The following are the outputs of the real-time captioning taken during the Thirteenth Annual Meeting of the Internet Governance Forum (IGF) in Paris, France, from 12 to 14 November 2018. Although it is largely accurate, in some cases it may be incomplete or inaccurate due to inaudible passages or transcription errors. It is posted as an aid to understanding the proceedings at the event, but should not be treated as an authoritative record.
***
>> MODERATOR: Good morning. Good morning, everyone. Thank you so much for joining in this workshop. We have a very small room, but then we are also able to have a more cozy discussion.
My name is Marta Capelo from ETNO. I will be co-moderating this session with my colleague, Ricardo predraw za. We also have our online moderator, Kristina Olausson.
We are bringing 5G, IoT, and AI, addressing digital inclusion and accessibility. We know this is a really broad topic that you all have already been discussing in other issues. But what we really would like to steer the discussion here, see where these technologies can actually help us to bridge the gap, to bridge the divide, or what are the risks of accentuating the differences in society.
We have a lot of opportunities. We think this should be embraced. With you we also have a lot of risks.
I am not going to be very long in this introduction. I just want to highlight that we are very privileged to have a very diverse panel of speakers of all of the stakeholder groups and regions, and we hope to actually have short interventions so that you can engage. So I would really appreciate if you could take your notes on the first half an hour of interventions and really, really ask questions and bring your points of view to the table.
And I think that's all from my side. Carlos?
>> So hello there. Good morning. I am Carlos Affonso. I am Director from the Institute of Technology and society, ITS, Rio, from Rio de Janeiro from Brazil. It's our pleasure to be part of this panel.
As you can imagine, this panel is a result of a merge of two different workshops proposals, and we want to make this as an experiment of serendipity in which I think this merge can provide us with a very interesting opportunity for us to look at the issue of new technologies or new ICTs through the lens of inclusion and accessibility. As Marta has said, we have quite a challenging task ahead of us since we are dealing with 5G, IoT, and AI, but I think it's going to be really interesting to see what are the challenges and what are the obstacles on those different technologies concerning accessibility and inclusion.
From my side, just to say that it's a pleasure to co-host this workshop. ITS is currently spearheading the Secretariat of the global network of Internet and society research Centers, which is a network that encompasses more than 80 centers dedicated to do research on different fields on Internet and society. AI is one of our most important topics to research right now. So we are really happy to be able to share with you some of our current research and some of the members of the NOC are here to share their expertise on those topics. So let's make sure we have an engaging conversation.
Just for a very quick housekeeping, our speakers are going to be kept short on fiveish-minute presentation, but I will hand over to Marta so that we can begin the first round of presentations that we have for this session.
So Marta, if you want to continue.
>> MARTA CAPELO: Thank you very much, Carlos, for those remarks.
So we will start with the accessibility angle, then we go more to the AI angle, so I would ask Alexia Gonzalez Fanfalone, telecoms analyst, the digital economy policy of OECD, that is going to give us an introduction on what are the main issues and opportunities that 5G can bring to accessibility.
Thank you, Alexia.
>> ALEXIA GONZALEZ FANFALONE: Thank you very much. It's an honor for the OECD to be here and talk about this topic. I must say that there's a lot of hype around 5G network, so I am going to try to be the realist in the room and say what are the implications really for network deployment, and what are the experience of our OECD countries to date and where this might be going.
So that's why the title of the presentation is The Road to 5G, because we are still in progress, and we hope that there will be more developments in the upcoming two years.
Next slide, please. What is the promise of 5G? Three main use case scenarios -- enhanced mobile broadband. The ultra reliable and low-latency applications, so this is critical IoT applications. And massive machine type of M2M, which is more like sensors and different type of IoT.
So what is different from this new generation of wireless networks? This is an evolution from 3G, 4G, now 5G, but what is different is the first standard to be conceived with IoT world in mind, where millions of connected devices will be handling different network requirements and an exponential increase in data that we will see in the years to come.
Next slide, please.
But what are the implications of this? So I am a telecom specialist, so I will just talk about how networks work today. You can see the macrocell in the top left. This is how 4G and networks have been deployed to date. 5G will also use macrocells, but there's a new trend that we've seen. The stand-alone standard still has to go through the second phase. The first phase was completed in June 2018. We do know that one trend that many of the initial deployment and trials is smaller cells. So when you see down there the cell site in a lamp post, that's the Verizon network in Boston, and if the lamp posts are the new cell towers, this means that we need much more fiber back deployment. As we have been saying for many years, wireless networks are an extension of fixed networks, so fixed infrastructure -- and this is important for many emerging economies -- fixed infrastructure has to be deployed deeper into the networks to have lower latency and better quality communications. So when we talk about digital divides, we have to think that it's not only access to basic Internet, it's access to high-quality Internet that will provide firms with the productivity enhancements that they need in order to compete in this global economy.
So in the right-hand side, there's other type of technologies, for example, P cells. This is a company, at the miss. What is interesting is these are fiber cables in the posts, and they create personal cells. So it's more like wireless and fixed become more similar. We don't really know where it goes. Companies like Verizon in the U.S. are saying that they will be deploying in the next three years 36 million miles of fiber in order to have functional 5G network. This is a significant investment, and it might shift the competition between fixed and wireless carriers in many countries, which makes us regulators -- because our constituents, as regulars of OECD, to put ourselves in the feet to what are the challenges in order to tackle this, to foster incentives, to keep on deploying networks, but also to have pro-competitive regulation so we have a level playing field.
Next slide, please. So what does this demand? What is it, this network densification that we keep hearing in the news? So more fiber backhoe needed for 5G. There will be new demands for wireless infrastructure, notably new IoT applications. For example, Intel has come up with this number that one fully automated vehicle in one day will produce 4,000 gigabits of data. Just to put this number in perspective, this is equivalent today at the average mobile user in the OECD 5,000 users. Okay? So what does this mean? This means network infrastructure has to be upgraded. There will be high demand on telecom operators, on regulators, on policymakers, to make this reality. And some countries are forging ahead, and in some regions, rural and urban, there will be some gaps. So we have to think about this.
Next slide, please.
This is the gigabyte per month today in OECD countries, the consumption for mobile broadband description. So as you remember, the first slide the first usage case for 5G is enhanced mobile broadband. But that's not the only use case. We have seen there is a trend of exponential increase of this over the years.
