 Yeah. Sure. Yeah. Hello, everyone. And welcome back for those of you who have been with us for their, for the workshop. This is the last event for the week. And also welcome to those of you who join us on the YouTube live stream today. So I want to remind everybody that exceptionally or differently from the rest of the week. Today's session will be is being live streamed on YouTube because it is the closing session of the workshop, but also because we've incorporated this event today as part of a series of events in the series of the International Year of Basic Sciences for Sustainable Development, more about which in a minute from Marco. As we have tried to do throughout the week, we are trying to bring a variety of perspectives to bear on the problem of ethical challenges in machine learning. We've heard for and for now from a number of scholars working cross-disciplinaryly throughout the week. And we're very pleased today to have four panelists and speakers with us today that will illustrate their own perspectives on the problems and introduce them in a minute. And then we'll close off the workshop and close off the event with a panel discussion around three o'clock. And again, very happy to have seen so much engagement, so much participation throughout the week. And I'm sure it will be the same today. Marco, if you want to say a few words about the International Year of Basic Sciences for Sustainable Development, perhaps before we kick off? Yes. Thank you, Roberto. I even have some slides that I will show. So as Roberto said, today's session will be also presented in the framework of the International Year of Basic Sciences for Sustainable Development. And the title of the seminar is Embedded in Ethics in Machine Learning. So what is this year about? So this year is about focusing on the links between basic sciences and the SDGs, the Sustainable Development Goals. And the idea is that this is a unique opportunity to discuss with stakeholders, with government, with institutions, and convince them that basic understanding of nature is really important, so that action will be more effective for the common goal. And STP and UNESCO are partners of this International Year. One slide about the SDGs. We mentioned that during this week a couple of times. I'm sure you're heard about it, but the SDGs have been developed in the framework of the 2030 Agenda for Sustainable Development, which was adopted by all UN member states in 2015, as they share a blueprint for peace and prosperity for people and the planet now and in the future. And in the framework of this agenda, they've developed 17 sustainable development goals, which are a call for all the countries developed and developing for a global partnership. And I think the main point about the SDGs is that in some way they have to be reached all together. So they go hand-by-hand. So the idea is that if you improve education that will also benefit health, it will benefit economic growth, and so on. And at the same time, we're going to tackle climate change and work to preserve oceans and forests, so these different aspects go together. In a graphical way, you can see them here. And as was mentioned yesterday and the day before yesterday, AI can play a role to reach some of these or to tackle some of these SDGs. We heard about health. We heard about education. We heard about decent work and economic growth, and so on. So this is the SDGs. What about basic sciences? Well, it was highlighted that basic sciences are the sine qua nom for sustainable development. And in fact, the essential means to meet crucial challenges, such as access to food, to health coverage, and to communication technologies. And in fact, to tackle the issues of 8 billion people on the planet and to reach this climate change and depletion of natural resources and so on, basic sciences are really, really important. And in fact, applications of technologies are kind of easy to recognize. In fact, we sometimes have the risk of focusing only on that. But on the other hand, contribution of basic sciences are not really appreciated. But at the same time, they're really, really important. Just to name one, we're using the web, and that was developed at CERN as part of a basic science research in some way. So again, this international year is to highlight the importance of basic sciences for the SDGs. And ICTP has planned a number of activities to support these teams. So this seminar is a part of these activities, but we have more of them. This year was integrated just a couple of months ago at the UNESCO headquarters in Paris and events will be organized around the world up to June next year. And in fact, ICTP itself will organize more activities about climate change, AI, human ecology, and open science. Finally, the website, you can find it here on the top of the page. And you're very much welcome to visit it so you can see what other events are going on all over the world. And that's all from my side. Back to you, Roberto. Thank you, Marco. That's really interesting and really a perfect fit for many of the things we've been discussing this week. So today, I'm very pleased to welcome our panelists, the first of whom is Dr. Jabera Matogoro. I think Jabera is here. Are you? Is it? Yes, I'm here. Thank you. Yeah. Hello. Welcome. Let me introduce Dr. Matogoro, who is a PhD in telecommunications, engineering, and a master's of science and computer science from the University of Dodoma in Tanzania, is a teaching staff at the Department of Computer Science and Engineering of the College of Informatics and Virtual Education, and project team member of Artificial Intelligence for Development is working to establish Africa's Anglophone Multidisciplinary Research Lab and is the founder and chief executive officer for Tanzania Community Networks Alliance, which is an umbrella organization for community networks in Tanzania. Or Tanzania, I'm not entirely sure how you pronounce it in English. Maybe Tanzania, please correct me. Dr. Matogoro is the recipient of the Open Internet Engineering Mozilla Fellowship for 2019 and 2020. And we are delighted to have him here with us today to talk about the experience from Anglophone Africa on ethical and societal challenges of machine learning. Please, over to you. Yeah. Thank you, Professor Robert. And I also shared you a deck of slide. I'm not sure if you can... Yes, I was going to display them now in just a second. Just to share my screen. Absolutely. Hopefully you can see it. Please just say next slide when you want the next slide displayed, if I can find a way to make it full screen. Oops. No, it's not. Can you see it? It's not the one. It's because when I go full screen, it's a bad day. So just a second. When I share it, it goes in the wrong screen. Sorry. Yeah, I see that. That's what's happening. I'm going to go back to Zoom if I may. Apologies, this was not what I was supposed to be doing. OK, let me try again. Maybe I'll try the PDF because I've got a problem with the... I think the PowerPoint gives me a problem. OK, let me try this. Share the screen again. My nose in PDF? I'm not sure. Yeah, I got a PDF here as well. Is that OK? Can you see this? I shut the PDF again. Can you see my screen now? Marco, can you see my screen? Yes, but it's Monday's presentation, I think. I'm so sorry. Absolutely. I'm so sorry, I got the wrong PDF. You're right. It's up in the actual. I got it here. I just, it wasn't the same. I'm sorry, it wasn't the same deck and the same email conversation. That's why I got it lost. OK. A little hiccup. Here we are. Yes. Yes, yes, yes, yes. OK, go full screen. I can try too. Full screen. Can you see this now? Perfect. Please go ahead. I'm sorry. So sorry. No problem. Yeah, as Robert mentioned, I'm part of the research team on AI for the Africa's Angrophon interdisciplinary research lab. And we are working to see how the artificial intelligence can help to address challenges facing humanity. And I'm happy to share experience on ethical and societal challenges of the machine learning and how we are trying to embed those issues in our research lab. Next slide. Yes, we actually made a official launch this year in general. And as part of the launching, we had the government of Tanzania. That's the Deputy Permanent Secretary from the Minister of Information, Communication and Information Technology. And we are happy that the government is supporting and we are working on behalf of the Angrophon Africa. Next slide. Thank you. And these are also some of the university management, the vice chancellor, the deputy vice chancellor, research academic and consultants. And we also have the deputy vice chancellor for the planning, finance and administration. Next slide. Thank you. Yes, maybe I need to spend some few times on the challenges. Why are we working on the lab? We are trying to address almost four critical challenges. One is the skills. We understand the artificial intelligence existed since 1946 when it was first mentioned. But the uptake is special in addressing the social challenges has been remitted, particularly in the academic context. So we understand some of the challenges, especially in the Angrophon Africa, is the shortage of human capacity. And the lab is working towards addressing the shortage of human capacity, where we are training youth, women and people in the Angrophon Africa to be able to participate actively in the AI industry. The next challenge we're trying to address is on the infrastructure. We understand that there is a raw uptake of artificial intelligence, especially in the Angrophon Africa, because of the lack of high computing power, especially that can compute and manipulate data, larger data sets. And also, there is also challenges on the African origin data set. You find in the academic context, people are learning about artificial intelligence, but when the training model and algorithms, they are actually using data which are not from the African regions. So as part of the lab, we are also working to address that issues. And the third challenge we are also trying to address as part of the lab is on the inclusion. We understand in the current situation, there is a poor inclusion of gender and marginalized group, especially in the Angrophon Africa. So as part of the activities of the lab, we also are trying to come up with a solution that will contribute to addressing the inclusion, especially in artificial intelligence and machine learning. And the other challenge, as I also mentioned, when I was speaking about infrastructure is on the data set. We understand