 أهلا وسهلا بكم في اليوم الثالث على التوالي وخير العمل نهاياتها نحاول أن نكون في الوقت اليوم تأخرنا بخمس دقائق لكن نحاول أنه خلال الفحس الأولى نسترجع هالخمس دقائق هذه was a pleasure to have this discussion yesterday either in a plenary session or in a different session about the statistics and how we collected about data mining about how to take maximum benefits from these من هذا وكيف نساعد نفس each other وكيف نقوم بكتبه في حالة مجموعة التصميم والتصميم والتصميم والتصميم والتصميم والتصميم والتصميم والتصميم للمديناتنا. لذلك, سننتهي اليوم مع هذه التصميمة الثالثة. and it's number six. So it's also about measuring emergency ICT trends. And we will start with Dr. Zhen Keau Shan, Deputy Chef Engineering China Academy of Information and Communication Technology in China. So Mr. moderator, you have the floor. Thank you, Chair. إنها my honor to be invited here to be the moderator of the session 6. As you know, nowadays the term of AI, big data, IOT, and so many things, so many new words are rising very quickly, but it is very different in various countries. So as you know, with the introduction of the new technology and the new services in data life and in the industry, we can observe the fast development of the ICT sectors, which you can take a look at the new devices, new services such as e-commerce, such as mobile payment, such as digital content services everywhere in the home or in the office. So we are very happy to be here to discuss the hot topics, what's going on in different countries, what's emerging trend for this new emerging technologies when we are ready to judge its influence or its inside millions particularly for the developing countries. So we have four panelists here, invited four panelists here to discuss these hot topics. Let me briefly introduce them to you. Number one is Professor Jonas Bohr, who is also the moderator of yesterday's session 3. He is the Chair of the Department of the Media and Information of Michigan State University and Associate Editor of Tech Communication Policy. Number two is Ms. May Emel. She is head of Carol Operations and Senior Data Engineer. Number three is Dr. Nicholas Friedrich. He is postdoctoral researcher. He is coming from Oxford Internet Institute of University of Oxford. Number four is Mr. Joe Noronka. He is head of the State Institute of Market Research. Anacom from Portugal. He was appointed the new Chair of EGTI. We hope to work with you in the coming recycle. Nowadays, I will first invite Professor Jonas Bohr to give his keynote speech. So the floor is yours. Thank you very much. Your Excellency, Mr. Chairman, honorable guests, ladies and gentlemen. It's a great pleasure and honor to be here this morning and to continue some of the discussion that we started yesterday. I would like to make three points this morning that transforming emerging technologies into economic and social opportunities requires three things. We need to document the new ICT value chain much better than we currently do. We need to inform policy and governance in new ways. And thirdly, in order to do so, we need to develop new ICT indicators. Let me talk to each of these three points briefly. We already heard yesterday about four digital transformations that are currently affecting our lives and our work. The Internet of Things, big data analytics I would like to add, new computing architecture such as cloud computing and artificial intelligence. These technologies are pervasive and embedded in all aspects of our life and will become even more so in the future. They have tremendous opportunities to advance the 17 sustainable development goals, but they also boast enormous new challenges for governance and business decision makers. Of the four, currently the Internet of Things is the largest as you see in the diagram on the right-hand side. Artificial intelligence is the smallest, but artificial intelligence most likely in the 10-year period between 2015 and 2025 will grow fastest, almost 5,700% during the time period whereas the others we expect will grow at about the rate of 300%. It is important and I'd like to come back to something that Lourdes Montenegro said yesterday already briefly to understand the technological and economic forces that are driving these transformations. Some of those are beyond our control. They are how technology evolves. You may have heard about Moore's law on the quotation marks. It's a regularity about the improvement of efficiency and semi-conductors which leads to essentially half of the costs will be shaved off every 18 months approximately. But there are others in place. Cooper's law does the same thing for wireless communication services. So we see in all these technologies tremendous performance improvements that translate into cost decreases. Therefore enable ever more ubiquitous connectivity at faster and faster speeds. So enable massive growth of user and machine based data collection and essentially force businesses to really completely rethink what they're doing. The slogan here is the key word is that many businesses need to start to become platform businesses rather than pipeline businesses. It's a very different logic of doing business where we can accelerate value generation by building networks ecosystems of partners. The fourth technology is essentially enabling technologies and they enable innovations many of which we don't know yet. So it is important to create a framework that is not constraining for entrepreneurial activity. Without going into details let me just give you a couple of examples. The Internet of Things commonly looked at as the next generation of the Internet will extend our communications capabilities to devices to create cyber physical systems that will enable us to achieve great improvements let's say in agriculture, in health monitoring, in energy systems and transportation systems. Big data analytics essentially is necessary to help us understand and make sense of the vast amount of data. So it is a complementary technology to the Internet of Things to a large degree. Again enormous improvements in transportation, energy and services industries. Cloud computing again complements these other technologies in that it makes us independent of time and space in terms of accessing these services. Lastly artificial intelligence the smallest currently but fastest growing technology uses machine learning to assist humans in making better decisions and also to make routine decisions that we could delegate to machines and robots. We already have enormous advances in diagnostics where artificial intelligence for