base in downtown Berlin to discuss artificial intelligence (AI). The idea evolves and becomes the world's most important conference for artificial intelligence. 40 AI experts are speaking at what is now the 4th Rise of AI conference, taking place May 17, 2018 at Deutsche Telekom's Berlin offices on Französische Straße. Some 500 people are expected to attend. Fabian J. G. Westerheide, CEO of Asgard Capital and initiator of the Rise of AI Conference, about current trends in the artificial intelligence market and Berlin as a center for AI.

Berlin is a visionary city. I feel very free here. Berlin has cultural openness, history, but, most importantly, no encrusted structures
Fabian J. G. Westerheide, CEO Asgard Capital/Initiator Rise of AI

Mr. Westerheide, "Learn, Understand, Adapt" is the motto of this year's Rise of AI Conference. What do you mean by that?
 

Artificial intelligence is already a part of our lives, but we're not always aware of it. It is therefore extremely important for us to understand the current status of artificial intelligence, to share our vision, and discuss the impact of AI on human life, business, society, and politics.

Why is there such hype surrounding AI at this time? 

The topic has experienced a renaissance in recent years. There are several reasons for this: First, processor performance has increased. Computers today are good enough to handle highly complex tasks. Face recognition on the iPhone, for example, only needs a tiny chip to run. In addition, large amounts of data are being generated by the many end devices we use to communicate nowadays. The increased computing power and the Big Data being gathered together form the basis of today's artificial intelligence. Software can already play chess, change money, and navigate cars and boats. That's what's called narrow or weak AI. But the sector is developing rapidly.

And what's the next step?

The next step will be so-called deep learning: Machines will be able to optimize themselves. Chatbots like Siri or Amazon's Alexa are already examples of this. This strong AI will allow motivated machines to be built that can learn from mistakes. 2012 was a milestone in the development of deep learning. Researchers at Google's X-Lab in Mountain View, California fed photos from YouTube into the "Google Brain", a network consisting of around 16,000 processors, for a week. The system searched the data for patterns and finally filtered out three categories: photos of human bodies, human faces, and cats. The experiment ushered in a new era in the history of artificial intelligence because it demonstrated how machines were beginning to approximate human thought processes. In 2014, Google bought Deep Mind, a British start-up specializing in artificial intelligence programming. Both events were turning points in the history of AI. Companies like Facebook, Amazon, and Intel got involved. Now the market is booming. According to a McKinsey study, companies around the world invested about $27 billion in artificial intelligence in 2016 and another $12 billion was invested externally in AI, for example by venture capitalists. That's three times more than in 2013.        

However, the use of artificial intelligence also carries risks... 
There are ethical, moral, and security aspects that we need to consider and discuss. Learning systems adopt the culture and mindsets of their teachers. In this respect, it is important that we develop a European AI. The pioneers in the market are currently the USA, China, Russia, and Israel. With a market share of just 3%, Germany is lagging far behind. 

What can Germany do in concrete terms so that it does not miss out?
We need more professorships in cognitive science, more capital being invested into AI research and development, and we need more applications. AI must be integrated into the processes of companies. Currently, artificial intelligence is mainly used in the field of communications, i.e. in sales and marketing, and not enough in industry. AI can make machines smarter and digitalize processes. No matter whether it concerns the production of cars, drones, trucks, or printing machines.

However, companies are often still afraid of losing control...
Sure, artificial intelligence makes mistakes, but so do humans. If machines make mistakes, that's fine, as long as they don't do it more often than humans. They need to learn, after all. Accidents involving self-propelled cars have been the subject of a lot of discussion lately. Studies have shown that the risk of accidents with smart cars is significantly lower than with vehicles driven by humans, when you take into the number of miles driven. 

Where is artificial intelligence already being successfully used today?
When large amounts of data are available, artificial intelligence is already superior to humans. By comparing data, it can analyze patterns and recognize interrelationships in the shortest possible time. This is already being used in the healthcare sector, for example in the early detection of dementia or cancers. In the FinTech area, computers help to detect forgery, money laundering, etc. And processes at Amazon now run completely via AI. The only human interface to the customers are the delivery personnel. Micropsi Industries is based in Berlin. It has developed software that teaches robots how to learn. They imitate human movements, optimize them, and then work more precisely than humans.  

As a digital capital, is Berlin also a center for AI?  
There are now a lot of reasons why Berlin should be of interest for the AI sector. These include the proximity of lawmakers, the many corporations headquartered here, the dense research landscape, a strong start-up scene, and the media attention. Berlin is a place where everything happens. That's why more than half of all AI-relevant companies are located here. However, the artificial intelligence market is global. In the US, for example, the corporations are younger and more agile. Berlin should therefore not rest on its laurels.

You've lived in Berlin for ten years, why?
Berlin is a visionary city. I feel very free here. Berlin has cultural openness, history, but, most importantly, no encrusted structures. I like this mix of laissez-faire with enough rule of law. Here everyone has a chance to go from trainee to millionaire. Everything's possible in Berlin. 

Photos:

Header: ©Shutterstock/Phonlamai Photo
Rise of AI-Konferenz: ©Saskia Baumeister

Back to stories