Next slide, please. So what will be needed, as I mentioned, more fiber backbone and backhaul. This is the percentage of fiber description in OECDs. Some countries like Swedeer are ahead, but other countries that mention industry 4.0 and the industrial revolution as important, such as Germany, are in the lowest fiber deployment in the OECD. So this is not to mention that fiber is the only option. There's many next-generation access technologies. But there will be an increased demand for investment in networks, and we hope that we can work together in order to make this a reality.
So this report -- next slide -- also mentioned can't talk about wireless without mentioning spectrum. One of the new uses of sprm spectrum that will come about with 5G is mill imter wave spectrum with very high frequency. And in 26 gigahertz. That's why we also need smaller cells because this type of spectrum requires smaller cells. But there's other frequency bands I will talk about in the middle, the 3.5, and lower bands are more good for coverage, and 700 megahertz. The report that will be published in January that I have been working on talks about the different countries' experiences in terms of spectrum but also trials and the 5G strategies.
Next slide, please.
But another regulatory question is smaller cells, that's great, but what are the implications of network densification? Who wants to be close to a small cell? And simply put, that means that under the current power density levels in many OECD countries, 5G cannot be deployed up to the lamp post. And if data is the new oil of the economies, for example, in driverless cars, we would need more deployment of cell towers closer to the user in order to reduce latency. So this cuts across many issues that we see in the OECD from transport to many other aspects, data, privacy, many other things. And we are looking very closely at 5G and intersect with Internet of Things as well as AI.
So next slide, please. We look into countries, many 5G trials. I just mentioned a few here, Japan and Korea. Korea in the Winter Olympics in February made one of the first displays of 5G services with virtual reality, augmented preality, but there are many other countries running trials.
Next slide, please. As well as 5G strategies
At the European level, there's a strategy, but also Australia, France, Korea. It's interesting, the Korean strategy, because they are thinking about DNA, data, networks, and AI. They really think that AI can help optimize networks in order to provide more coverage. So this is something that we saw yesterday, was talking to a company in Finland that has been very groundbreaking and providing large coverage of their 4G services. They have a network operations fully automated, and they are using machine learning and AI to enhance quality of services. So this is just to say that there is an intersect between what we are talking about, network deployment, and new developments in the industry.
So finally, I'll just go back to the basics. This means that many of our core traditional telecom issues, like streamlines, rights-of-way, efficient spectrum management, access to backhaul facilities, how to faster investment and deployment of networks will become even more crucial. So we have to -- it keeps regulators on our feet, but also it requires new partnerships that are arising between industry verticals and horizontal players like the telecom operators, as well as new partnerships among countries, like the 5G corridors in Europe in order to make fully automated vehicles a reality.
So I have just set the field here for discussion and happy to take your questions after ward. Thank you very much.
>> MARTA CAPELO: Thank you, Alexia, for a very comprehensive introduction, very insightful. I will now ask {Mongi Marzoug from Orange. You have very extensive experience in digital industry and Ministry of ICT Indonesia, so very broad background. Can you please give us our view in the topic? Thank you.
>> Thank you. More than half of the world's population remain unconnected according to last ITU estimates. And they expect more than 49% of the global population will be online by the end of this year.
For emerging countries -- as you know, Orange is running more than 20 mobile networks in Africa -- mobile is the platform for digital economy in this country, innovation and cution. Nevertheless, connectivity and inclusivity remain a challenge for the unconnected into more -- mainly in remote areas.
The IGF community has identified five main dimensions for and policy options for increasing connectivity. The first one, deploying infrastructure, physical infrastructure, broadband, wi-fi, spectrum, mobile, and universal access. Second, increasing usability -- applications, online service, local content, and media accessibility. Third, enabling users, human rights, inclusiveness, user literacy, digital stewardship, and entrepreneurship.
Fourth, ensuring affordability of access to Internet. Mainly based on cost of access to the network. And the last one, creating enabling environment access should be universal. Secure, reliable, affordable. And mainly government regulatory authorities and private sector has to contribute to that.
It's clear that government policy plays a central role in access expansion, ensuring efficient spectrum management by promoting digital infrastructure investment, encouraging network, including spectrum, network sharing and spectrum sharing, and providing affordable and reliable access to electricity, on grid and off grid; fostering online service on local content; and ensuring a level of playing field for the players. In fact, telecom operators still face significant technical and financial challenges in expanding networks into more remote regions. According to the Broadband Commission for Sustainable Development report.
ITU estimates that connecting the next 1.5 billion people will cost more than $450 billion. And how will 5G and IoTdom for many developing countries, access to electricity, water, education, healthcare, and transport remain the main challenges. Digital advance technologies like 5G IoT will play a critical role to address more efficient these basic needs. Indeed, 5G will increase and encourage mobile connectivity, which is the core connectivity form in emerge gent countries, while IoT will provide smart and optimized platforms and application. Both will improve and modernize the management of public utilities, business, in case of lack of legacy infrastructure, and trigger innovation and new service to the general public.
So 5G may offer a great opportunity compared to previous cellular technologies. 5G technology is expected to bring -- sorry -- to bring ultra capacity, ultra high rate, ultra low latency, massive connectivity, expand IoT for the support of reliability and latency, but as for ultra and low-energy consumption and ultra low-cost networks for low-density area and more advanced traffic management, multiservice (?). All these pictures will imagine and produce services and application well adapted to local needs. But what about network and energy costs? Network costs, despite the 5G technology, may prove less expensive per bit than previous generation. McKinzie foreesees that total cost of ownership for mobile access network will decrease dependently and this dependently on growth and (?) circumstances. The number of sites is the bigger factor driving the cost of open issue, followed by maintenance and engineering costs. But in fact, as I used to be an engineer in telecommunication, starting from 2G up to 3G and the 4G, I can tell you that the main factor for coverage is frequency, is to have low frequency, frequency lower than 1 gigahertz. It's the main factor in the case to have better coverage, more than what we can bring with technology.
So for 5G, it's better for coverage to cover remote area to have, for example, 700 bands in this quantity.
In a 5G network, automation of performance and operations process optimization, site visit, drive testing and advanced radio access network configuration, cloud or centralized radio access network, virtual radio access network may help to reduce maintenance, engineering, and save rental costs. The concept of frugal 5G with only cost-effective features, we implement in 5G only cost-effective features, introduced in India may be considered in many developing countries at least in a first step to expand Internet connectivity to remote areas.