example regularly outpaces the skill of diagnosticians in predictive maintenance of plant of engines, aircraft engines and so forth. We do not however have a general purpose artificial intelligence yet. We made huge progress but we don't have a general purpose artificial intelligence technology which limits the ability to situations where we have large amounts of data on which we can train our machines to make such choices but I think over time we will relieve those constraints as well. Now it is important to understand that these enabling technologies completely change how value is generated in the ICT industries. On the left hand side you see the old stovepipe model of how value was generated. As a regulator this was a very very easy time to do your job. It was easy to use those performance improvements and really create more efficient and better services all the time. In the new system though what we have is a highly interdependent ecosystem of players many of them are not regulated and one of the challenges for creating a reliable and forward looking indicator system is how to get the data from companies that are essentially beyond the control of many of government activities currently. This ecosystem works as a fast paced digital innovation system. Those of you who may know some of the American literature may have heard of Alice in Wonderland. In Alice in Wonderland there is a scene where there is a red queen that says in order to stay in place in this country you have to move ever and ever faster. Game theorists took this up. Currently what we see in these technologies is a red queen effect. A technology race where connectivity application services and then now the Internet of Things cloud computing big data analytics artificial intelligence each are related in a virtual cycle but whether that virtual cycle can be optimized and we can harness the benefits for society will to a large degree depend on the center policy and governance how do we enable these technologies to work. So let me now briefly look at what the implications are and what we need to know to inform policy and governance. The first thing I would like to emphasize here is that ICTs are neither good nor bad. I know this is a risky proposition in an audience that inherently believes in the benefits of technology but I would like to warn you whether we can utilize those benefits depends highly on the social the political conditions in which these technologies are unable to unfold. We do not know in the second point I would like to emphasize the full effects of these digital transformations. We have very very fantastic versions of how the future could be but we also have dismal versions as to how it might end up in ways that we don't like. So ICTs despite bringing huge benefits for the social development goals might also bring challenges such as unemployment. Robotics, artificial intelligence can replace human labor very very easily. We have already discussed next generation digital divides where the great advances that were made for the past three decades in bringing many of the lower and middle income countries to better infrastructures might actually be jeopardized because now the race moves to those next generations of technologies. Lastly, technologies of freedom such as mobile communications the internet could also be technologies of surveillance and control and we do not want them to be abused in this way. To protect society we need sufficiently intelligent models to control these technologies and to shape these technologies in ways that benefit society. The main thing the third point I would like to make here is that there is not one single best model. You all come from countries at different income level at different infrastructure deployment levels and the best policy needs to be appropriate and customized to those conditions. This is something I think that is different from what we thought 30 years ago 20 years ago this is not the best model how we should do things. So this is encouraging because these technologies provide the flexibility to correspond to them and to respond to them in unique nationally appropriate models. There will be probably three or four or five types across the globe but appropriate models will differ from country to country. The good thing is that many of the tasks and many of the accomplishments that were achieved in the past decades go in the right direction because to have a network infrastructure and the deployment of services in place is a very important precondition but there are additional things that are needed by policy and those might challenge ICT regulators and policy makers because you're historically perhaps not used to work with other policy makers in education in economics ministry and so forth. So we need complementary user skills. This is not just education in computer science or in data analytics or to have a digital work force these are important but we also need a new digital mindset that enables business people to work as platforms and that enables individuals to take advantage of the great opportunities of software. One of the great pluses of these digital transformations is that they become easily usable but we need the right mindset to use them. Big data analytics as we will see in a moment offer so many new opportunities that can be done by many many people but we need to have the right mindset to work with the data and ask the right questions. Lastly we need policy responses that enable entrepreneurship. This is a big challenge for many regulators because regulation has historically been construed as a constraining force. Regulation initially started as limiting monopoly power and market power and to make that mind change from limiting the abuse of market power to enabling entrepreneurship and innovation is a big change that needs to be made. Lastly let me say a few things about how we can create a system of indicators that will inform those important issues. Again I would like to make a couple of very important points without really going into detail. I think we can only use and the experience shows this. These technologies to the best of society if we use what is called human-centered design approaches. We have to start our analysis from the effects on individuals on humans and organizations to use these technologies and then work backwards to understand how we can best harness them. Doing this requires continuously updated and reliable information and we can of course do this as governments that collect data and this is a very important part but we can also trust