As a child of Arab parents in Germany, RWTH alumnus Hassan Sawaf learned at an early age how difficult it is to translate words from one language into another. He studied computer sciences, became fascinated with speech recognition, and worked on technological solutions for translation problems. Today, Sawaf lives and works in Silicon Valley – and has good advice for you regarding start-ups.
FutureMag: Mr Sawaf, when you founded your first start-up in 2000, the kind of network of advisory institutions and investors we have today – like the RWTH Entrepreneurship Center or the Founder Region initiative – didn’t exist yet. Where did you get the necessary know-how and support back in those days?
Sawaf: Back in 1998, some of my colleagues and at that time doctoral students shared my vision that we were working on some amazing things at the university with massive commercial potential: the technology to recognise speech from audio files; another technology that enabled us to understand and interpret speech; and a third that made it possible to translate text automatically. So in 2000, we founded the company AIXPLAIN AG to apply this technology as a commercial concept.
The first steps of founding and managing a company weren’t easy. But I had been able to gather previous experience, for example as one of the founders of ACME Computer GmbH in Eschweiler, as a member of staff at Catch Computer GbR and as a senior member of staff at GESYCOM GmbH. In addition, I regularly sought the advice of experienced entrepreneurs.
FutureMag: Was the knowledge and experience you gathered in your time at the RWTH helpful?
Sawaf: In terms of technology and technology management, the experience I gained there was of fundamental significance. To this day, I still consider maintenance of the connection to the RWTH to be extremely important. The high-quality theoretical instruction of students in combination with internships is of immense value. And my doctoral supervisor back then, Professor Hermann Ney, was an invaluable mentor. From the very start, he was a member of the company’s supervisory board and helped me keep on track.
FutureMag: Last summer you switched from eBay to Amazon, as the Director of Artificial Intelligence there. What are your tasks and responsibilities?
Sawaf: We want to minimise friction losses in everyday use. So we need to understand what the customer wants, in which contexts he or she expects specific things from Amazon, and possibly even predict what he or she will be interested in next. We also want to optimise processes inside the Amazon system in order to cut costs. This will allow us to make the best offers. And we also want to make the communication between the user and Amazon’s products and services more human.
FutureMag: What exactly do you mean by that?
Sawaf: Well, until recently, the interaction with Amazon – as with the majority of search engines – was primarily initiated with keywords that you type into a text field. The user had to learn to formulate his or her wishes in such a way as to get the best possible results to the query. That can be quite tedious. Human communication contains natural language, and in a wider context also contains images. There’s a good reason why we say “a picture is worth a thousand words”. Communication between people always uses a wider context, one that we see, hear and recall. Our ‘Alexa’ (voice command in Amazon’s loudspeaker system ‘Echo’; editor’s note) shows where we’re trying to get to, and the visual object recognition in the Amazon app on iPhone and android mobiles does that, too.
FutureMag: Digitalisation and artificial intelligence – a lot of people are sceptical about it, in terms of data protection and the ethical and moral challenges…
Sawaf: I agree with you that artificial intelligence brings dangers of an ethical and moral nature with it, and that, as researchers, we must acknowledge that this concern needs to be addressed. We need to get a range of people to come together – like technicians, scientists, philosophers, lawyers, politicians and economists – and create policies, regulations and platforms in order to maximise the positive potential of AI and minimise its potential hazards. To shape this discussion effectively, we need education and open dialogue. First steps in this direction are initiatives like “OpenAI”, which has committed itself to the democratisation of artificial intelligence. Amazon is in the OpenAI consortium. If we ensure that the regulations and platforms are properly adapted, I think that the benefits will far outweigh the risks.
FutureMag: You once said that far too little thought is given to the subject of education.
Sawaf: My hope is that, one day, the basics of artificial intelligence and its underlying mathematical foundations will find their way into schools and universities. I also hope that using the tools will become increasingly simplified so that not only specialists will be able to find solutions through them. At least in that respect we’re well on the way. Platforms are being made available by Amazon, Google, Microsoft and others, and the development of these platforms is open. Many scientists inside and outside the companies are helping to develop the technology further.
FutureMag: You’ve been working for a good few years in Silicon Valley and are well-acquainted with the businesses there. What makes them so successful?
Sawaf: There are a lot of different components at play. For example corporate culture. There’s more readiness to take risks. Many successful companies actually see it as negative if a certain portion of initiatives – serious and well executed ones – don’t flop. It’s taken as a sign that the risks taken were too low and so was the capacity to innovate.
The hierarchies are very flat. Even companies with 50,000 employees often only have five or six management levels. CEOs like to talk to people with new ideas, even when they are just fresh out of university. The concept of continuous learning is deeply engrained in everyone. And experience shows that improvement often comes from looking at a problem with fresh eyes. For me, neural networks for speech recognition and image processing are the perfect example of this.
Companies don’t only see the value of a member of staff who is a manager, but also respect what we call the “individual contributors”. They are often the most important employees for software companies and are highly appreciated – and correspondingly well-paid. Managers are more like senior staff members and mentors – “leaders” rather than “bosses”. In other words, employees are invited to see themselves as participating in decision-making processes, and this helps them to share a feeling of “ownership” with the senior staff.
Another important aspect is financing through capital providers, I mean venture capital and investment. Often, the VCs are not only bankers but also former successful entrepreneurs. This helps the communication between investors and borrowers and also anables the VCs to lean a bit further out of the window. Experience shows that these are the best investors and the best investments.
FutureMag: What qualities, ideally speaking, should a young entrepreneur founding a start-up today have?
Sawaf: Well, first of all self-confidence. What I mean by that is not just a self-assured appearance and manner. You also have to know what you’re talking about. And you need to be willing to learn. The world is moving so fast that you simply can’t learn fast enough if you really want to keep up.
Another important thing is being a team player. No-one can do everything, but you can always gather a highly motivated team together that covers many different aspects. The team needs to show mutual loyalty. That, along with motivation, is more important than having the best scientists or technicians.
And, for a successful start-up, steadfastness is called for. Even if someone else comes up with a similar, perhaps even better solution, you can still learn from that. In most cases, it makes sense to keep going. You can see the competitive situation as a confirmation that you’re probably on the right track.
And last, but not least: read, read, read. Papers, news, books. For my part, I do that every day. “Information is King.” Not just for artificial intelligence but also – and more so – for us. And it’s a good idea to keep broadening your horizons. My first positive experience with learning from other fields of research was when I got seriously into thermodynamics in the mid-1990s.
FutureMag: The Campus Melaten currently being created here in Aachen is one of Europe’s largest research landscapes – in a sense also a kind of “Silicon Valley”. What do we need to pay attention to when developing this sort of “hot spot”?
Sawaf: Having good mentors – I mean people who come from research and have lots of experience in implementing research findings and results – they’re essential for an initiative like this, even as critics. But most of all, the strong involvement of the RWTH needs to be maintained, and the same applies vice versa. What would Silicon Valley be without Stanford and Berkeley? What would Stanford and Berkeley be without all the successful start-ups in the region?
FutureMag: Thank you for talking to us.
This interview was first published in the alumni magazine “keep in touch” of the RWTH Aachen University (Nr. 62, WS 2016/2017)..