Now, access to power to electricity is a main issue to expand access to Internet Africa. What is about energy cost for 5G? Massive mobile with low spectrum, low-band spectrum, and massive (?) would improve the gain of antenna and complex 5G broadband will increase energy costs, which already represents an important percentage of site running costs. In 5G, several critical functions are being automated to ensure energy efficiency, artificial intelligence, according to traffic load. In addition, low energy will help to ensure cost-effective sources of electricity to mobile networks and devices. And as we know, in Africa main country in the world, more than 140 or 45 countries now introduced, they open the production of power from rural energy to private sector in order to develop and to provide more power in this area, mainly in Africa.
5G and IoT --
>> MARTA CAPELO: Go on, if you can conclude.
>> Yes, one or two minutes. 5G IoT expanded to expand connectivity, connectivity 5G will significantly furnish mobile connectivity for the benefit of consumers. It will create opportunities to enhance user experience for digital service, application.. (?).
5G features and IoT applications will be much more important for the developing countries such as smart transport system, eHealth, education, smart grid agriculture. Thank you for your attention.
>> MARTA CAPELO: Many thanks, Mongi. The angle from Africa and developing countries is particularly relevant for our discussion.
I will now pass the words to -- the floor to commissioner Juan Manuel Wilches from the Commission for communications regulation in Colombia. You all also have extensive experience in telecommunications, and you can bring us the very insightful angle of the public interest of the regulator. Thank you so much for joining us. Please go ahead.
>> JUAN MANUEL WILCHES DURAN: Thank you for the invitation to the panel. I apologize for being a little late. I was in another panel participating.
I was thinking about the -- what to say at the beginning of this panel, being a regulator. You usually tend to think regulators like to regulate. In the case of CRC, we don't, we don't like to regulate.
And when you talk about technology adoption, 5G, IoT, and how governments can try to approach those new technologies and how to manage to develop those kind of new solutions in any countries, what I thought was that you need to think on the best way to promote the development of digital services, the new application of those technologies into the day-to-day lives. We don't need to think about net who, probably. That's something that comes with the implementation of technologies. We don't need to think too much or regulate too much what needs to be done and a lot of rules, and you need to do this and implement this kind of specific standards or whatever for these technologies.
I guess what the CRC has done in the last few years is to think about how to become more a promoter of the digital development and digital transformation of a country. Instead of trying to regulate or to impose charges or things, impose rules or new regulations on telecom providers.
We need to think about the best way for telecom countries to develop those digital services that complement their traditional telecom offer. And for that, we need to think on how to best promote the definition of digital transformation initiatives that encompass the network slide of things with the application, how to takes that to a regional perspective with the industry and how to apply those technologies in day-to-day productive activities, for example, or any other economic sector.
That will be, obviously, based on policies for the country. And we need to coordinate a lot that regulation that regulators are issuing with the policies in terms of data transformation and how to best construct or pave the way for digital telecom companies to be the ones who provide those solutions to the industry and to the different companies and entrepreneurs that need to develop those kind of solutions for the provision of services to the final user.
Based on that, there are some things also that need to be discussed in terms of how to think about when I am saying we don't regulate is to simplify regulation and to regulate as little as possible. So there are a couple of recommendations. There's a document from GSMA from February 2018, which talks about what is the best way for regulators to build that regulatory environment to promote the adoption of these technologies. And that includes six points that I guess Alexia also touched on. The first one is spectrum management. You need to define the different frequency bands that are going to be used. You need to promote deployment of front and backhaul. That means promote competition in those telecommunications markets that will allow to connect the different cell sites in all the different areas of the country. So it's just competition and trying to promote competition among stakeholders, among the different market players. You need to find ways to develop new sites. A lot of new sites will be needed for 5G. A lot of new sites would be needed to connect all the different IoT devices that are going to be located everywhere in the country. So you need to find coverage and need for the companies to do that.
The fourth point is to have network sharing agreements to be Friday among operators. The regulators shouldn't impose obligations or prices on those agreements. We already took a decision about 18 months ago because we didn't Symme pose any obligations on operators to share. Obligation just to share essential facilities. Because we didn't want to impose rules or restrictions on the way in which operators were collaborating to deploy networks and infrastructure.
I guess I would like to talk about network densification and how we need to promote that development and that deployment in the different regions of the country. And to also talk about define the power density limits, the radiation that cell sites produce, and how you need to work, and that's one of the key things that we need to work in Colombia. It's to work with communities, to work with municipalities to allow the deployment of infrastructure. If we don't work with communities, if we don't work with the local -- the mayor of the municipality, they are not going to allow the deployment of those cell sites or those new small microcells that need to be deployed are not going to be deployed if we don't work as a government, as a Ministry of ICTs, with the municipalities to allow the deployment of that infrastructure. And that's, I guess, one of the problems that not only Colombia but many other countries have. A lot of the communities are thinking about health issues, environmental issues. Like in the case of Colombia, there are a lot of people that take advantage of that and that they get to the communities with messages about you cannot allow for this cell site to be deployed, but then they are charging a lot of money to the operator for the deployment in a specific location.
So we need to work with municipalities, and we need to work with communities to make them understand the benefits that they have of having infrastructure close to them. It's better for the infrastructure to be close and not to be farther away.
I wanted to mention, just to close my first intervention, report that was posted by ITU July I guess this year, and this's a conclusion there, they talk about the deployment of 5G could probably increase the digital divide because investment in infrastructure, these new technologies, could take place only in larger cities, densely populated cities, and the infrastructure in the rural areas is not going to be deployed, so I guess it would be good to discuss what's the best way for governments to manage to make a good balance of we want to deploy 5G, but we need to complement that with coverage and service in other rural areas and how to connect people in those areas that probably are not connected. Around 50% of people in Colombia are not connected to the Internet, and we need for that people to be connected. So how do we deploy that infrastructure? How do we manage to -- for them to appropriate that technology if they don't have 5G? They wouldn't have 5G in, I don't know, two, three, four years, maybe more than that. But people in cities are going to have that. So how are we going to manage in our policies of digital inclusion that connectivity in those areas? Even if it's not 5G. Thank you very much.