that most of the data sets are data variable, especially from Africa's. You may find they are not ready for the AI usage. That means they may not be usable for training model and the logarithms. So as part of the lab, we are also working to contribute on a contraction or development of the African origin data set. I understand there are also other initiatives towards addressing this, but as a lab, we are also working toward having African origin data sets that can help in addressing issues in the society in the Angrophon Africa. Next slide, please. Yes, I remember Marco was speaking about the sustainable development goals. And for our lab, we are working on a number of sustainable development goals like goal number nine, industrial innovation and infrastructure. Also on goal number two, zero hunger and goal number 13, climate action. And also we are working on goal number three, good health and the well-being. So from the point of what Dr. Marco's mentioning, so we find there are a number of synergies and we're looking for working ICTP and other stakeholder in this area. Next slide. Yes, so the lab is working on four thematic area. The first one is infrastructure and the data ecosystem where we are working towards supporting AI training, research and innovation, but also contributing to the African origin data set. And the second thematic area where our team is working or our lab is working is on healthcare, which we are trying to apply AI for improving our diagnosis in diseases, treatments and drug, drug, drug, drug, drug, drug, drug, drug, drug. Yes, thank you. And another area which we are working is on digital economy, where we are working to see how AI can accelerate digitization and automation of African industry. And also the fourth one is on environmental conservation and agriculture, where we are working to exploit the use of AI for environmental conservation and by the vast monitoring. And also using AI for climate crisis management, where as I mentioned, the AI for the Anglophone Research Lab is being executed by the University of Dodoma in collaboration with the Nelson Mandela African Institution of Science and Technology based in Russia, both in Tanzania. Next slide. Yes, I think I should mention on what are we working how are we working toward a responsible AI? We understand that in order to have a responsible AI means we need to ensure inclusion. That means gender and my generalized group should be part of the equation. Also we understand that data sharing and the protection policy is a critical part towards a responsible AI. And we are hoping that our Research Lab is also working toward contributing on the data sharing policies and protection in the Anglophone Africa. But another concept that we are also trying to bring on board is on the heels of a human-centered design approach where we are taking on board stakeholders before developing AI machine learning solutions. Next slide please. Yes, from principles to practice. In terms of technical, we are core nurses and co-designing of AI systems. In terms of operations, we are trying to have a conducive environment that fosters AI uptext in Anglophone Africa. And also in terms of organizations, then we are raising issues and concern with AI systems. So as part of our day-to-day activities as the topic of this week, we have been discussing on the ethical issues as well as addressing certain challenges. And this is what our lab is working towards having a responsible AI in the Anglophone Africa. Next slide. Yes, we are also doing this in collaboration with some of our partners, CSIRA in South Africa, Ifakara Health Institute in Tanzania, Data Research, ICT Africa, Zindi. And we are excited to have many more partners joining us because we understand addressing the uptake of AI in Anglophone Africa is not the one-man show and we would appreciate having more partners and stakeholders working on the same. Next slide. Thank you for listening and I'm happy to be part of this conversation. Thank you. Thank you. Thank you very much, Jabera, for this very nice overview of all these efforts. What I suggest we do, we now go to Monday, was also similarly gonna give a short presentation and introduce him in a minute and then we'll take questions on both your presentations and discussion at the end of this first part if that's okay? Yes, yes. Perfect. Let me first introduce our next speaker. Next speaker is Dr. Monde Adenomon with a senior lecturer in statistics in the department of statistics at Nazarawa State University in Nigeria is a trained statistician and his research interests are in machine learning and in finance and economics and econometrics, time series analysis, financial time series analysis, spatial econometrics and interdisciplinary statistical analysis. Monde is also the chair of the International Association of Statistical Computing, the ISSC African members group from 2021 is a team leader of the ambitious Africa Nigeria team from 2020 and is the lead organizer of the Northern Nigeria-LISA 2020 symposium. He's also the founder of the foundation for the laboratory of econometrics and applied statistics of Nigeria. And he's gonna talk to us about teaching machine learning with R in Nigeria, challenges and prospects. So very much focusing here on the educational aspect of machine learning and I'm gonna go and find your slides, Monde, in a second. All right. Okay. Sorry about that. I've got them here. Let me know what you've got there. There you go, those lights. Let me share the screen now. Here we go. Can you see my screen? Yes, I can see the screen. Perfect. Please, over to you Monde. Yeah. Once again, I really appreciate the organizer of this short course or workshop. Of the truth I've been, I've enjoyed so many things and we hope that what I've learned from this workshop will pass across to my undergraduate student and postgraduate student. I will talk on the second slide, what we are doing. Sorry, the next slide. The outline slide? It's on, do you see it? Yes. The outline side will talk about what we do. We're gonna look at introduction. They'll talk about the wide open source, then why are, and the machine learning, that the challenges and prospects they will not conclude. The next slide will be talking about some of the things I do or we do. I also have team members. We have Lisa, Lisa Miss Laboratory for Interdisciplinary Statistical Analysis. It was a vision from one professor, Eric from USA. The essence of this network is to see how to build capacity of students, build capacity of researchers, how they can move between theory and practice and also see how to influence decision making in African countries using data, using data science. For the African member group, AISC, they's called International Statistical Association, Computing, the African member group. What we are doing in that group also, want to see how we can also build capacity of researchers in Africa in respect to statistical computing. Then I'm also connected to our user group. Actually, our user group are doing great things in Nigeria, but they are supporting us in terms of funds to organize capacity building program like workshop, like training. I'm the coordinator for the Nassarawa user group. For that of the fund is the foundation that supports women and guests in data science. This foundation, recently I'm collaborating with E-Base in a camera room. What they are trying to do, they want to see how they can ensure that education is obtained by women and guests and likewise the disadvantaged children. These are the things we do. But for today, I'll be focusing on how we teach machine learning in our country in Nigeria with respect to not part of the country. The next slide. For us from experience in academics, we found that the feed of data science is still young. And apart from that, we don't really have a university or a polytechnic that run undergraduate program in data science. And apart from that also, we also found out that even the curriculum we have or we use, we are using, data science have not been integrated in the curriculum. In fact, sometime, even in terms of the computing aspect, which is the technical computing, is still lacking in our curriculum. But we use that foundation, add a platform to see how we can easily, how we can also influence our students in using these codes and using R because of the advantage of using open source software. Now, the next slide, I will be asking, why open source software and why are? I know most of the participants here know the advantage of open source. In quotes, you all talk about, it's a free software, yes. And it's doing great things for us in Africa and developing country like Nigeria. Because if we want to go for paid software, they can be very expensive. That takes me to the next slide. We will be working with some of this software. Like recently, I know that Minitab is trying to in bringing machine learning into Minitab and some other software. But when we look at this software, they are great, they are doing great, but they are expensive. Because some of us, we use our personal facilities like me, like my university, we don't have a lab, it's that school lab. What we do, we use our, like me, I use my personal laptop. They were calling our students, they can also get their personal laptop and that could be, that is just a challenge, but we are breaking through since we have other options by having our own laptop and also having our own software. Therefore, for this software, like eView, they are expensive, 650 USD. If you look at Stata, Stata is a great software but it's expensive. The next slide talks about SaaS. For me, I have not even used SaaS before because in my university, we don't have SaaS and SaaS is not open source. And if we look at this software and some other software that are not mentioned yet, they are expensive. Now, coming to open source is a great relief for researchers in Africa and especially for us in developing countries. They will move to the next slide. We're talking about R. I won't be spending much time here, but I got introduced to R in 2012 when I was to start my PhD program in statistics from the University of Elori, meaning that at my undergraduate program, many programs I've done before the PhD, I don't even know about R. Not until recently that we are trying to bring R, some R into the curriculum using different means because of the techniques we help our students as its epinodes and then move on. And R has been so, it's a fantastic language that's very vast in statistical method, machine learning and