Ulli Waltinger: 'As more interconnected we are, cyber threats will rise and the security will get more important in industrial sector'

Autor:

  • Karla Juničić

14.10.2019.

Hanza Media

Ulli Waltinger, founder and Co-Head of the Siemens AI Lab at Siemens Corporate Technology

It was a Wednesday morning when Ulli Waltinger received a message from his colleague: 'You flight has been cancelled, the airline company is under financial stress'. The first thing he thought of is: 'is my phone being hijacked, how is this possible?'.

We would think that this is something only a person like dr. Waltinger can think off, a person who leads Machine Intelligence Research Group and the founder and Co-Head of the Siemens AI Lab at Siemens Corporate Technology. But it's a wrong assumption. We live in the world where fake news are posing a security threat and can impact our everyday life.

- As we digitalize and as we are more interconnected cyber security needs will increase – said Waltinger.

Adria Airways airplane did stay on the ground that day and comapany collapsed in bankruptcy just like in the chain reaction following its counterpart Thomas Cook in UK. But despite all the confusions in reality or cyber reality, dr. Waltinger landed safely for his interview with Euractiv.

 

We often argue how the AI can take the digital transformation to the new level but only with the human help. Still it seems like innovations in AI go much faster and further than human knowledge itself. If people are not familiar with something it makes them excluded from the whole process. How can we achieve better connection between new technological inventions and people?

As broad the application field of algorithms and AI is in the world of softwares, as broad should be community in the entire life cycle of software and algorithms in the AI deployment. In this manner we need to think of data selection use, securing it is not biased, optimize systems etc.

What should be more enforced is: 'How do we make sure that results are somehow explainable or at least communicable in the real world to the community?'.  At the moment the training and the development of the AI , are devoted to the very narrow set of people which are not reflecting the broad application and implication it has. In terms of digital aspects I think we live in the society which is driven by consumerism. It relies on three pillars: convenience, selection and price.When you are interacting with social networks, business platforms or shopping platforms you don't care about the complexity but about the convenience. You don't care if this product is shipped this over three continents, three countries or three regions . You want to have it tomorrow!

The second one is that you want to have a possibility of choice which is not to complex . You want to choose between: small, medium or large.Third is one is the price – we want to have pay per use service.

We live in the world where complexity arises. We are more engaged at one point and we demand more transparency in technical complex system. On the other side, we already have such a complex system so we demand actual simplicity. As for example: we all know our data is being used in social media, but we accept it because benefit of having a free e-mail account is much higher. We also live in the B2C world ( business to consumer) where you are being turned into a product.  If you are not paying for it, you are the product. You asses even when you know someone is reading your e-mails with machine learning technology but at the same time you are not paying for this convenience of having a free communication.

Let's consider the fact: 'the more networked the plant is, the more important is cyber security'. Why is the cybersecurity so important in today's business environment and what it the role of AI?

In the drive power station, where the assets are connected and their products are being connected, they also somehow enter a space, a cyberspace where new threats arises, especially security threats. So how do you make sure that this is secured and not hackable - that nobody will bring it down or will manipulate data. We have new technologies arising and that's why digitalization and the development in AI comes along with the needs.  The needs will increase in cybersecurity. If you look on security aspects the tool driven by AI and new technologies is being used for fishing, for penetrating firewalls or infrastructure. We try to use our invasive competences to improve our defence.The majority of threats in cybersecurity are not technology based, they are human based. We need to make sure and be trained to know what are the pitfalls when engaging in digital aspects and AI aspects.

As cybersecurity threats continue to rise, the security will be more and more important in industrial sector. We need to accept that and enforce collaboration as a partner and as a network; ensure  we have environment which we, ourselves and everyone within industry engages and trusts each other.

How do you comment the rise in the frequency of cyber attacks in today's global environment?

It will increase as we digitalize, as we are more interconnected. Consumer is much more affected every day. Our smart homes we gonna live in will be more affected. I think mobility will be next aspect to ensure security. How to make sure that semi platooning, semi-autonomous drivings are not threatened by technology?

If we continue in developing technology we should in the same point invest in security. That is what we in tech community call the ecosystem. For example Siemens has pushed a Chart of trust which brings together partners who have common trusted values. We support each other in arena of cybersecurity.

Are people aware enough how the cyber threats can impact their daily life?

Human reflection and human aspects in this loop is the important. It is important to reflects that a message received is the spam and not the fishing attack and be cautious about click on the attachment. In industrial sector we cope in a similar way with virus security and software. But the question is if the current industrial players are rather ready for that? 

In home we need to do a checklist of security: do you have firewall installed, is the virus programme up to date, is it installed in every device. We do similar checklist in the industry and plants. We need this processes that help facilitate transparency and in the digital world it is considering the digital strategy in cyber security.

With the use of autonomous technology increasing attention is being paid to relationship in between man/woman and the machine. How can the trust between humans and AI be achieved on a broader level? 

Trust is not necessarily about transparency. It's more about interaction that we have.  I think it is not about how can we trust machines but are we trusting the provider, who is providing this technology to us.