>> KS PARK: Thank you. So as we can see from this first set of speakers, what we are discussing here is a moment in which if we do not address seriously issues of connectivity and inclusion (this is Carlos Affonso) our shiny new technologies might end up setting us more apart than we are right now. So instead of bridging the gaps, we will be enlarging the gaps from those who have access to this technology and those who don't have. If we think that the digital gap in terms of those who are connected to the Internet and not connected to the Internet, it's a serious enough problem, the tendency is for those problems to be really stressed out and really make it way larger with the deployment of new technologies, such as those that we are discussing here.
So now what we are going to do is to turn into a deep dive on AI. So the framework might be different, but the ideas we are discussing are the same. So how can we make sure that from the start, from the set of those technologies, especially on AI deployment, we make sure that connectivity and inclusion is right there from the start? So we have three very different perspectives. We will begin with Smita Prasad, offering an Indian perspective from the CCG, from the national law University from Delhi. So Smita, if you want to lead on.
>> Sure. Thank you, Carlos.
So as was mentioned already, I think AI is probably one of the big technological developments. There's been a lot of discussion at the IGF and just generally in the recent past, and we are all here because it's clear that this trajectory of discussion is going to continue over the coming years, and inclusion and diversity within this space is an issue that many people have raised, including Carlos and other colleagues on this panel. In this context, I think I want to foe kous on India, but moving beyond development economies when we speak of AI and inclusion. The conversation on AI has typically been dominated by actors from developed countries in the Global North. One of the big reasons for this is the technology is often developed in these regions. It just makes it more accessible for people from these countries to talk about it. This also means that while there is discussion on inclusion and accessibility and diversity, et ceteraics, all of these, and rights-based issues, many of these discussions are also focused on the issues that are raised in the developed countries in the Global North. While, of course, this discussion is much needed, I think the emphasis here is that the effect of these kind of technologies is felt all across the world. Many of these large tech companies that are developing AI are aggressively marketing their services and products to emerging markets, to people in the emerging markets, and the technology that they develop will affect everyone. But at the same time, the way in which we identify and deal with issues of inclusion and diversity in different jurisdictions differs vastly.
I mean, just basic thing likes our politics, our religions, and our cultures differ. And this means also that the language that we use to discuss discrimination and other issues like this is different. I mean, that's literally and figuratively.
So you know, on a lighter note, like makers of self-driving cars they joke that they know they achieve their goals when they build a car that can survive in Mumbai. There is also bigger issues where the way that, you know, the use of social media, filtering mechanisms and algorithms affects elections in countries like Brazil and India very differently from the way it would affect the U.S., and something that we have been discussing a lot in the past.
At the same time, there's also issues of like the use of AI in healthcare, where common diseases in different jurisdictions may just not be heard of in more developed economies.
And then there's also the more basic issues that we just discussed in terms of access to Internet, literacy, digital literacy, different issues here.
And so I think it's important that we take all of this into account and build in the not-so-homogenous Global South values into the data sets that we are developing -- that we are using to develop AI. But the other important thing is also to make sure that there's representation in the communities that actually engage in building and developing these technologies. And this is something that, like to step away a little bit from the rights-based issues, this is something that many countries have been discussing and they've taken note of. This is not specific to the Global South. A lot of European countries are working towards catching up in this AI race, but also a lot of Asian countries are working on this. And many countries have developed policy frameworks to kind of identify how they will engage with AI and how they will leverage new technology in their countries, whether it's for public services or private actors.
And here I would like to discuss a little bit in the Indian example, so India is one of the countries that has policy frameworks that address AI, and AI has come up in discussions not just in specific targeted policy frameworks on AI, but also our policies on data protection, on e-commerce companies. So this is a big issue. They want to facilitate innovation in India, and they are trying to identify how to do that, and unfortunately, the issues that I discussed earlier on the rights-based issues, the discrimination, the inclusion, is not something that's really shown up much in these discussions, but what has shown up is the question of how to facilitate innovation. And one of the ways that they are talking about doing this, for example, is to mandate storage of personal data in India, and then make sure that this personal data is then -- the data sets are given to Indian startups so they can use these data sets to build AI technology.
And we've seen this in draft policies so far. It's not something that's been done yet. But I think it begs the question of whether -- ensuring there is better representation and enshiring technology is built in the Global South as well as the Global North means this is the only way to do it and if this is the best way to do it. And I think there's need for more discussion among all stakeholders to address these kind of questions. Especially given that there are other implications. There are implications of privacy and other human rights involved in these questions.
So I will stop there, and then maybe we can take this discussion further.
>> Thank you, Smitha.
Then we turn to Christian Djefall from the Hobolt Internet Institute.
>> CHRISTIAN DJEFALL: Thank you very much, Carlos. My name is Christian Djefall. There is a J in my name just for you to know. It's a Nigerian name, so it's not easy to spell. I know that.
>> CARLOS AFFONSO PEREIRA: I will try my best next time.
>> CHRISTIAN DJEFALL: No reflection on you. I am working on a project on digital public administration. I am also involved as a member of the civil society in the open government partnership, working with the German government on opening up their automated system, including some transparency in automated systems of the public administration, and I am also a member of the eGovernance concept in Germany.
Having said this, I will give you a German perspective today, focusing first on AI a little bit, and then touching upon two other points, which is in relation to AI organization and strategy.
What's really interesting to me in the AI discourse is that we keep revisiting very basic topics of AI, and I am talking about concepts, which could be due to the fact that AI is an emerging technology, and there is a lot of technical development, so this would be in parallel with the 5G narrative that you have, like, 3, 4, 5G and AIs building up. We are talking now a lot about evolutionary algorithms, so there's new technologies creeping in. But what I would suggest to you today is that I think the reason why this discussion is more and more vital is that we see the social purposes more clearly. And this is due to the fact that AI is a general-purpose technology, and we are just about to encounter the different social purposes it can have. And you could see that actually, at this very IGF in different panels, where talking about discrimination and inclusion, in some panels it was specifically mentioned that we are worried about discriminatory AI systems. This discussion started with, of course, also systems in e-government that actually distinguish between different classes due to ethnicity. But then, and there is a very interesting use case in school administration in the U.S., but also in other places, when they found out that the algorithms discriminated, they actually used those systems, turned them around, and took the information in order to tackle discrimination. So here you have a clear example of how just the social purpose we are focusing at could turn around what these applications mean, and this is what I mean by general-purpose technology.
It can be used in very, very different ways.
