 Thanks for joining us here in Geneva for the AI for Good Global Summit 2018. My guest is Thomas Vigant. He is Professor at TU Berlin and Executive Director of Fanofa HHI. Thank you for joining us, sir. Thanks for having me. So, Thomas, you were involved with the creation of the summit, weren't you, as one of the early organizers or people helping with launching the summit last year. How do you think it's going? I think it's going great. We have, the summit has grown beyond our expectations. The dialogue that is happening between the UN agencies, the social entrepreneurs, philosophers, engineers, that's a great dialogue that we're having here. It's really become a global multi-stakeholder platform to engage people. And this year, I suppose, the difference is it's all about practical applications, isn't it? Yeah, we have tried to progress the conference towards also creating projects that people hopefully will work on between now and hopefully the next one. And hopefully something practical will come out of it. As an engineer, I very much would like that. Okay, so what direction do you think AI is taking at the moment? Are we having the right conversation? In part, yes. In part, no. I think it's important to understand that behind AI, there are algorithms. And these algorithms have their limitations. They are broadened by mathematics, and proven by mathematics what these algorithms can do. And there are lots of things they can actually do well because of the large amount of data we have available and the compute power. But there are also things they cannot do. So we should be, how to say, practical and reasonable about what AI can do for us. Because I'm tempted to ask you, what is AI after all? It seems to be a different thing to everyone who's coming to the summit. How would you describe it as an engineer? As an engineer, it's a tool to solve problems that engineers can use in their work. So you can try to use AI for 5G, for example. We have created a focus group at the ITU, 5G in machine learning, where, for instance, traffic is routed through machine learning algorithms in 5G, or where you are tracking participants for a MIMO antenna. So you could use it to improve medical diagnosis. But nevertheless, it's about an algorithm, or multiple algorithms, that work on the data and provide you a classification or regression result that you can then use again for the technical system you are designing. And can you use it to fix the world? What can you use AI for, or can you not use AI for? I think the parts of the world that broke in small steps, and the way to fix them is also in small steps. And a lot of engineering progress in the past has been through increasing efficiency by the use of more resources, use more water for agriculture, use more energy for communication, et cetera, et cetera. We actually need solutions that solve engineering problems by using less resources. So make something more efficient when it uses less resources. And then also observe another aspect, which is the Jevons paradox, which is as of 150 years ago, Jevons had the efficiency by which a resource is used more efficiently, then the overall consumption of the resource will increase, meaning if you make more fuel-efficient cars, the overall consumption of gasoline will increase. So we have to be careful by the efficiencies that we are making that they also lead to an overall improvement when it comes to the resource consumption. Thomas Vigant, thank you very much. Thank you.