as was already discussed yesterday machine generated collection of data and of data processing. Now what's the role of the public sector in this new more diverse framework? I think there's three important ones in addition to be a collector of information. First of all we need to come up with new frameworks to collect standardized information that is of broad importance. This is information such as as the indicators that collected by the ITU and others that everybody needs to know and everybody needs to have but that cannot be all. I think we need also to rely on machine learning the collection of data from the networks through open algorithms and open data not all of this data is necessarily needed for everybody so we have to make crucial decision what's the accessibility of this data to help us make our decisions. Lastly I think there's a very important role of the public sector to be a curator and an archiver of data a facilitator of data generation rather than the data collector itself through open repositories and others. We need in order to achieve these goals indicators but we also need better ways to model and understand what is happening. And here the most important thing is again to focus on what's the goal. Indicators are not just indicators by themselves but they are indicators to help us achieve certain purposes such as the sustainable development goals or other economic and social goals. So that's the starting point. Then we have to ask ourselves how can we create frameworks that help us better inform decisions to lead to those goals. And there are certain steps that we need to undertake to develop such an enhanced system of indicators. One is we need to create better direct indicators of those emerging technologies. One challenge there is that these technologies are embedded. Artificial intelligence is widely diffused. Big data analytics is widely diffused. It's difficult to count devices like within the number of access lines. So we have to come up with new ways of measuring that could include in addition to the number of devices where this is feasible what's the percentage of the installed base let's say in businesses in government offices that uses technologies with certain capabilities. What's the the share of revenues that is accrued with these activities. Secondly, we need to have additional information on basic services and software such as M2M connectivity the availability of big data analysis software and lastly we have to augment our indicators with new indicators on the applications and services sites such as what's the percentage of businesses that use cloud solutions or artificial intelligence In addition though to these emerging technology indicators we need to better understand the enabling conditions and where we are with regard to meeting those enabling conditions so we need data on network infrastructure again this is already collected so nothing new here but we need better data on skills we need also better data on policy arrangements for example what's the number of how many cities to have open data policies how many government agencies to have open data policies what's the share of unlicensed spectrum indicators like this lastly we also need to have better understanding of our outcomes how are these technologies related for example to income how are they related to unemployment and the quality these are issues that go beyond the immediate remit of our telecommunications regulators and telecommunications policy makers therefore you need to work with others last but not least let me mention that one illusion needs to be eliminated and that is that big data speaks in and of itself it does not in order to make sense of the data we need explanatory models we need predictive models and most importantly to help decision making we need prescriptive models that help our decision making now I know that you cannot read this but I encourage you that this is a table that is taken from the new report that provides a matrix of these indicators that we can use to substantiate these information requirements in many many cases this is information that is already captured that is already in the indicators that exist so we need to augment them with those that we need for the future so let me conclude with two recommendations specifically one is short term in the short term I think it is important to build on existing initiatives such as EGTI the partnership on ICT for development to develop an enhanced system of indicators that enables us to document better those four emerging technologies in the medium term though and I was tempted to say in the long term and then I realized there is no long term in this fast pace the technologies the medium term is relevant here in the medium I encourage you to think boldly and develop a new system of digital national accounts along the lines of the matrix that I just showed you that enables us to fully comprehend the pervasive and massive impacts of these emerging technologies and make better technologies going forward it is very exciting to be in this field those days you might think it is daunting the challenges are so big but it is exciting these are enormous opportunities that occur only once in a lifetime and I think we are here and we are starting a process here to harness those and in order to do so successfully I always say you need two things you need hard heads because you really need to think through these complicated dynamic issues with a clear and elliptical framework and you need soft hearts because without soft hearts you will not have the compassion that will help us realize the enormous benefits of these technologies thank you very much thank you Professor Bohr you have again a very influences and I think many of the many participants willve be interested on your recommendation and proposal but we we will be a little bit later so we have to speed up the coming process so I will invite you the 2nd speaker يأتي من الأكتبار، ميس أمو، لكي أخبرك، فالفلو هي أكتبارك إنه مرحباً لكي هنا، شكراً جميعاً باستخدام الإنطلاقية. في العام وحسب الثلاثة، لقد always been the ultimate goal of computer engineers and scientists to develop machines that can model the complex decision-making process that a human brain can naturally do. لذلك، لا أسأل أن الأسلاحة كانت تصميمة بها by the brain and its neurons as the most advanced computational organ in the natural world. الأسلاحة كانت تصميمة بحيث تصميمة بشكل كامل or artificial neural networks that have massive computational capabilities. تقوم with those networks with other computer technologies such as