graphic analysis and all that. We'll go to the next slide. The advantages are huge in the sense that if we are to go with the paid software, sometimes there are some limitations. But with the R program, with the R soft language, already somebody have mentioned that because of a lot of, it's a community of contributors. Sometimes I want to, most of the work I'm doing, I'm using R. One of my students, we're working on Paneva, we are using R. I'm doing my work on Bayesian, better to aggressive model, I'm using R. The advantages are huge and we are seeing the, it's the greatest experience using R. Then the next slide. The machine learning. Actually, most of the things I've been doing before now has been on statistical computing in terms of building capacity of students, researchers on how to understand some of the statistical method we use how to also get an auto-op idea in what we call statistical collaborating skills and all that. But already many definitions are out there and we have talked more on this. Is machine learning talks about getting insight from data. We'll talk about the training part and then the predicting part. Then the next slide. Yeah, for now I'm focusing on supervised machine learning. That's what I'm focusing on in terms of teaching my student and also when I have opportunity to teach some, we'll have a workshop outside the university we're dealing on the machine, the supervised aspect of machine learning. The next slide. The next slide talk about, we have many packages. If I'm teaching myself, I tell them that we have many of them, but I have an interesting one among this, in this list called Carrot. Carrot is a very robust machine learning package. In Carrot, you can do your support version machine. In Carrot, you can do your random forest. In Carrot, you can do your decision tree and so many. Almost like when I have opportunity to teach my student and also answer them, instead of going to look at many, many packages we just focus on Carrot. Carrot is very vast in terms of machine learning and all that. The next slide. Now, this is another interesting thing we're trying to put forward to the community of people teaching statistics, data science and all that. We're coming up with a framework called team R. The team R, the T that talks about teach. And the next slide, we'll be looking at different way we are teaching our teaching. Now, one of the way, since I belong to different platform, this platform is called the LISA. In the LISA, we have family of labs across African developing countries, not just African country, developing countries. We have a start lab in my university. We have a start lab in India. We have a start lab in Pakistan. We have a start lab in Ghana and many countries in Nigeria. We have many of them. And these are Israeli epinodes in terms of because our curriculum does not even have much statistical computing in most of these courses. Through this platform, we use it to teach our students by using the short course, using workshop and all that. And I don't have much time there. Let's go to the next slide. Now, as I mentioned, they introduced me. I am the African session, the chair for African session for international statistical computing. Now, what we did, I reached out to this IAC that we don't have African session and they give all the permission to have what we call African member group. And it's doing a lot of things. What we do through that platform, we have webinars, we have short course. This is one of the short course we had in April last year. These are the participants, the participants across students and also researchers. And because of the, we had little funding, the program was entirely free. The part of the webinar we had, we had one of it recently. You can see the title, but talking about data science in academics and also in industry. Then the next, this sort of picture you see, these are college students. One of the, we had a workshop for Northern region. We're trying to say, these students in Secondary school need to know how to idea on statistics. Sometimes they don't know what is statistics as a course. Are we part of our vocation is to also introduce them to a course called statistics. The next slide. Now, this workshop we had because this one of the first workshop now in machine learning. I was the facilitator. It was head form members of Nigeria's high school association. It was this year, September, 2022. People that came for this workshop, most of them are from the ministry, government ministry and from the industry. And it was a lot. We use machine learning using the carrot. The next slide. There is another way of teaching, the way we teach it. We use real life data sets. Just like my first speaker spoke about, sometimes when we are teaching this thing, we may have data set online, but sometimes getting Nigeria data sets, sometimes it will be a little bit difficult, but obviously one of the website I'm using for financial model is WW West. They have some data set for Nigeria's talk. We also teach. We also try to influence our students that there's feature in statistics. There's also feature in data science. We also have mentoring. The last one is relationship. Now what we do with relationship is that we are trying to tell our students we can move from theory to practice. Even after you graduate, you can see we can still maintain relationships so that when you have a job out there, you need a resource person. You have an opening for us. We can link up to us and we can, that can be an avenue to also impact in those settings and quarters. The next slide. Now we look at the challenges. As I mentioned before, data science is not well integrated into our curriculum. Even in statistical computing with R, sometimes you see people, sometimes when you go to some places, you see somebody from economy, somebody from sociology, they're applying a lot of idea in computing. We're surprised, but that may not be so for us in Nigeria. But we are trying to see that it's a problem for us to really move to the world of data science. It must be integrated into the curriculum. Then the basis is sometimes some people complain that coding could be a little bit stressful. It's one of the challenges of R. They say coding could be stressful and all that. Then so far as at now, we really don't have a university that run BSE data science. And sometimes we want to do it online, it could be expensive. But all the same, we also know that there are some other free online, you can learn some basics. This is a little challenge and we are looking forward that at some time to come, university will be running data science at BSE level because we help us in terms of getting the rudiment and then move to the postgraduate level. The next slide talk about even the university that run data science related courses, they're not too much and they run it at a postgraduate level. So my department of copter science and all that, we also have one in my university, Nasrawa City University. These are some of the university that do data science related courses at the postgraduate level. Then the next slide. Now, we have a lot of prospects if we continue this way, try to see this interdisciplinary nature of the data science and then the machine learning. One of the area, we talk about fraud detection. In one of our discussion, somebody mentioned from my country that we have a PMA system and there was a fraud that engulfed a lot of funds in billions. And thank God that we've learned some things here that not just giving everything to the machine that there should be involvement of the human. You see that fraud detection is very important which we can also do. The electoral process, like one of our lab in Nigeria in University of Ibadan was able to have a collaboration with the Electoral, the Independent National Electoral Commission. And the election is coming up next year and you can see that AI or machine learning can also be used. We have agriculture, we have the climate change, like the climate change, somebody talk about deforestation. Because of sometime the cost of gas, cooking gas is there, kerosene is there. And people are going to what is called the charcoal. Choco is getting from, they're going to work on the fire wood, burn the wood and they get the charcoal. And sometimes when there's, when they're for those tree, they may not replant and we can also use this one to also monitor and also help out. Security problem is there. Medicine and education can also be applied. A lot of, there are a lot of prospects in terms of machine learning in Nigeria. To conclude the last slide, in conclusion, it's a call which we are still doing from our own local perspective that data science component should be integrated in all courses in our tertiary institution. And apart from that, the NUC, which is the National University Commission, should see how to encourage university to start BSC program in data science with all this room that the use of data science, both in the academic and in the industry will be achieving the future. Thank you very much. Thank you very much, Monday, for this very interesting overview of the challenges, prospects and opportunities as well of machine learning in Nigeria. And I suspect many of the things that you said apply more generally elsewhere as well. So let me start by inviting questions to both Monday and Jaberra. You can either raise your hand and we can give you permission to open your mic or you can put them in the chat and I can then read them aloud to our panelists. While you think about your questions, I'll start with a question about, for both our speakers, the question about infrastructure because we've seen that we have different kind of approaches, for example, to education, but also different kind of needs in Africa, for example. But the assumptions that sometimes are made in the big private companies or Western countries where AI and other machine learning tools are developed are not necessarily always fulfilled. For example, all of these big data sets that require lots of storage capabilities or lots of computer power or lots of processing power, even just to train the models. So those methods even may not be applicable in a situation where there is perhaps lack of data science, education or also lack of infrastructure. So how do you see what is the next step in order to avoid the pitfall of not being able to catch up or lead in AI development given perhaps infrastructural or educational needs of the moment? Maybe I can go fast. Yes, please, please. Yeah, thank you, Robert. And actually