To give you just -- and I think this is really, really important to stress to give you some examples, just to open up the space of imagination in that regard. I mean, electricity, for example. You can run a dentist's chair or an electric chair with the same technology, and if I get a little bit more -- if I include a little bit more in that, since we are in a building of the United Nations, we say (?) to plow shares, which also tells you about the general purpose of metal.
So this is very broad point that I would like to make and really to stress to you for the second argument I would like to make, and this is that when we talk about AI, we talk a lot about micromanagement of applications. So all the very good initiatives that come up with principles of AI talk about nondiscrimination and fairness and things like that, and I think this topic of inclusion reminds us that there is not only the microlevel but also a macrolevel, a general level, that is very important but that we don't really see. And I think this is very much about governance and the broader topic of inclusion I think reminds us about this macrolevel, and I would like to stress this with two points, and my general metaphor for this is actually the very room we are sitting in because we are at a multistakeholder conference talking about inclusion, and I have seen so many people passing by looking in unable to come in, which is actually -- I am actually mad at myself because I said how can I be part of such a setting where I said I want to have an inclusive discussion, but like a sheep, I accept. Of course, there is reasons for that. I completely understand them, and I don't want to start a revolution here.
(Laughter)
But I think nevertheless, it's very important for us. This is such a macro reason, which shows you how the architecture in a way frames us and brings us into a setting where we automatically accept certain things and leave people out. And I have seen dozens of people actually looking in and passing by.
So what are my points for AI in this regard? One is organization. We see a lot of movement in that regard. And I think one is the struggle of the United Nations to find the right forum for AI, something which has not really addressed in the fora I went to, but there is -- of course, it's always an organization, you know if you work for an entity, even in civil society, everybody tries to in a way get onto issues that I think lies in human nature, but we are actually at the moment currently looking who is responsible in this organizational setting founded in 1945 and developed in waves. For me, the size of my Erie moment of understanding was this March, I think, or May, when the ITU and the UN Development Agency hosted AI meetings in parallel on the same topics and could not agree to merge the meetings, even though they both tackled humanitarian and development issues.
And the other thing I would like to comment is in terms of organization, very much the democratic design and multistakeholderism, which leads me to my second point, strategy. We are now, I think, in a point where this architecture I was telling you about is made. So as we are speaking now, the German government will decide on its AI strategy, and two points I would like to mention that are already known to the public is that they will include a big bunch of measures that actually link the German efforts in public administration but also in other fields to development policies. So this has been a big part of the strategy which I think is -- we haven't seen in a lot of strategies so far, and like the Canadian government, they also want to found an AI observatory, really understanding social impacts and looking at the development of technology as we go.
And I am -- actually, I am in many ways getting more and more optimistic about the strategic development due to the fact that AI, the strategy game started out a little bit as a weapons race, and it was commented in many instances, but now I think only three months ago the Chinese government stressed that they would like to cooperate, and they were open to partnerships. The Russian president talked about sharing knowledge about AI, and the European Union, these are the examples I know, but you can complement them. The European Union actually tried to have a common strategy, including also non-EU Member States that was also focused on social and ethical issues.
I would like to close and also to point my questions to you because I think in a way we have to also discuss what the role of the IGF is in that regard, and I think it should also be mentioned and I think scholars can say these things more easily, that the IGF is in a way at a crossroads. We had now successive meetings in Geneva, Paris, and Berlin, which might be a sign that in a way support is shrinking for this forum, and I think if we stop to be engaged and innovative, this could be a problem for the future, and I am not worried that the IGF will be abolished, but I am coming back to Carlos. Carlos's comment. I think the big danger here is that the IGF will be a fig leaf for an actual widening digital divide and a fig leaf for many bad developments that could take place if we don't be active. So I am looking forward to the discussion, and I thank you for your attendance.
Thank you.
>> CARLOS AFFONSO PEREIRA: Thank you, Chris. I think you end up raising a very interesting question, which is when we think about inclusion and accessibility, you bring this layer of Internet governance, and when we question who is going to handle the global discussion on those technologies, and especially on AI, there is this big question mark on the decision of different international fora, and of course, different fora will offer different opportunities for inclusion and engagement. So there is an important layer of international governance for us to consider here.
So very quickly, I will turn to Eduardo Magrani now fromfrom ITS Rio now to offer Brazilian perspective on the topics. After that we will open up to the audience to make it a more lively debade in the final segment of our panel. So Eduardo, please.
>> EDUARDO MAGRANI: Good morning, everyone. Thank you, Carlos. I am very happy to be in this panel, first of all because thanks to the merge, we can talk from ICT to new personhood of robots. This is a very interesting opportunity to talk about those issues.
This Italian ethicist says that humanity enters now in the era of design. Let's hope it's the era of good and not bad design. That's where my speech starts to connect a lot with the speech from Christian.
Our behaviors are being guided now through architectures and through online platforms. That's why we must pay a lot of attention in those new environments we have today. When we talk about IoT, the first step is connectivity, so that's why it's very important for us to address issues from the Global South. For example, connectivity today, access to the Internet is now a human right. This is a statement from the UN from 2014. And it's being more and more important. Digital democracy and digital issues we have on our civic relation with the state also. So connectivity is the first step. Half of the population does not have Internet access, and in Brazil we have pretty much the same scenario. Almost half of the population in Brazil simply doesn't exist online. How we can jeopardize many civic rights we have today in the Internet environment.
This problem of digital divide can create a huge gam, as Carlos was also addressing, between developed countries and emerging countries in terms of what we can do with digital innovation, digital awareness, or capacity building. I think the problem of digital divide nowadays can create a huge gap on the labor force in those developing countries, such as Brazil, India, or some other countries from the Global South. So that's the first step we should address when we talk about AI and IoT. We should not forget that access to the Internet is a human right and is the first step.
So coming to the IoT scenario, when you talk about IoT, we could also create a synonym that Internet of Things is also the Internet of sensors, but also the Internet of our things, intelligence robots, and people. We could also address IoT as the Internet of people because it can process big data all the time, and big data many times comes to personal data. So the conclusion is big data are us at the end of the road. So IoT is also important for human beings and should be addressed through human values, and I think that's an important statement also.