computer vision, تقوم with machines to understand their surrounding environment, make decisions, and take actions. ونحن now we're living in a world surrounded by those hyper-connected devices, devices that have massive cognitive abilities. most of these devices are very smart making decisions yet these decisions are very transactional. They go linear in one direction. دوائس that have very high IQ but no EQ. It's turning into a world that's devoid of emotions. And here comes our team's bold vision. We imagine that in the next five years, دوائس will be emotionally aware. We need to bring empathy into the way we connect to devices and back into our lives. دوائس will be able to read and sense your emotions and respond back accordingly just the way an emotionally intelligent friend would. From here, let me give you more background about my organization. أفكتيفة was font out of MIT Media Lab in 2009 was co-founded by two leaders in the field of effective computing Dr. روز잖아 بيكارات the publisher of the effective computing book and Dr. روندو少بي هوהو بيو가요 or current CEO. و يمكنني أن أخبركم أن أفكت سيوة is a leading pioneer in Emotion AI, but why emotions? Why Emotion AI in particular? Emotions influence every aspect of our lives, from our health and well-being, how we connect and communicate with one another, the purchases that we do, how we do businesses, every decision we make, big ones and small ones. Yet being surrounded by all these devices, we are now spending more time connecting to these devices and through these devices, more than we connect and communicate with one another. I think most of us can relate calculating the number of times we spend with our laptops and with our smartphones. How many of you have watched the movies Ex Machina and Her? Good. What's fascinating about these movies that machines featured were not just super-super-smart, but they were highly emotional intelligent. And because of that, they were able to get the guys in the movies to like them, persuade them and motivate them to take actions they wouldn't have taken otherwise. Of course, the movies took darker turns that we will forget about for now, but it's been shown over and over that humans that have high emotional intelligence are more likely to succeed in their lives and more likely to get liked by others. And interestingly, this translates to how we connect to machines. Imagine a word that doctors can objectively measure your emotional state just the way they measure other vital signals. So when you step into a doctor's office, they don't ask you what's your blood pressure. They just measure it. What the golden rule in mental health is filling out surveys. What if we can detect early depression and Parkinson's behaviors, avoiding more suicides? What if your car can sense that you're angry and frustrated and makes your breaks more responsive? These questions and many more in different domains drive our team in affectiva to build technologies that can read and respond back to your emotions. And the starting point was the phase. The phase happens to be one of the most powerful signals that we all use in our daily communication. Everything from frustration, confusion, curiosity, and yeah, different emotions. It contributes by 55%, while 38% is how you say the words and only 7% are the actual words. So when you're texting, only 7% of all your emotional and social states are communicating while the rest are just lost in the cyberspace. Facial signs have started over 200 years ago by this guy called the Shin who used to electrically stimuli people's facial muscles in order to study their movements. This is very painful and we don't do that anymore. We use computer vision and machine learning. Fast forward 100 years later came and Paul Ekman and his team who started giving each facial muscle movement an action unit name. So for example, Action Unit 12 is when you put your lip corners to the side. It's a contributor of a smile. Let's try it around everyone spreading more smiles. And it's an indicator of a positive emotion. While AU4 is when you draw your brows together forming all those textures and wrinklers, we don't like this action because it's a huge indicator of a negative emotion. We have 45 of these Action Units and they combine in different ways to form different emotions. Yet, teaching a machine to learn the different between these Action Units is not an easy task because these actions can be very subtle. They can happen very fast and they combine in many different ways. Take for example the smile on a smirk. They look somehow similar but they mean totally different things. A smile is positive while a smirk is often negative. And it is important for a machine to learn to detect the difference between them. So how do we do that? A machine is like a student. It needs to be provided by examples so that it can learn about the subject it needs to learn about. So we provide our networks by thousands and thousands of examples of people who we know are smiling. People with different shapes, different ethnicities, genders, different age groups. And we do the same for smirks. We pass those examples into deep neural networks and using deep learning techniques. The networks are capable of extracting the differentiator feature between a smile and a smirk. And it will learn that all smiles have common characteristics while all smirks have subtly common different characteristics. So the next time it sees a new face it's like a student sitting in an exam. If the student is studied well like a machine being provided by unbiased examples enough examples, it will learn to detect if the face is smiling, smirking or doing any different actions. Now we have classifiers that can detect 23 of these action units, 7 emotions and 3 appearance matrices. Gender, ethnicity and age. We've just added speech into the mix. It's just a couple of months old now and we can map how you say the words to an emotional state. Extracting features like tempo, pitch, energy and mapping all those features into different emotional states. Now we have 4 matrices for speech. Arousal, laughter, anger and gender. Till now we have amassed 60 billion emotion data points. It's the word largest emotion database. The data points are collected from 6 million faces analyzed in 87 countries. It is very important to have this spontaneous type of data and so most of our data is spontaneous collected in the wild. Lots of ethical questions revolve around data collection but in short one of the affectivist core values is to maintain the privacy of people so we don't record or