as part of addressing the infrastructure in the lab, I also forgot to mention that our app with this country being funded by CIDA and IDRSE, and part of the funding we have, we also put in place physical infrastructure within the campus, where we are buying GPU data saver so that people can also run the model from within the campus. But also we also, because Robert, maybe something I should mention that and also share experience and what we have been going through for a couple of one year. We know when we won, we are so excited. And as we are waiting to put physical infrastructure, we also found a challenge. And one of the challenges we find is that as a government institution, we also need to go through the tendering process with the government. But it also happened that an organization won to supply the devices. And because in Africa, a few organizations are putting physical infrastructure. So the guy who won the tender was thinking like it's just like an ordinary normal savers, normal machines. And then they found something different because things like GPU, which are some of it expensive for an institution to own, it was not something that is possible. And probably, Robert, I think this has contributed so many for the African, especially in Anglophone Africans to have a raise in uptake of the artificial intelligence because we have seen a lot of startups struggling so that to have a physical infrastructure or some real infrastructure for them to have simulations and also be able to run their model. So as a wrap, we appreciate the support from CIDA IDRC. We are putting a physical infrastructure within the campus and we are happy that the infrastructure will be used across the Anglophone Africa. It's not only about Tanzania or East Africa or Saudi. We are working for the Anglophone Africa. Thank you. Thank you, Jabir. That's very interesting. Monday, do you want to add from your perspective? Yeah, just as he mentioned, the government in our part of the country is also trying. We have a funding agency called the TET Fund. They also give priority to infrastructure. Like in my university, we have the ICT Center where we have some set of systems. But apart from that also, in our individual way also, if we get funding, we can also see how to also have some of this infrastructure on our own and also encourage our students, some of them that may have that a boy and they can also get their own, especially with the laptop. But just some of the feedback from this workshop, I'm seeing that what we also need to do, we need to extend hands to some of these companies because in our university, the presence of companies are not just there because I have gone to some country, you see in some countries, sometimes I see something like maybe Nokia, sometimes I see like Microsoft Lab, but in my university, precisely, we don't have such, we hope that in the future, we may also reach out to them to see if they can also support us in terms of infrastructure. Absolutely, absolutely, infrastructure and location that go hand in hand. Thank you for that. Gips, do you want to ask your question? Please go ahead. We can hear you very well, maybe it's a microphone problem, it's a lot of echo. Okay. Again? Very difficult, there's a lot of distortion, maybe the headphones, maybe? Is it better? Yeah, much better, yeah. Okay, thank you. My question goes to Mondia and Dabura, I just want to know, we know that data science and machine learning is a very applied field and usually you can get a lot of tutorials from online, but usually they don't teach you the fundamental aspect like maybe the mathematics behind all those algorithms and all those stuff. I just want to know on your own activity, are you also trying to teach the students the intuition behind the method like before using a package in art of any of those stuff? Are you also trying to give them, okay, this is the idea, this is the mathematics behind it, in order for them to have a very deep understanding of all those algorithms? Yeah, that's my question. Can I go ahead? Yes, please, Mondia, go ahead, thank you Gips. Okay, thank you for that question. Actually before now, that has been a challenge. Sometimes in academics, will you just focus on the theory without giving the answer on, which is the practical. Part of what we do is to teach them the theory and then go for the answer on. Because in the, they are majoring like, let's take for instance, my students are majoring in the field of mathematics and statistics. Therefore, I'm going to teach them more to have that busy. But somebody that is coming to apply it, we may just teach the very fairer and then teach the answer on. Just as you have asked the question, we teach them the theory also. So there is the two aspects together. Yes, the two aspects together. Thank you. Gilbert, do you want to ask you, go ahead and unmute yourself? Yeah. Yes, maybe before that, but I wanted to add on what Gips on asked. Oh yes, yes, please. Yes, yes, as part of the lab, the capacity building is one of the very high section that we are giving priority. We don't teach them just to analyze data. We teach them how to develop model, how to create algorithms, and also how to read and existing algorithms, it happened, someone is using it. So as part of the lab, we are building that capacity because we understand it's very critical for the access and the uptake of AI, especially in the Anglophone Africa. Thank you. Thank you, Jabera, for that. Gilbert, please go ahead. Yeah, thank you for the presentation. Actually, my question will go to Dr. Jabera Matagora. I wanted to ask on this issue of the AI for lab. I saw there's different section and one of it I saw was on the datasets. In Africa, but again, I don't know what is this take on this because the issue of dataset in Africa, it's really a complicated issue because it cuts across different challenges and one of it is funding. Another thing is, of course, the lack of capacity of people probably to collect the data that is required. And for us, for that element to be met, then these underlying challenges have to be addressed. So I don't know how Jabera Matagora is thinking about this because if really this has to be a success, then those issues have really to be addressed through collaboration partnership and NTC. So I don't know what is this take on this matter, but. Thank you. Yes, I think I can go, Roberto. Please go ahead, Jabera. Yes, please. Okay, thank you. Yeah, actually, you are right. As I mentioned, it's not only about the dataset. AI machine learning is not a one-man show. It's something that need a cooperative and many stakeholders. It's like a multi-stakeholder initiatives. So as part of the lab, when we started, we also brought the government on board because we also noted that you cannot have an extensive usage of the dataset if you don't have policies and the regulation in place, like personal data protection rights, policies or registrations. So we acknowledge that. And in my last presentation, I also called upon many other stakeholders to join hand with us. We are the only artificial intelligence for the development lab in the Anglophone Africa, supported by SIDA and the IDLRC. And we would appreciate to have many stakeholders joining so that we can have an unexpected impact in the field of artificial intelligence and machine learning. Because we understand this field is the cornerstone for the realization of the 40 industrial revolution. Thank you. Thank you, Jabera. Yeah, many challenges there, but hopefully working together in partnership can absolutely help on all of these fronts. Marco, Teresa, did you have any questions, anything you want to ask from our speakers so far? Please. Hello. Monde, yeah, go ahead. If you want to write something, please, from a perspective, absolutely, yeah. And some, Arindzi is asking whether what is our efforts in ensuring that BSE data science are not that should start. Like my university, we are making efforts now where we have gotten the curriculum for the data science. We want to push it further to see how it will be approved by NUC. I know that UI also, they're also making some effort also on seeing how BSE data science can also commence. Those are the efforts. Thank you. Thank you, Monde. I see a question from Matteo Marcuzzo in the chat. Do you have any issues with students and practitioners moving to different countries to look for better resources or really brain drain? In Italy, for example, we often hear about human capital flight where people with advanced training move somewhere else or maybe they go somewhere else for training and then they don't come back and you lose brain drain problem. What's your take on this? So maybe starting from Monday, brain drain, brain drain? For me, I'm not looking at data from a brain drain perspective because we are trying to struggle from Africa. If I see an offer, let's say from US, to say come over, come and learn some things, I don't mind going there. Then if the country don't want to help us out, we are willing to go out there and contribute. Just as we have mentioned, even for me to be connected to this platform, I need to have my, I buy my own data sets that are internet facility, my laptop, my own then. When I see offer from the outside world, even this one is not part of the company. This program you invited me to come and you selected me to attend. It's part of capacity building. Therefore, if I go to any country for training and even stay there to work, I'm not seeing it as a drain, but I'm seeing it as capacity building. Thank you. Absolutely, but I guess if people go for training and then they don't come back, then it becomes perhaps a drain. Jabira, what's your take on this? I actually wanted to mention on the same point that if people goes for their training and then they don't come back, then it's become a challenge. But there is an opportunity of engaging the diaspora and I'm also up to say that I'm working with the diaspora team to see how we can collaborate with the diaspora around the world who are from Tanzania. So I think that is also another angle to see that. But maybe I could also share experience from the lab. Our lab AI for the diaspora research lab has around 14 team members and the PI and the lab coordinator. The PI went abroad for his PhD and he came back to the university and also the lab coordinator went abroad for his PhD and he came back to the university. And then I also went to many other countries just for setting knowledge and we come back to the country, to the university. And then after coming