From a post-humanist perspective, we could state that things have agency, such as BleauLatour likes to address. Christian was telling that our data model is guided through architecture or digital platforms or things we could say. So those things that have agency needs to be regulated as such, I should say from different levels of agency. When we talk about connected things or AI, we should differentiate things that have narrow AI or strong AI because they can act or interact with us in a completely different way. We should not regulate a simple tool in the same way we should regulate intelligent algorithms with the capacity to auto-program themselves, such as many things that we can -- we could see right now that are producing (?) such as autonomous cars or autonomous algorithms from Microsoft and others, they are already producing damages.
I think we should come to a different ground of concepts of connected things and also AI to think about different regulations from those connected devices.
When we talk about the importance of inclusion in this scenario, we could also address the importance of inclusive engineering. Many problems, many dangerous we see today on the AI scenario begin with the exclusion part in the design phase, so that's why I think inclusive engineering, including minorities and have a better gender balance on AI development is crucial for us to have a (?) in this.
Also, the GDPR and the Brazilian regulation data protection talks about the right to explanation. And on the AI scenario, the inclusion discussion also passes through the right to explanation debate. AI technologies should be explainable since they guide our behavior all the time. And transparency is a crucial part of those new technologies. We should not interact with black boxes; right? Those new technologies should be more and more open and transparent by default.
In talking about by default, I think the rule of law should change a little bit in this scenario, paying more and more attention to the design phase. We should think about value-sensitive design and how to include human values on the design phase of these new and intelligent artifacts.
So just a quick remark that we can explore more on the debate with the audience. Thanks.
>> CARLOS AFFONSO PEREIRA: So thank you, Eduardo.
So now we have our two segments on 5G and AI, exploring issues of connectivity and inclusion. For this next part of our session, we would love to hear from you in the audience if you have remarks, questions, comments that some of our panelists might address. So let's open up the floor to you. Do we have any comments, questions, remarks from the audience at this time?
Please feel free to do so. So to the gentleman down there, if you want to just say your name for the recording. Okay?
>> My name is Olivier from Nigeria. My question goes to Christian. You expressed your fierce on the continued importance of IGF. Although I was in a particular forum yesterday and we also discussed the sad occurrence of coming together to discuss all these issues year in, year out, and somebody even made a joke that the slides that were used two years ago could still be used this year, and it would still be relevant. And I find it almost very true.
So from your perspective, what would you say? How would you advise? What would be the advice for the improvement of whatever IGF is doing for the IGF itself to be able to achieve its purpose?
>> CARLOS AFFONSO PEREIRA: Chris, before you answer, do we have any follow-up questions to this first round? If not, Chris, go ahead.
>> CHRISTIAN DJEFALL: Thank you very much for this question, and also it's these sad jokes that really make you reflect the most because in the beginning I was laughing, and then I was thinking that's actually a very, very sad joke you made. And I think I cannot give you the recipe to go forward, but I think what we should do is think about the strengths of the IGF and the things that make us come here and that make us come together and that are unique about this forum as compared to other fora. And I think if we have these strengths in the back of our head, then we should also not stop to be innovative and stick to the ways things have, in a way, settled at the IGF, but also maybe from time to time try some new things and try to especially -- I mean, I think it's interesting that the IGF tackles more and more technology issues, so maybe we can think about different formats for AI, for the IoT, for 5G that in a way develop also the forum along as we tackle new technologies, which is not a very concrete answer, but maybe we should think about actually connecting a little bit and really try to not only reflect but to be innovative, so I will try to find you after this session just to make contact.
>> CARLOS AFFONSO PEREIRA: Thanks. Please, go ahead. Say your name.
>> Thank you. Thank you very much for the interesting contributions. My name is Susan Taltra. I am interest from ITU in Geneva.
I have one small comment, then I have a question for the panelists. My comment is about the -- I hear a lot in many of the sessions here and elsewhere that half the population has access to the Internet and the others don't. And I used to be in charge, actually, of producing the data, so I know it quite well, and I just wanted to make a small comment about that because the data about 50% refers to Internet users. So it is about people actually being online and using the Internet. It's not about access in the sense that if they wanted to have Internet they couldn't buy it. Because we also publish data on mobile population coverage, and we have now about 5 billion people that are living somewhere where there is a 3G network. Okay? So it's not just about access in the sense of not being able to buy a package and being able to connect, but there are other issues like affordability, skills, content that are very important. I just wanted to mention that because it's used all the time, and it's not precise.
Okay. But having said that, and actually, 4G population coverage is also already higher than 50% of the population. But I have a question for the panelists because I think there is an interesting discussion going on here between 5G, IoT, and AI. And my question is a lot of the policymakers in developing countries have mostly focused on expanding the mobile network, and we have heard about that a lot in terms of different panelists, and anyway, we hear it all the time. But we also hear like the OECD presentation focused a lot on the fixed backhaul and backbone infrastructure. I would like to hear a little from the panelists about that link, and especially if we want to bring AI, IoT, and the ability to exploit that to remote areas, what does that mean in terms of the technology infrastructure and especially also when we talk about the backbone and the backhaul.
Thank you.
>> MARTA CAPELO: Thanks a lot. Maybe if I can complement your question, one of the use cases that you talked about is the mobile broadband enhancement, and there's also fixed wireless. To build on this question, do you think that can be a tool to actually bring coverage to more people?
>> {alexia That's an excellent question. Thank you so much for both. We also stress at the OECD the difference between Internet usage and access. There's different ways of measuring access. One is broadband penetration, but there's also coverage, and within coverage, population coverage is one, geographical coverage is another. So even if the population is covered, they mostly are in urban areas, then we still are lacking the geographical component. But this is great, and affordability is specifically important for countries, like my country -- I am Mexican -- where for many years this was one of the main barriers due to lack of competition. So we try to stress the importance of institutional framework and legislation that promotes competition or to bring prices down to eliminate at least one of the barriers, with is affordability.
In terms of 5G, in general, wireless networks and fixed networks, although there's been a lot of talk in Europe and many other fora about fixed-in wireless substitution of services, indeed, the networks are complementary. The core network of wireless network is fixed infrastructure, and the difference that we see with the evolution of wireless is that the it needs to fit more in with fixed infrastructure. What we will see more is what we call network -- just exacerbates what we say in many countries. In 2011, I had the privilege of doing a review of Colombia. The fixed network infrastructure was highly incipient, and many times regulation didn't tackle the fix because most emerging economies think wireless, wireless, wireless.. We keep saying this, not only fiber and backbone is the part. In 70% of countries, mobile data is downloaded into the wi-fi. So the quality of wireless networks increases with more fixed infrastructure, even in the access part, not only the core networks. So this is something definitely to keep in mind, and will also being more and more important as we evolve into the generations.