analyze people's facial expressions unless they opt in and agree to share their data. The diversity of the data that we collect allow our algorithms to work and detect people's facial expressions from all around the world. It's not just about the diversity of the data that we collect it's about the diversity of the team. We have two affectiva consists of two team members sorry, two teams in Boston and in Cairo where I come from with members from more than 10 countries Bangladesh, Netherlands, Iran, Taiwan, Canada different parts of the US and Egypt and many more. Mindful that a diverse team is more capable of building diverse technology, diversity breathes creativity. It's mind blowing the number of domains that Emotion AI can be integrated in quantifying something as personal as facial expressions and emotions. Let's take a couple of examples in different applications in the health domain and education. In education, the MIT Media Lab has integrated affectiva SDK in TAGA. TAGA is a social robot platform that supports educational interaction with children. There was an experiment conducted with children learning Spanish as a second language and the group of children was divided into two groups. One group worked with TAGA and the other didn't. The group who worked with TAGA was shown that they learned more words in shorter duration because TAGA was able to read the emotional state of students everything from the confusion level, boredom, frustration and it adapted the content accordingly. While in the health spectrum, Brainpower partnered with affectiva building the world first augmented reality smart glasses system empowering patients in the autistic spectrum to understand the emotions of people that they are communicating with them and teach themselves to more cognitive and social skills. The merger of IQ and EQ and tech nowadays is inevitable. Emotion AI has the potential of democratizing the access to services like education and healthcare. Closing the gap of the socioeconomic state but this will not happen unless we as leaders in the space of AI ensure that the technology is being used in the right direction. So from here I'd like to call policymakers in different countries to prioritize the development of AI and increase the circle of machine learning scientists and applied AI developers in their regions. From our side, it's our goal to emotion aware every device in the internet of things by providing an emotion chip that will track your mood throughout the day and give you personalized recommendation by the end. There is a lot of stuff to be done in the emotion AI field and that's why we have packaged the technology into a free SDK that's supported by different platforms so that developers from all around the world can download the SDK integrated in different fields different domains and the sky is the limit. Emotion AI will not just transform how we connect with devices but how we connect and communicate with one another. Thank you. Thank you Ms. May. You have given us a very interesting presentation of the emotion AI that is applied in daily life. I totally agree with you. It's not a device by itself but it's widely used in the field of daily life and our office work so it can be applied everywhere. So we can discuss the topics a little time later. So let me invite the third speaker Dr. Nekras Fidwick from Oxford University to give his speech. So the floor is yours. Okay, thank you. Thank you very much for inviting me. It's an honor to be here. I want to spend the short time that I have to alert you to some stark divides in the global digital economy which I think have important implications for the topic of this session. To do that I will use studies by my colleagues at the Oxford Internet Institute often focusing on Africa and these studies highlight in particular the extreme geographical divides in digital production with the United States and Europe dominating and no signs of low income countries closing the gap. Divides in digital production are not often acknowledged because we tend to focus our attention on adoption. This year 2017 I'm sure you're all aware is the historic year in which half of the world's population is going to be online. This map even if it is from 2013 I said there are now many more far more internet users in the so-called developing world compared to the developed world. However adoption and other indicators of usage are not at all good measures of digital divides. They're not comprehensive. 100% of people are using the internet as inherently the maximum that any country can reach. So for adoption since the north has widely reached saturation catching up of the global south is a long conclusion at least if we assume decreasing cost. But when we shift our gaze to digital production we see a drastically different picture. Let's start with use a generated content. This map shows you that in 2013 the number of Wikipedia articles about central western and southern Europe was greater than the number of articles about the entire rest of the world. In fact not only do we see drastically more content creation but in the north the north also dominates the south in content creation about the south. This map shows us again with Wikipedia edits this time. There is obviously a language barrier from people from non-English speaking countries to contribute to online content about their home regions. But the stronger cause as you see here for the lack of participation is the nation's income level and connectivity history. You can see here for example the language language information that is available about them. Next what happens if we define digital production more narrowly? The data paints a similar picture one of stark divides between the north and the south. In this more recent study we use data from Github which is the world's largest repository of code and a database of website domain registrations. We then shows academic articles as a comparator so this measures traditional knowledge creation. Academic articles we thought were interesting because they are notorious for being extremely skewed towards the global north and to English speaking hubs like the US and the UK. So this study showed that even when taking a comparator like academic knowledge production which is already extremely skewed digital production is actually even more skewed in favor of the global north. So you see here in these charts that the global north accounts for about 80% of digital production measured according to these two measures while Europe and the US actually only have 20% of the population