back, we started collaborating as the faculty members to put together a proposal which was strong and some more competitive and then we won. Maybe a message I could mention is that even if you go abroad for knowledge searching, but it's possible also to support your home country on the knowledge perspective. Thank you. Exactly. Thank you. Thank you both. Let me now close for now the discussion but we'll reopen it later at the end where we'll bring together all the four speakers of today. For now, thank you very much Monde and Jabera for both of your excellent and very interesting presentations. For the second part of today's session, we're going to hear from the perspective from India and our next speaker is Dr. Anu Kumar Das who is an information specialist working with the Center for Studies in Science Policy at, and I'm sorry if my pronunciation here is incorrect, Jawaharlal Nehru University in India. He has a PhD from Jadavapur University in Kolkata and his research interests revolve around open science, open access and research data, digital inclusion in data science and information policies is the joint convener of Open Access India and advocacy group for promotion of open access and open science. Anup is also the co-chair of the Co-Data task group working towards a paradigm shift for open data in planning smart and resilient cities. He was a consultant to UNESCO New Daily and the Commonwealth of Learning. And we're going to learn now from him about machine learning as envisaged in India's AI policies. And I got to use slides here. Anup, please go ahead and unmute yourself while I share your screen. Welcome. Yeah, if you have, actually my bandwidth is low so can you share my slides? Yeah, I'm going to share your slides now. Can you see them? Yeah, yeah, we can see. Okay, let me go to... What is it? What screen? Here we are. Please go ahead. Okay. First of all, course directors are inviting me to give this case study on India and this is about the India's AI policies and the machine learning policies. So, exactly. So, we just started our AI policies. So, this is one website, OACD.ai, that is OACD policy observatory on artificial intelligence. So, from India, we see that 23 policies are mapped and most of them are from the government sites and few of them are from the corporations and other practitioners. So, this is the starting point when we can see that where to start about the AI policies for India. So, next slide, please. So, we just saw that we have one major policy document which is called National Strategy for Artificial Intelligence and it is hashtag AI for all and this policy document was released in June, 2018 and it is released by Nithya Aayu, one entity under government of India and this policy document focused on five sectors that are envisioned to benefit the most of from the AI in solving societal needs. So, they are mostly focusing on the sustainable development goals and which are upcoming for our different sustainability issues, particularly which are related to healthcare, agriculture or we can see that zero hunger or there's a sustainable development goal is there and education, the quality education, then we also have some goals for the smart cities, infrastructure and smart mobility and transportation. So, all these five sectors, this policy is really covering and we'll come more details about the sectors, particularly healthcare, agriculture and education or the slides. So, next slide, please. So, Nithya Aayu, this think tank also released one document called Responsible AI, approach document for India, but one principles for responsible AI and this document was released in February, 2021 and it has seven system considerations and few societal considerations. So, seven system considerations are understanding AI systems functioning for safe and reliable deployment and post-deployment can be relevant stakeholders of the AI system understand why a specific decision was made and consistency across stakeholders, incorrect decisions leading to exclusion from access to services or benefits. Yesterday, we had one presentation about the food distribution system in India, particularly one state from Tamil Nadu. So, our speaker discussed about this thing, social exclusion from the access to services or benefits. So, this is also part of the system consideration and next consideration is accountability of AI decisions then privacy risks and security risks. So, these sevens are very important considerations from the systems perspective and from the societies perspective, this document talks about that Malaysia's use of AI impact on jobs, particularly sometimes we see that many news are coming that many people will lose their jobs due to the implementation of AI and machine learning in the industries. On the other hand, some other news also telling that some more jobs will be created due to the implementation of AI and machine learning. Next slide. So, the part two of the discussion paper that is called Responsible AI, a potential point for India, Provisionalizing Principles for Responsible AI and this was released in August 2021 by Niti Ayo and it has two aspects, one for the government and government enterprises and second is for the private sector and research institutions. Come the government setup that designing ideal regulatory and policy interventions, creating awareness, accessibility and capacity building and facilitating precise procurement strategies. So, these are the operationalizing principles for the government. So, we have federal structure, we have state government as well as central government at the grassroot level, we have local government. So, all will follow this kind of operational principles for adopting responsible AI. And then for the private sector and research institution, so some of the principles are that incentivizing ethics by design, creating frameworks for complaints with relevant AI standards and guidelines and the promotion of responsible AI practices in research. So, we are having around more than 100 institutions where both private as well as public research institutions were working on the various aspects of machine learning, artificial intelligence and data science. So, they'll also adopt these principles for responsible AI. Next slide, please. So, that third discussion paper was released in this month, November 2022, and it is called Adapting the Prima, use case approach on facial recognition technology. So, this is open for the public comments until 30th November, and it provided a study of Digi Jatra program, which is facial recognition kind of program for the airports, Indian airports. So, some of the airports also carrying that prototype of the Digi Jatra and the travelers or flyers, they can download the Digi Jatra app on their smartphone and they can proceed with the boarding of the airplanes. So, there will be smooth process for the Digi Jatra users and they have to use less number of queues for getting the access to the boarding gate. So, in this particular policy discussion papers also, some examples of facial recognition technology systems deployed in India. So, that is mostly for the surveillance purpose from the law and order agencies and also from other agencies for having some public interfaces. So, that is the discussion paper which is presently seeking public comments. So, next slide please. Very recently in September 2022, UNESCO newly office, they released one 2022 state of the education report for India, Artificial Intelligence Education. So, this report is produced in collaboration with the Minister of Education of Government of India. And it also provides a different recommendation, particularly, recommendations for the AI education in India. And some of the recommendations are like, that consider the ethics of artificial intelligence in education as an utmost priority, to rapidly provide an overall regulatory framework for artificial intelligence in education, effective public private partnerships, ensure that all students and teachers have access to latest technology, expand AI literacy efforts and similar other things including the ownership of data by the students. So, they'll cover most of the areas which are for the AI education and that is not only for the school level, that is also for the tertiary level and also for the vocational level. So, all kind of educational systems will be following these recommendations which are mentioned in the UNESCO report. Next slide please. Also, there is one industry body called NASCOM, which is catering to the information technology and information service industry. They are partnering with the Government of India and they released one NASCOM Responsible AI Resource Kit in October 2022 and this is released to see the adoption of responsible AI at scale. So, upscaling the responsible AI across the country with their member industries and to harness the power of AI ethically. So, NASCOM also has different industry partners for the development of this responsible AI tool kit and it is also helping prioritizing user trust and safety. So, this way the industry will adopt the responsible AI and they're sensitizing their own stakeholders. Next slide please. Also, there is one National AI Portal which is IndiaAI.gov.in and this portal was launched on 30th May 2020 and it is a joint initiative of Minister of Electronics and University of Technology and National E-Governance Division and NASCOM Industry Body and this portal serves as a central hub for AI-related news, learning, articles, events and activities in India and beyond. So, particularly all the policy documents and all the documents related to AI and machine learning in the industry. So, those are available from this website and they also maintain social media accounts like LinkedIn and Twitter where we will get more information about the AI activities, machine learning activities in the industries across the country. Next slide please. There is one new publication which was launched in the Independence Day which is called 75 at 75, but AI start-ups were built in India. So, it showcase 75 AI start-ups and according to the economic startup India and 2021-22 India had at least one start-up in 555 distance. Most venture capital funding is now going to AI projects in the e-commerce, banking and healthcare sector. So, in some specific sectors, we are having more start-ups. And in 2021, there was funding of $1,100 million in Indian AI start-ups in the field of AI and analytics. Business analytics is also part of this group. We saw increased amount of funding from the venture