We are not asking 'do you trust your mobile phone'', but you are asking:  do you trust your operator; do you trust provider in the background or the App Developer; do I know what happens with my data; do I know what do they do with my data; am I profiled; to what level? Therefore, this are the aspects that fuels and harnesses trustworthiness and will gain more attention. It will lead us also on European level with the guidelines of trustworthiness and endorse liabilities and transparency.

We witnessed lately many scandals of privacy, including recently the scandal in London where security cameras on King Cross where using AI technology to capture informations about passengers. How can we work on the issue that the security systems and controls don't cross the border of violating human rights?

In Siemens we developed a framework called 'Responsible AI' which structures the capabilities and launch in monitoring is one of this capabilities. Sometimes world is not black and white, sometimes you need to reflect on the capabilities but also on the risks. From our perspective when we do ask for monitoring that doesn't mean necessarily on the security level or profiling. We send drones over construction sites, we do landline analysis... But we also give guidelines and best practices to management of the pitfalls, the risks and what to do about privacy, faces and people.

On European level, especially B2C level, face recognition is something I'm really worried about. It's something we need to reflect more, and we see similar aspects already in the USA where San Francisco banned entire face recognition, Microsoft has put some active principles rejecting it.  The world is complex since security is very favorable aspect of our life. We need to discuss more in society and community how far should technology be pushed.

Currently we have many different AI strategies being developed. We have our European strategy being developed, USA strategy... But we always see a Chinese strategy in AI representing sort of a threat. Is China really representing threat?

There have been numerous AI strategies. Countries and cooperations think an AI strategy is in the first place a good thing because they see significant implications on technologies such as AI and machine learning in the society, people, products and processes.

AI strategy in US is driven by the tech giants and advance is really magnificent in the field of research and contributions towards AI, being more B2C centric.  I think that the strategy of China is very interesting to see. It is not a threat, it represent really high hopes, high expectations of a country. In technology they are focusing on talent, they are talking about training and education. They are talking how to actually collect and provide data and disussing about implication on infrastructure and energy efficiency. This is not necessarily a threat. You see similar approaches in Europe: European data space, trainings, networks of labs, networks of infrastructures.

I wouldn't go in the direction and say that the strategy of the certain company because they focus on education, talents and data is a threat to Europe. I think opportunities are there and if you draw attention on industrial AI and enterprise level AI than you see that the tool that is being used are also European like RapidMiner,Knime, Airfield Lighting control and monitoring system (ALCMS). First autonomous driving ideas are all concepts in history which are Europe based.

From my perspective Europe has a bit of a PR challenge, meaning we could establish better promotion of what has been accomplished and what are the future challenges. This is being crushed by public press talking about US and China.

Currently it seems there is a big rivalry in development of innovations or pushing the 5G, especially in Europe, where few countries as Germany or France are leading the way?

It is complicated. So 5G network is one of the main pillars in digitalization. 5G will enables us to get more efficiency and more latency basically in connecting devices and networks. The strategy in Europe is multi vendor specific, meaning multi vendors are providing this infrastructure. We have to catch up with certain regions including Germany sometimes in network connectivity being done.

We see on local level in companies that it is about networking and organising, not defying. We see that on global scope aswell. Europe is always very active in negotiating and having different opinions on the plate in a democratic sense. Most of European countries identified the strength and opportunities Europe as united has. This is what we need to see. I think there is a positive impact of European Horizon program. It build a solid backbone in terms of collaborating. We are engaging currently just within my little group in 4 different proposals together with more than 30 industrial players across Europe shaping the impact of the AI in Europe.

What would you recommend the future EU AI strategy should include?

Three aspects have been proposed on the European level which I think are quiet good and not everything is optimal to be honest. The first one is focusing on funding on the European cross country level,enforcing collaboration between research institutes like universities and industries so they can work on vital aspect and contribute towards European aspects.

The second one is about data: how do we cope and what is the role and European position on data privacy, data sharing, GDPR. Not everything is perfect but it is the first attempt, we need to improve in certain aspect and make a better practice on European level.

The third one is reflection on trustworthiness.  Promoting this engagement on a global scope is something we should work more.

Currently the demand of finances in research is on such a high level that most Universities can't afford research. How do we make sure that the AI research can be performed on University levels without spending a fortune? Currently all the machine learning models and research papers are devoted by large industries who can afford this research. I feel EU should broad network of universities who have ability to do AI research because currently only few universities are having financial powers to apply research.

And how to improve gender equality in the AI development which is currently dominated by men?

Diversity is important because diversity makes us beautiful. If you look at the role who drives currently technology, it is the little group actually. Diversity is not only about gender but it is much bigger, as broad our society is from country scope, language scopes, education background etc. We need to include all communities in the development from training to deploying the service.

We will need to reflect on ourselves of what our values are, how conscious we are and how unbiased we are when engaging with technology. There will be trust not happening because something is always biased. In Siemens we have support on every angle and we see importance in building networks and supporting networks. There is a say from Madeleine Albright: 'There is a special place in hell for women who not support women'. That is important not only in the AI but in the broader sense of people development.