so we see this 80-20 divide. In particular you see the jump for North America obviously due to the US jump upwards and Asia and the MENA region lose. Digging deeper into this and going to the country level we then start to see clearly the absolute dominance of the United States. Here we're again using github data. This graph shows the export-import ratio of followers on the github network so this is a kind of trade surplus of software developer influence. There are a lot of interesting aspects to this measure which I won't have time to discuss here but the clear message is that the US has a very particular role as a global influencer. With almost 2.7 times as many followers from the outside of the US into the US and from within the US to the rest of the world. We did a similar analysis that I'm not showing here with data on stack overflow which is an online forum for software developers and the results were basically the same. We can also see from this github follower data set that there's very little south-south exchange. For every African region we look at African regions. Here for every African region about a fourth of followers goes to the US a much smaller but still a significant percentage goes to Europe and only tiny fractions go to Asia, Oceania or the Middle East. We also see that only north and southern Africa have significant intra-regional followership. Likely because they have larger economies and more established tech industries to begin with. In east-west and central Africa software developers almost exclusively follow others outside of their home region. And here's the result of that in the economic terms. At one point in 2016 the five largest companies in the world were the five global digital technology companies Apple Google, Microsoft, Amazon and Facebook. At that point their combined valuation was equivalent to the nominal GDP of the entire African continent. At the time investors felt that Google alone was worth as much as the GDP of Nigeria. For the time being when this trend is only increasing in 2017 none of these companies have decreased in value. Two Chinese contenders have entered the fray, Alibaba and Tencent but digital companies from countries other than China and the US are nowhere to be found. Now I'm aware that company valuation is not an entirely meaningful metric for development and that this is of course only a snapshot of the absolute top end of the global digital economy but this does illustrate to me that the value that is being created through the enormous growth of the global digital economy is captured only in a very narrow subset of places. Specifically Silicon Valley in the US few places in Europe and a few places in East Asia. So where do we go from here where does this picture leave us let me revert my somewhat pessimistic talk a little bit and tell you that there is hope. I base this statement on information which is not well qualified which we have so far not been able to quantify in any meaningful way but from our extensive qualitative work we do know that a lot of technological innovation is happening across the African continent again I'm taking the African continent because that's the focus of our research and that this has important positive development impact. So this new economic practice digital entrepreneurship as we've come to call it is unlikely to have vast development effects in the short term or in the mid term but it does have important long term potential. And this is because it's one of the very few ways for developing countries to control their own technological destiny. It's not about protectionism but it's about creating the conditions in education and training and the integration of markets to enable local enterprises to adapt global ICT trends in ways that work locally for them to grow in scale but also for them to create new pathways of innovation that positively impact lives regardless of the financial bottom line. The key is that digital entrepreneurship can be a way to create and capture value locally. So let me close by highlighting the implication of this for you for this audience. I'm calling upon you but also upon us as researchers to measure digital production and digital entrepreneurship specifically much more carefully than we've done so far and I'm aware that this is a challenging task but I'm happy to explore this challenge with you now in the Q&A session or afterwards. Thank you very much. Okay, thank you. Dr. Neckloss you have just given us very detail about digital divide and we have to do something to bridge the digital divide nowadays you can easily find the new kinds of digital divide particularly in the field of new emerging technologies so I think many of our participants will have interest on the discussion how we can do something particularly for the regulators and the ministries and you also touch on the topics of the digital skill digital skills is not only related to the industry but also to the digital skill of the human of the people so we all invite the last speaker of our session the new chair of the EGTI from the particular Anacom Mr. Joe Norlman the floor is yours Thank you very much for the invitation to be here my presentation is on measuring the internet of things I will start by trying to define what is the internet of things the internet of things is a network of interconnected things or devices most of them are connected to the internet and use standard communication protocols they have certain capabilities they are sensors or actuators or they are able to be programmed and they generate a large volume of information these devices and information can then be packaged into services in practice what this means is that we will have billions of devices that are interconnected and will generate big data and will cover all areas of activity for instance we'll have connected homes smart farming industry 4.0 smart cities connected hails smart retail etc some people are calling the IoT the new electricity and because of its relevance for our lives and our societies the public policy and regulation must take the IoT into account Professor Bauer already explained in a more detailed way why this is important at this stage I would only like to say that telecoms will be the infrastructure of the IoT and so public policy makers and regulators must make sure that telecoms does not become a bottleneck or a barrier to the development of the IoT so issues like standardization interoperability, numbering addressing and also coverage accessibility availability universality all these issues must be taken