capitalists as well as from the investors and cost around $1,100 million US dollars. Next slide please. So, this portal also showcased some of the start-ups, the start-up of the week and present day. It is swinelogi.di. This is one start-up. It was funded in 2018 and this is an AI based digital patient platform for preventive care from home at zero cost for users. So, it is kind of a smart mobile enabled applications where patients can see information about healthcare and they can get some information about preventive healthcare. Next slide please. Also, in the last month, we had the National Cyber Security Awareness Month and AI is also helpful for the cybersecurity. AI and smart algorithms are playing a significant role in analyzing network traffic and learning to recognize patterns that suggest various intentions. Some of the most significant applications of AI that we see developed in 2022 are likely to be in this area. So, these are the cybersecurity areas which are being focused, that is faster detection, phishing detection, network security, secure authentication and preventing online frauds. So, these are the high-focused areas and National Cyber Security Awareness Month focused on these aspects. And AI applications will be there for mitigating these issues. Next slide please. So, in the first document, the AI strategy document, we focused on the healthcare. There are many AI based applications available for the healthcare like effective high-care screening tool using AI and AI-enabled mental health care that is for improved predictability, greater accuracy, support, empathy, and 24 into 7 availability of chat boards. So, this is also promoted during the first mental health day last month. So, also we have one app or one tool to identify probable COVID-19 cases. So, this is also AI based. And this tool will combine an analysis of solicited top sounds as an objective measure along with identify the probable COVID-19 cases. Also, there are some applications like AI-powered stroke treatment and early detection software for breast cancer. All these are currently developed in India and these are being promoted by different startups. Next slide please. So, in the AI portal and also Government of India they are promoting laboratory to market initiative. So, whichever is developed at the laboratory or institutional laboratory, machine learning or AI laboratories. So, those products should be reached to the common people. So, that's why they have one lab to market initiative. So, last year they had received over 50 applications from renowned institutions like Indian Institute of Technology, Kharagpur, IIT, Dhopal, Indian Institute of Science, Bangalore and more. So, top applicants were incubated in NASCARM 10,000 startup incubation program and these are also they are seeking new applications for these initiatives and they have extended the last until 15th of November 2022. So, they are looking for the academic innovator and perfect opportunity to gain right visibility and industry exposure for your ground-making AI solutions. So, this can be any areas like healthcare, cybersecurity or other, many other areas which are mentioned in the policy document. Next one please. So, in conclusion, so we can say that responsible AI and also responsible machine learning support in such development goals and practical learning goals related to good health, quality education, clean water, zero hunger through food security and also innovation, innovation, making cities and human settlement inclusive, safe and resilient and for the climate actions like energy and environment related actions also will have some AI applications and public and private sector corporations in India are expected to undertake competitive strategies in implementing responsible AI for mitigating sustainable futures. AI ML researchers, science communicators and information professionals have been engaged in sharing knowledge resources to open access information portals. We talked about one portal in the AI. There are many other portals related to science technology and communications. So, they are also carrying different views of AI and ML applications. And these individuals or researchers also have initiated awareness raising on responsible AI besides other aspect of data-driven decision making. So, we already have some ecosystem of having the people around the responsible AI and responsible machine learning. So, we'll have more initiatives in different sectors in coming years. Thank you all for this presentation. Thank you very much for this wonderful overview of so many different aspects. We will talk more about it in a minute. So, please stay with us. We'll come back to you and the rest of the panel at the end of the next talk which is the final talk for today and for the week and the final talk is by Dr. Sharik Hasan Manazir who is a science technology and society scholar and a researcher in the area of digital democracy and exclusion in South Asia. He's currently with the Bahá'í Art Institute of Public Policy at the Indian School of Business in India and looks after the course curriculum as well as the training of senior government officials. Sharik as a PhD from Javarar Nehru University, New Delhi is a 2013 UN Volunteer Award recipient and is a member of the Science and Democracy Network at Harvard University. He's also a member of GIGANET, the Internet Society and a founding member of the Digital Inclusion Research Forum in India. And his research interests more broadly are in the Digital Governance Inclusion and Democracy Studies. And this talk today is going to be on digital democracy and exclusion biases and impact analysis of my Gov portal of the Government of India Sharik, I'm going to share your slides in a minute. I have them here. First of all, welcome. Great to have you here. Thank you for accepting your invitation. And I'll bring up your slides. Can I share my screen if that's fine? If you want to, that's fine. If that's okay with you. I can, I can. Okay, no problem. Sure. Let me see. Do I have access for that? Yes, you should. You're a co-host, aren't you? Okay. Yes, yes, you should be able to do that. Mm-hmm. Just a second. I'm kind of habitual of using Teams, so... Yeah, exactly. There's two different camps, isn't it? There's a green share button at the bottom. Can you see the share screen? Yeah, I can see. Okay. Before I share the PPT, let me thank everyone at ICTP and especially Dr. Marco, Professor Roberto, and everyone there, actually, for organizing this workshop, it's actually very rare when you see people from pure science reaching out to academic work and all the, you know, thing going around, especially in digital area. I have been working in this domain to be very honest. I'm not very satisfied with the research work that we see every day around. Most of the time, it kind of feels bad when you see a lot of reputation in the work and the lack of connect with the pure science in academic domain when we, you know, deal with a lot of research work, especially in my personal view, the whole of the machine learning and digital humanity has to one day go back to pure science and this initiative was really a refreshing thing to, and that's the reason why I thought, you know, come on, let's have an interaction, let's share some work. So I will just begin my activity. Okay. I hope you all can see my slides. Yes, we can. Yeah. So I'll be talking about digital democracy and exclusion biases. This, in my talk, I'll briefly share one of my studies of the my go portal. I'll tell you what it is in India and how the government is using the participation platform for public policy consultation and how machine learning can, you know, improve the situation in these kind of hotels in South Asia. What are the, you know, in areas where improvement is required and how the, I'll just take you through the next 15 minutes with how the public policy domain has evolved globally, academic work specially and how we have reached and where we have reached in India in terms of academia and in terms of practice. Just a second. Okay. So what you're seeing on screen is the, you know, conventional public policy making cycle. So it's basically proposed by Nekmes and Fleminger. So the, you know, system begins with agenda setting, policy discussion, policy formulation, policy acceptance and it ends with, you know, policy feedback and evaluation. It's basically a very conventional way. So when public policy academia differentiated from the public administration academia, seeking a need for a wider discourse on public policy, that's how the, from here the discourse shifted. Just a second. Over here you can see various theoretical and academic literature that came up, especially with the theoretical frameworks from across the globe that discussed about how public policy, you know, evolves, how people participation is incorporated, how technology as a tool is incorporated and at various and at what stage that technology comes into picture in public policy for people consultation and government consultation, starting from Laughville in 1956 to Berkland in 2005 and I think it's not till Berkland we have been seeing a lot of theoretical framework evolving day to day. This is a very basic evolution of the conventional public policy, you know, framework where now when people who are working with big data and machine learning now, there's a basic discussion that, you know as compared to previous time now since everything is available every information is available in real time, you know, we can do intervention at any stage of public policy formulation and we can meet the whole process more holistic, robust and beneficial for private cooperation society, government, whatever. Now, when we talk about, you know, digital platforms the debate and academic discussion and policy discussion on digital platforms, these are known basically by three, you know, there's basically three names which is used in common violence in the academic world. One is internet-based platform or digital platform, a more polished literature with a little e-participation platform, then the cyber ecosystem. All three of them, they mean the same thing. Using a study it's using study basically divided into three structure. One is the evaluation study. So most of the time we have government to private cooperation and large consulting firms, which conduct online and offline surveys and they come up with a set of reports, consultation papers, drafts saying, you know, we can do evaluation studies and these are the findings these are the