Innovation is never mediocrity
Matthias Zimmermann about his passion for machine learning
Matthias Zimmermann was born in Zug, the cradle of Crypto Valley. When looking at his latest projects – blockchain, AI and the BSI Lab – you might consider his birthplace to be an omen.
- What do you do with 40 minutes of free time? Drink coffee, read, observe people
- What would you like to learn next? To deepen my knowledge of blockchain and machine learning and to implement something concrete before the next explorative phase comes
- Definitely want to: go into space
- Don’t want to: arrive – the question of what I want to do when I grow up still remains unanswered
- Morning ritual: giving myself enough time to wake up – and then a cappuccino
Matthias, your hobbyhorse, machine learning, is totally in at the moment. You have been preoccupied with it for nine years, dedicated your Master´s thesis, doctorate and post-doctorate to this field. Why is it now suddenly so red hot?
Ten years ago, machine learning was a specialized topic that didn’t particularly interest most people. While academic progress was made, the machines often weren’t advanced enough to deal with the complexity of “real” tasks. The breakthrough came in 2012: recognition rates were massively increased with the help of deep learning in the ImageNet competition. This new technology triggered a quantum leap in the field of artificial intelligence. Today, neural networks are experiencing a renaissance with deep learning – not only in the laboratory, but also in the real world – which is also attributable to the growing amount of data and improved computing performance.
That really frightens a lot of people.
That is understandable. There is an increasing number of areas in which machines are outperforming people – from medicine to the granting of credit. Keyword: narrow AI – for sufficiently limited tasks with enough relevant data, machines can perform as well as or even better than humans. Nevertheless, we are still far from a general form of intelligence. There is not even a consensus about the definition of general intelligence. As soon as AI needs to think around two corners, it is still in its infancy.
So are humans still ahead of machines?
Machines are lost when the environment strongly deviates from the trained examples. Even today, machines are hardly able to explain their decisions upon inquiry. A judge, however, is usually able to explain his decision even to lay people. Machines are still unable to do well in such situations.
Then what advantages to they offer?
That is rather obvious: no unions, no minimum wage, no naps. A machine only needs electricity and sufficient computing power.
What impact does this have on our society?
Robotics in combination with machine learning has a massive, widespread effect on society. It is currently assumed that in 20 years, half the jobs that are common today will no longer exist. The challenge lies in mastering this transformation. It will attain the quality of the industrial revolution and ultimately be even more comprehensive.
«Sometimes I get the impression that mediocrity is given too much attention.»
Matthias Zimmermann, BSI Business Systems Integration AG
How do you see the collaboration between humans and machines in the future?
Machines will increasingly relieve us from our daily routine work and assist us with making better decisions. They are new tools that make humans even more successful.
Curiosity and innovation also mean: no focus on short-term figures. Is that difficult in Switzerland?
Trying new things in Switzerland can be hard. When you fail, people are usually looking down on you. In the USA, you are congratulated for trying and wished better luck next time. Sometimes I get the impression that mediocrity is given too much attention. Average performance is viewed as solid and sometimes even appreciated. But innovation is never mediocrity! You have to move outside the box, at all times! Initiatives such as Crypto Valley from Zug to Zurich encourage and prove that it is possible to play on the world stage of good ideas from within Switzerland.
Where do you meet like-minded individuals who share your passions?
In meet-up groups and in the BSI Lab.
Now that you mention it: we have opened the BSI Lab in the Zurich office. What are your objectives?
It is both a place and program for free testing of new ideas around software and hardware. The point is to encourage motivation among people who want to accomplish something. My job is to channel these ideas, avoid duplication and allow participants to make mistakes too. The Lab should lead to even more ideas being tested and to create novel findings that may prove to be useful both internally and externally.