The reason why I try to start with the presentation saying this is the road to 5G and try to bring back to reality the hype that we have, the benefits that we will see for 5G in the usage case scenarios will only be there as much as we work in the foundation of the digital economy, which is the access part. And access, not only the access to just simple 256 kilobit Internet connection; access to high-quality broadband for all, like, within countries and among countries. And we are trying to work to promote this. I was recently in Southeast Asia working on a toolkit for how to connect SMEs, and many of their issues there resonated with what I see in Latin America. They said yeah, but what can we do to make broadband affordable? That has a lot to do with institutional frameworks and how sound they are, how they promote investment, how they promote competition. So we are trying to work together, and the ITU as well, that they are doing these efforts as part of the Sustainable Development Goals.
And just to finalize this, I was pleasantly surprised in February that The World Bank signed a partnership with the GSMA for IoT and big data for development. And The World Bank is bringing this new perspective that they really think the IoT will help developing countries come into the digital transformation, and we also think so at the OECD. So I leave you with those food for thought.
>> Thank you. Mongi, can I ask you also to relate this to what you see in Africa?
>> Thank you. The issue of access and coverage. I addressed in my speech the main need for better coverage in remote area is to make available low-band frequency at 700 megahertz, at 800 megahertz. It's the main, even before technology 2G, 3G, 4G, and 5G.
The second thing, it's still (?) a common network between operators in low-traffic areas to share spectrum, to share network. I agree with you that if we want to connect radio sites, you know in many African countries the fixed network, to install copper into the home is very low. The usage is -- I don't have the figures, but lower than 5% or they have broadband, fixed broadband in Africa, mainly in Sub-Saharan Africa.
So in fact, the main -- the mobile network is the main forum to access to Internet in Africa, in perhaps all African countries. So to connect radio sites with fiber because until now, 2G, we have used only microwaves. So to connect with -- in order to allow more bandwidth, more traffic.
As you know, more than 50% -- I think 60% or more -- of broadband traffic is video. But we can, in fact, provide and develop many applications without needing video or without needing high-definition video so we can optimize traffic in order to optimize coverage, at least at the first step. This was, I indicate the concept of frugal network, frugal 5G network as defined in India.
>> MARTA CAPELO: Christina, it seems we have two questions from remote participants?
>> Yeah, we are sent two questions from Yenna Fung at net mission dot Asia. She is asking how we can eliminate the gap between the labor force and the rate of rapidly changing technological world. And she also is very active in Asia with Asia Pacific youth IGF. She wonders how we can help eliminate the digital divide when there is an urge to have technology advancement.
Thank you.
>> CARLOS AFFONSO PEREIRA: So who would like to start from our panelists?
Please.
>> ALEXIA GONZALEZ FANFALONE: Thank you very much for the question. Recently we published a report on rural digital divides and the different strategies that governments are doing, but also the technology. And one point of view is that technology keeps advancing. It doesn't have borders. What we have to update is our policy frameworks and how we get up to bring everybody online, how to foster investment in the networks for everybody to enjoy the digital transformation.
And this is a concern, the rural and urban digital divide is not only a concern in emerging countries. Sweden is also very rural, and part of their main concern is how to get the last 10% up into high broadband connectivity targets that they have. So this is a common concern, I think, for the whole world. And as much as we try to build policies in order to foster deployment of networks by private players where they can and complement that with public-private partnerships or EU funds or government funds, where it's not commercially viable, we'll be able to close these gaps. But one precision is it's not only access to basic Internet, which is what has been declared in many constitutions as a human right. It's the access to high-speed broadband that will actually bring the productivity enhancements that we need everybody to take advantage of the digital transformation.
>> Just add to that, the pace of technology is very fast, and sometimes public policies are very, very slow, such as legislations. In Brazil, we had a national plan for the high-speed broadband, and it was a fail. It took many, many years, and when the project was launched, it wasn't adequate anymore. So this is a problem. I think governments should be more aware of the pace of technology to try to make it faster, to make it adequate. Right? And it connects also to the speech of Christian talking about strategy plans. Rio de Janeiro, for instance, received the mark of smart city. The problem was that the project itself was completely elitist. It wasn't guided by principles such as openness, transparency, inclusiveness, and nondiscrimination. And I think governments should pay attention not only to the pace of technology, to try to regulate it better, but also the principles behind all those public initiatives.
>> CARLOS AFFONSO PEREIRA: Juan, if you want to add something.
>> JUAN MANUEL WILCHES DURAN: I guess when we speak about digital divide, governments, policymakers, regulators can do a lot to try to connect people in a country. You can have development of backbone, backhaul, you have a lot of networks access, get the networks to the people. But even if you do that, you might find out that people don't use or do not connect, do not use Internet because they don't have any use for it. We found there was a very big comprehensive policy that was developed in 2010 in Colombia to 2019. It's finishing up this year. The first things that we found out in 2010 was that people didn't find a use -- Colombian people didn't find a specific use for the Internet.
There were, I don't know, a lot of programs, adoption programs. The Ministry of ICTs connected 250,000 people -- around 200,000 people using subsidies. They subsidized the broadband connection. But people didn't continue to pay for that, and the stability of that is unviable because you have to pay year after year those subsidies. It's very difficult for the government to achieve those goals if they didn't combine the deployment of infrastructure with adoption initiatives. And adoption initiatives need to go to understand what are the needs that the different communities have in the country, what are the productive industries that are located in those regions, and how to make a match between the connectivity and the productive industries that are there so that you have people that use the connection, the access for a productive thing, whatever they do, and try to continue to use that connectivity, too, for their day-to-day lives. So you need to find that connection. That was part of the policy defined in 2010. You have a digital system in which you have offer and demand. You offer a lot of connectivity. But if you don't have demand, you cannot develop that appropriation that you need from the population.
And I guess it's not an easy job. The new government which started in August this year is still in the process of -- their goal is to connect 100% of the population, and they haven't defined yet the best way to do that. Governments have tried through wi-fi hot spots in different regions of the country, subsidies, connectivity, access, working with public-private partnerships with operators, tried to connect everyone. But I guess you need to complement that with the complete digital solution, which includes connectivity and content.