consultation reports. Then you have e-governance platforms. These are the platforms and by and large we all know across every country and every government have public service delivery platforms and that's where people go from basic services to be the application you go and you apply and we all have a set of e-governance platform in every country. The third category is the e-participation platform. These are the e-participation the whole concept of e-participation comes from digital people participation in public policy formulation starting from, you know, loss well and then just forgetting okay, I'll tell you if I remember it. So the whole idea was, you know when you have people participation in public decision making starting from agoras, ancient time agoras to right now when you have when government take any decision and they have to justify particular decision they have to in a conventional legal system they used to run polling and consultations, manual consultations and as a digital platform evolved with time from e-governance to more interactive versions with more enabled machine learning and capabilities and to be very honest when I'm discussing machine learning I'm not reaching to AI when I say machine learning it's a road I might might not into AI. So e-participation platforms are basically meant for people to come across digital platform for various purposes. So if I just see, okay, so these platforms could be you know social media platforms like Twitter, Facebook where you go, these are run by private corporations then I'll share in later on slides that these governments they today have their only participation platform where they run online consultations to have involved people directly into the interaction for a quick decision making and judgment and from there the whole discourse of you know digital ethics comes into picture that how valuable how useful how fruitful that decision or when we participation comes up which we will be discussing next slide. Okay, so what is the research gap and what is the research that is available right now when we talk about you know people participate e-participation platforms so there's a lot of research work across the globe on e-governance platforms in various ways you know quantitative analysis qualitative analysis you have government ministry website content analysis you have you know research and social media platforms they fund researchers across the globe for various things then you have privately owned e-participation platform where like you know you have various e-petition platforms in India and even abroad where if you have certain cause or there's some issue which you want to discuss and you feel that you know there is no space in public domain to do you go and you do the discussion but then owned by private corporations and individuals. But when you talk about government e-participation platform there's a huge academic gap because you know it conventionally it comes under government domain so study and analysis is less first of all then to reach out to content analysis and everything it becomes a tough task so it's less accessible but then it has its own set of benefit analysis which we'll see now we talk about e-participation platform so we have it in USA, UK, Finland Germany, Australia and I'm really not sure whether it is there in Italy or not I have my language limitations but someday if there is someone who would like to collaborate we can probably have a joint study for e-participation platform in Italy if there is one okay so for now I can talk about South Asia which I have been broadly studying there is no specific people party e-participation platform which is you know being run by government in South Asia especially when we study Nepal Bhutan, Bangladesh, Sri Lanka or for the matter even Pakistan in India we have MyGov MyGov is basically a digital participation platform let me explain how it works so for example if government wants to come up with any specific policy like Anu, Professor Dr. Anu who was discussing just before me so he shared with you the AI consultation papers so these consultation papers if the government come up any department or ministry of government comes up with these consultation papers they not only go on the departmental website they also go on the MyGov platform so what person can do is if I feel that this specific part of the document has some issue or it could have been done in a better way as an academician or as a concern citizen I can simply go over there I can post the comments and then I can expect a feedback and that thing gets incorporated into the draft and that's just one example so when I was doing my studies so I started my study with you know there was this thing called Smart City Campaign in India where central government decided that though they were having different levels of smart cities in India but then it was one of the initial campaign to start and designate official smart cities in various states in India and what should be the policies for those smart cities so the concerned ministry they went to MyGov platform city wise and then they posted and they asked the concerned department of the local government to you know come up with a policy draft and post it on the platform so that the people from that city can actually suggest improvise or recommend things or if there is something which they feel should not be there they can suggest at least remove it. Now this was a good opportunity to study how people participation is actually going on you know with India being one of the largest population largest internet base and this was a good opportunity to map how you know things are evolving and how it could be a case study for a foreign country also. So what I did was it was a mixed methods approach where starting from GIS mapping, literature review of socioeconomic cost sensors then we theorize the formula how to we will calculate the volume of digital people participation on these platforms ranging from content analysis of the whole comment sheets on the over the smart city campaign going towards the textual analysis of the various response and you know the new sense of indicators mapping the gender mapping the you know the literature mapping the nature of the response usability of the comments so that's how we proceeded with the research. Now there were two public policy cycle that I showed to you one is the conventional one then the other one is where now in a common discussion most of open the AI researchers learning researchers post over here this is the you know if you can see the screen this is the how I mapped through my own research that that's how right now the public policy formulation cycle works in India where you have a designated public sphere where public interest group and political parties exist and people still go then you have digital sphere where you have government participation platform then you have private only participation platform where people go and when people post their queries these queries they culminate and they go to government in various forms and capacities to different you know layers of the government from executive, legislator, judiciary and from there it goes towards the citizen service you know portals whether it's government projects you know legacy based government projects and that's how the cycle operates now I hope you all know the person in the picture either very famous guitarist so I call it blue effect and that's how I start the digital ethics discourse so you know what starts with is the bias relevance with no offense to blue but then bias relevance exclusion accuracy and uncertainty so like I said at the beginning of my talk when we talk about machine learning and AI there are two problems the whole discourse is one it's disconnected with pure science second we think that it's going to solve all the problems of the world without mapping the complexities of the problem and these are the issues which I thought that they are the prominent one when we deal with you know government-enabled participation platforms too like you can use machine learning over there to various stages and that can improve the reliability as well as the outcome of the policies and if taken care of now okay so you know when you talk about machine learning and the ethical dilemmas the first of all is data-driven decision making now how will machine learning respond to these platforms and how can we incorporate in the whole the debate of digital platforms largely and people participation in governance and democracies through digital platforms specifically because you know if we use it in a positive way it can increase the reliability and trust of people so if I post one response that this policy draft has this issue or it can be improved in this way now if you have 10 million users even if you have like 2 or 3 comments it becomes tough for a department to respond to every comment or go through all the document and incorporate all the changes at one place and I'm talking about my go being the only portal like discussing the different portals so we can use machine learning very well if we try to incorporate these changes to make it more reliable aspect now moving to exclusion now officially theoretically speaking first of all digital exclusion is somewhat you know we have yet not defined differentiated digital divide and digital exclusion especially in the South Asian context so a larger debate and when I talk about you know this course I'm talking about majority of the academic work which does not deal with South Asia it talks about the issue of accessibility so the idea is if the digital devices are accessible probably it may be solving the problem but that's not the case when we talk about South Asian countries we have different layers of issues so if you go through the whole social identity theory which Amat Singh talks about in his work you realize that you know there are layers of issues ranging from gender demographic location age group, physical health, mental health caste, religion social barriers and these social exclusion aspects they get extrapolated over digital affairs sorry for interrupting do you think you could conclude the next couple of minutes or so we'll leave some space for questions as well I'm really sorry I'll just try to it's fine we're absolutely fine but if you could wrap up in two minutes or so we can have some space for questions and discussion as well which would be nice thank you so that's how it was then we have e-participation so the reliability of people participation where what