care of by regulators on the other hand the telecom sector itself is being transformed and so for regulators market analysis becomes more difficult Professor Bauer stated that life for regulators was much easier before and in fact when I started working we had about four or five operators in three services and now we don't even know how many operators we have and or where they are so life is more difficult we have tight oligopolies we have our operators entering other markets and other entities entering telecom markets some operators are based on other countries so it's difficult to enforce national laws and there are new issues like net neutrality for instance so in order to address all these issues it makes sense to start collecting data on IoT what kind of data should we collect okay in the short term I think there are two sets of data that are relevant coverage and usage concerning coverage well we have to collect data on mobile coverage since a lot of IoT applications will be based on mobile networks the good news is we are already collecting data on mobile coverage for 2G 4G and there are international groups that are at the moment studying how to measure 5G coverage then there are other areas of the IoT like LPWA LPWA stands for low power wide area applications we must make it's feasible or necessary to collect data on LPWA we must also collect data on fixed coverage because fixed coverage is important to the internet of things first of all 90% of wireless traffic is supported by fixed networks and even 60% of mobile traffic is offloaded onto fixed networks and then we have a large share of the internet of things which are the short range applications which are supported on fixed networks so it is important to collect data on fixed coverage the good news is we are already collecting data on fixed coverage regulators, ministries and the ITU for instance are starting to collect data on fixed coverage there are also other infrastructures that are associated with the internet of things for instance internet exchange points data centers, cloud we should also study if it makes sense if it's possible to collect data on the infrastructures then we have the issue of usage and in order to determine exactly what we are talking about it makes sense to see what kind of applications the IoT covers we have basically three groups of IoT applications we have wide area critical applications which require ultra reliability availability low latency and high data throughput an example of this is of course autonomous or self-driving cars then we will have wide area non-critical applications that require high connection volumes low traffic low energy consumption and low cost devices and an example of this is fleet management and then we will have the short range applications this will typically involve connections with a range 100 meters or less and some smart home applications will be short range applications now these IoT applications are supported by networks for instance wide area critical applications may be supported by 4G and 5G networks wide area non-critical applications will be supported by other cellular technologies and also by the LPA LPWA proprietary technologies and then the short range applications will be supported by wireless and fixed networks some of this networks also support other types of applications what are the data sources for all these applications and networks well for cellular applications for cellular networks we can rely on supply site indicators and our let's say traditional data suppliers mobile operators for LPWA we might try to collect data from these operators although there might be some issues we'll see that later and then there's a short range applications in here we cannot rely on supply site data because it's simply not available we have to rely on alternative data sources device vendors the internet as data source user surveys or machine generated data as we have seen in the previous applications in previous presentation now I will show some examples of these new indicators and data sources later on but before that we should look at what kind of indicators we should collect and what are the challenges associated with collecting these indicators for cellular technologies we will collect mostly M2M type indicators we are already collecting M2M type indicators but the internet of things will demand certain refinements for the existing indicators for instance it will probably make sense to split M2M by technology or network because different networks will involve will supply different applications it will probably also make sense to collect M2M indicators for specific applications like connected cars or smart meters depending on our policy needs then there are certain technological developments associated with the internet of things that will impact our indicators and that we must take into account for instance there's these seems or simultaneous or multi-homing connectivity meaning that certain devices will have the possibility to have one or two or even more connections and this will impact our indicators because it might lead to double counting and so our definitions must be adapted to take this into account and then there's the issue of mobile penetration we must split or reverse mobile connections from machine to machine connections otherwise we'll get 500% or 600% mobile penetration which is meaningless then for LPWA networks we can collect number of devices clients, traffic or revenues the problem is some of these operators are transnational operators that are not based on our countries and that might make it difficult to collect the data for the short range applications well we can collect some people are collecting a number of devices type of devices or type of applications and what we must take into account is that this is always going to be only a partial subset of the universe of short range applications but on the other hand when we use user surveys to collect data on these applications we must also take into account for instance that consumers may not be aware that they have these devices or applications it might be embedded on other services or devices now this is an example of LPWA we at Anacom in Portugal collected data from LPWA providers this involved determining which were these operators but this was sort of easy because of the licensing procedures and we collected number of devices, clients, traffic and revenue and conclusions were as expected that there are a significant number of devices low volume of traffic per device and low number of corporate