>> MARTA CAPELO: Thank you. You would like to ask a question? You are a lecturer. Can you give the angle of side of Asia?
>> Thank you. I am lecturer of Beijing University of Posts and Telecommunications in China. I have been doing research on intelligent Internet of Things, especially the vehicular networks. So like we are submitting proposals for the intelligent transport systems with many vehicle manufacturers, and while talking with them, I realize that instead of the technical barriers, there are many other factors that hold them back from doing massive, like, production of self-driving cars for commercial usage.
So one of the challenges is the legislation for these self-driving cars. If we take the (?) problem as an example, like up to now, all the confirmation of liability is in traffic accidents is human centered, like the policy identifies which driver causes this accident and let him take the corresponding responsibilities, like paying fines or even, you know, going to jail or whatever. But in the future, intelligent transportation systems, there's no human involved. Like the reason of the accident may be embedded system failure, maybe design flaws or algorithm flaws or even remote hacking. So without the participation of humans in decision-making, how can we identify the responsibilities among these manufacturers, like makers and designers. As the commissioner said before, it's very important that we bring these new technologies to people's day-to-day life. So we should -- is there any like policy frameworks we can follow to identify the responsibilities among the suppliers so that we can bring more international and diverse companies to the development for these new type of applications?
>> CARLOS AFFONSO PEREIRA: Thank you. Yeah, I think we are almost running out of time, but we have -- yeah, we have time for this one additional -- two additional questions. Go ahead, very, very quickly.
>> (Off microphone)
>> CARLOS AFFONSO PEREIRA: And let's get the second one.
>> Thank you. Very quickly, congratulations for this workshop. It has been very useful because actually, I think those technologies are largely related. We need 5G to foster IoT, and we need the data generated by IoT to train that artificial intelligence algorithms.
And regarding artificial intelligence, by now, OECD, ITU, UNESCO are working on that field. On a global level and in your view as panelists, do you think it's the best way to address these global purpose technologies, or we need something like IGF but for artificial intelligence governance?
Thank you.
>> CARLOS AFFONSO PEREIRA: So Chris, if you want to take this question.
>> CHRISTIAN DJEFALL: Both questions I think are really worthwhile. So we have one time, rather, at the national level or regional level and one time on the international level, and so my approach would be to look at what we want to achieve with that, starting with the international level. So depending on what fora we put it in, I think this frames the results. As I was trying to say that with my -- with this metaphor with the room. So depending on how we build it and who we give the topics to, the results will differ a little bit and also our discussion will differ.
So my feeling is that we should use existing fora, but only if they are up for the task, so only if they are also willing to adapt to a certain degree. So I think for IGF it would be necessary to maybe find new ways, new forms of discussion, and to -- because I think AI governance is different from Internet governance. The principles at stake are different. If you look at the core principles that have been discussed here for years, and I think there's a lot of clarity. There's similarities like data protection, but there's also differences. So my suggestion would be to think about the IGF also because I think what we need in AI governance is meaningful participation and inclusion, and I think the IGF is a very good example how such a process of discussion can be started, but obviously, as we heard also from the participant before, there is a need to reform.
And very briefly on the national, just to -- I think it's in a way the same thing. Here we also are experimenting with new ways at my institute in a way to have, for example, random samples of the population either deciding or discussing issues in order to find new ways of governance of AI. If you look, I looked only in a few states, but there is a real proliferation of organizations tackling AI issues, and I think, again, there needs to be an inclusive organization, and for example, one worry for me is that we see it a little bit too much, at least in Europe, through the lenses of data protection at the moment, and I think that has to -- this view has to broaden, and we also need to reflect this from an organizational standpoint. So I don't have the ready-made solution, but I think it would be wrong for me to -- even to do it because this is part of the discussion we should have, and I hope this is a little bit satisfactory for you.
>> MARTA CAPELO: Thank you very much, Christian.
We are a bit over time, but we just realized that our session is starting only in five minutes, so I am going to thank our panelists and ask you for like a challenging last task. Can you just give us a key line in 30 seconds, each one of you, starting with Eduardo. What is like the key line for you from this workshop, and then I would invite you to refer to the conclusions and to the report. We will try our best to wrap it up.
>> EDUARDO MAGRANI: So my key line would be we need new person who foreign intelligence robots, because of her question. They are becoming more and more autonomous. There become no causal nexus to give liability to those agents.
>> MARTA CAPELO: Commissioner.
>> Government regulators, policymakers need to understand what's happening. We need to go deeper into getting to know our sector industry and impact that new technologies are having in that specific environment. And try to intervene as less as possible, as little as possible. Try to let things flow a little bit, probably some basic principles, and that's it. Not try to control everything.
>> MARTA CAPELO: Really like that.
Christian?
>> CHRISTIAN DJEFALL: I gave mine already, so I can be very brief. I think it's a formative phase at the moment for AI governance and for you as a multistakeholder, no matter from where. This is the time to get engaged and raise your voice.
>> I think that 5G and IoT will bring many benefits for emergent country in order to improve access and inclusivity, and I think the policy and regulation will play an important role to, in fact, benefit from those features.
>> {smitha Thank you. I just want to take off on what he said. The discussion on the fora and that discusses these issues and takes it forward is really important because each of the forums that we have mentioned here, whether it's UNESCO or OECD, World Bank, or national governments, the private sector commented with different perspectives, but the issues and the technologies that we are discussing are so all encompassing that we need all of these perspectives and to find some ideal way to make it work.
>> ALEXIA GONZALEZ FANFALONE: Thank you. I guess the core message of the presentation I made is as we evolve into gigabit networks and 5G and the next evolution in general of communications infrastructure, we really have to go -- it goes back to the basics. It exacerbates our traditional telecom issues which are streamlines rights-of-way, efficient spectrum management, how to deploy more backhaul and backbone, and how to have access of these wholesale facilities, and in general, power density regulations that might need to be thought of.
>> I guess my message is we really need new frameworks for policymaking to encourage diverse companies to be involved in the development and deployment of new technologies.
>> CARLOS AFFONSO PEREIRA: Thank you to all of our panelists. If I can ask you to join us on a round of applause to our panelists.
(Applause)
Marta, if you have any closing remarks? That's it. So okay. Thanks, everyone. Hope to see you soon. Thank you.