clients we have here an example of a consumer survey on connected cars this was done by Ofcom they asked consumers what were the types of devices and services that they had on their cars for instance automated driving features or in built infotainment so this is possible and feasible in that it's being done by some countries yesterday we heard from the OECD that they were looking to develop a model questionnaire on the IoT so this is clearly a way to go then we have internet as data source sources for instance a showdown which is a search engine collects number of connections and location of devices that are connected to the internet so this is a new area that we explored to measure the internet of things so in conclusion what I'm saying is we should continue to collect data on fixed mobile coverage and develop 5G coverage indicators we should start computing our mobile penetration separating machines from people we should try to refine our M2M indicators and mobile indicators to take into account that it makes sense to split indicators by network or technology we might want to start collecting data on specific applications and we must also investigate what are the effects of ECMS or multi-homing on our existing indicators and we should clearly explore alternative data sources LPW providers device vendors internet sources or machine generated data as was mentioned before and lastly we must adapt our consumer or user service to the IoT thank you very much thank you so now we have ended our 4 speakers contribution on the emerging technologies in ICT sectors so we have only 5 minutes left for the discussion so let me give you a proposal that we can go back to the topics of our sessions I suggest that we can have some idea or comments or questions on the speech of the 4 speakers you can evolve around the topics of what is the context for the emerging ICT trends a new digital device arising from the uneven spread of these innovations worldwide how can they be tracked how can effectively evidence-based policies be designed to accelerate the adoption of these emerging trends and ensure that nobody is left behind but you can also have the other topics so now the floor is open we may have 2 or 3 questions or comments from the delegates okay you have the floor you have the floor you have the floor thank you thank you do you have any feedback to help them okay so any other comments or questions asking for floor okay so what is actually to Johannes Bauer thing is that his advice is about in the 4 components around the digital transformation the big data one is very important one but what the underdeveloped countries is struggling with initiating big data project we are struggling around the big data and then how to capture them and how to visualize them for the policy makers we have challenges around the expertise we have challenges around the technology and how to manage them because these things are very important for the digital transformation so what is your advice for the underdeveloped countries where we are struggling and we are creating lots of social media and what we are using for decision making at the government level and we have facilitating facebook to use for the government employees what types of discussion is there how decision is taking and what types of visualization should come up from all those discussion around it thank you give some feedback well let me answer briefly which is very complicated these are both very very big and very very important questions and I think we will need much more discussion on those and I cannot really capture the significance of these issues in the short statement but to the first statement from Tunisia you are absolutely right it is important to expand what we consider relevant indicators to capture also those effects on society because that's really the metric or the effects on this sustainable development goals because in the end an indicator is only useful in relationship to a goal to a purpose and I think that needs to be really going forward the second question from the delegation of Bangladesh in the long run the biggest challenge of big data is probably to make sense of big data because it's much easier to harvest the data and visualize the data than to explain what is going on and the capacity building that needs to happen is to develop a sufficient number of data scientists and people with mindsets who are capable and interested in analyzing the underlying processes that are going on because we know much less right now about how to use big data for prescriptive analysis and decision making than we do for other purposes in that context I think it will be important to emphasize open algorithms so there are transparencies to how we make those decisions of proprietary algorithms that make decisions for us and we have no clue why this is happening okay, do we have any feedback to the question from Bangladesh just a short note on both of those questions to concur with Johannes basically that I think what matters is how the effects of these technologies trickle down to permeate into the lives of people and the realities of businesses and I guess that's also what I was trying to do in my talk that these technologies are not globally homogeneous doesn't exist everywhere in the world in the same way and so I think much more work can be done with representative surveys maybe with actually engaging with stakeholders on the ground how do these technologies enter your lives or your business processes because often what we see in our research is that the effects are actually and realities are dominated by other things or there is a lack of control over these trends so these trends are kind of imposed from the outside but there is a lack of control so I think this is where the most work has to be done how do these technologies impact the actual processes and the actual business processes and the actual lives of people okay, do you have any further comments to the questions do you? yeah okay because the timing is very limited so I have to end our discussion on the topics so at the end please allow me to invite you to give our warm applause to the wonderful presentation and the answer to these very hot topics thanks so I have to hand back to your majesty thank you very much it's always a pleasure to hear the analysis of the future and how to build this mariage between the technology and life technology and society technology and business technology and our family life and our own behavior so good thank you very much for all the panelists and for all the audience