I said is that eco-chamber is the common context understanding on digital democracy academia that you know people do participate people do suggest their recommendations but are they being considering considered in the final drafts of the quality that comes out or not is still a question now this isn't the main issue so you have digital divide you have policy, trepidation, accountability data privacy, social identity social exclusion and digital identity issues which these platforms face that can be resolved with machine learning but then machine learning has its own set of ethical issues like we discussed right now and probably we'll discuss in the question now and that's where the question is I'm not concluding my work I'm just leaving the dice open for extrapolation and discussion across the globe for everyone who's listening thank you this was another excellent set of perspectives and stimulating reflection so thank you very much for leaving it perfectly open so we can go straight into the discussion so thank you and thank you Anup as well for really interesting talks so we have about 15 minutes together and so I would like to open up the discussion to first of all to questions about the two talks we just heard but then more generally to questions about any of the issues that we've been touching upon with all of the panelists and the speakers for this session and will include also Marco and Teresa in the discussion so there was a question here in the chat start from that question for Anup what in you but obviously anybody else please feel free to contribute your perspective what in your viewpoint can countries in the global south do to accelerate machine learning or AI adoption and catch up with the global north so that's a really important really deep question let's start from Anup if you want to give us your perspective what can the global south do to catch up actually I feel that we have the talk of global south institutions there are many platforms so we can utilize those platforms for capacity building and also awareness raising and policy making AI and well related things and in India also get different ministries they are also involved in south south collaboration and also cooperation we can engage with those activities different bilateral or multilateral initiatives which are available with us also and TWAS also having some scholarships available for the global south researchers and they can visit our laboratories researching on ML and AI related areas so we can also seek some kind of collaboration through the funding from TWAS and other agencies also thank you Jobele do you want to come in yes I just want to mention that if you compare the uptake of AI in the global south with the global north sometimes you may want to give up but we are also trying to do our best so that we are able to join the rest so I just wanted to mention that the artificial intelligence for development in the anglophone Africa is also working to make sure that we are able to bridge that and we also have to collaborate with the global north in order to increase the capacity building and the uptake of artificial intelligence in African general thank you absolutely thank you and you're right we must not give up there are so many problems that sometimes we want to give up but that's absolutely what we must not do anybody else wants to come in on this question Monday or Sharik your perspective on what can the global south do and my question is it the responsibility of the global south to do something or connected to this question is the question of as been mentioned before the interplay in public administration public government but then public private sector help or support usually comes with strings attached and so that's another interesting perspective you know whatever the private sector gives usually they want something in return see first of all the whole AI research or the digital humanities research of now wherever across the globe and I'm not much south Asia and this trend has came to south Asia from global global source of academic work the privately funded research like you said Roberto you know there's a string attached to it so there's an area where you're not supposed to go there's an area where there's no benefit of going and these are the gray areas which can actually help in making the system more robust so I agree with what you said about for example in India the capacity building at various ministries at government level the things are going on but then the problem with the digital ecosystem studies and implementation is we have somewhat moved away the conventional ways of research of reaching out to problem and academic work the approach of academic work that used to be is not there right now so I'll just share one of my experiences with another one leading university we were discussing about one of the projects to do this was an informal discussion where like this is one large multinational digital company that's going to fund and then you know we have to do this and we can do this and this happens with more or less all the researchers who are dealing with and that's why you will see the implementation and applicability of AI has somewhat limited to the benefit of largely how it will help the corporations and industry and the focus is on more on that because that's from where majority of your donors are coming up so how it can be helped in a common parlance to society is not being discussed like for example Metaverse Metaverse has benefits but then there's a lot of issues which is being discussed globally right now and somewhere down the line there's no research centre which is actually ready to even talk about those issues and let's forget about you know having a fellowship to discuss and you know study those issues so there's a problem that's like in the beginning of my talk I said that academic work is required in digital sphere and if we cannot learn something from pure science at least we can learn how we can be neutral and approach the problem rather than you know jumping to conclusion just because they do nothing going on around. Thank you Monday did you have something to add from your perspective to this topic? From my perspective I think collaboration with the EPOS out between the south and the north collaboration was very important at this point just like we like I'm working with some people out there because they are out there in the south we are still collaborating we are making progress from the north collaboration is a way forward thank you thank you Jaber did you want to add anything? No I actually wanted to support what Manda is mentioning that and collaboration as Marco is also mentioning on the sustainable development I think is on 17 so it could also be included I remember a few of us mentioned that so it's sound to be as an alternative solution to bridge the gap between north and south especially on the emerging technologies thank you Yes absolutely there's a question from Teresa in particular on the echo chambers how are the echo chambers considered in the analysis of the government participation platforms and I was interested also I don't know if you were here Sharik for the talk by Petra yesterday when she talked about their collaborative crowdsourcing shall we call it approach to fair AI so this government platform seemed to be a very interesting way perhaps of doing some of that work as well depending on participation and so on so Teresa's question is around echo chambers how are that dealt with So you know the whole digital democracy aspect there's a lot of work from Ann McIntosh if you want to have a brief reference and one Moon scholar I'll just share the link if I remember his name mostly Ann McIntosh's work is there and that's where her work is mostly used in a lot of UN report you know e-governance survey and all they quote her now talking about echo chamber you know even in the normal democratic process where people participation is discussed in political science academia this whole notion of echo chamber is there now like I said when technology came into picture and the whole problem got escalated the issue of echo chamber 2 got escalated and when we talk about you know people participation platform being run by government and there's always an intention there's no doubt that we can think that you know the intention is not good but then government always wants to reach out to people and they want to have a better access to people for various policy issues now the echo chamber discussion comes into picture why because there is a sense of lack of you know credibility that has been there academically in past you know if you remember I don't know how it works in like in India conventionally there used to be a basic consultation so if there is a policy issue like if there is a policy draft on AI there is a website you go and put your comment there is my go platform you go and put your comment but before these platforms were in picture there used to be a department you like we are 50-20 people we go there they used to be around table discussion you put them and you do not know how much of your inputs were actually being considered into the debate and discussion and most importantly into the final draft now when we talk about digital platform so if I'm putting up my effort and if I'm giving going on the platform and reading the whole draft and suggesting some changes as a citizen it is my right to know what part of my suggestion was actually taken this is from people's perspective and this is a very basic understanding but from government perspective you have to understand that if you start taking every individual's perspective into consideration without keeping a track who is a specialist specific domain or who is not specialist on that particular domain and that's the whole dictomy issue comes into picture and I feel that a machine learning responsible machine learning can actually solve this issue for example if we can have instant reward for our food services delivery services if we want we can have instant reward on the suggestion that we post on these platforms but again the idea is how we can actually put it on the platform into reality so that the situation improves and I wanted to go to Anupra because we've got only a few minutes in closing I'd like to go and round the panelists once more sorry thank you very much for this very interesting answer Anupra you want to come in on this and then there's a question for you in the chat as well I'm talking about the eco chambers so we have very robust kind of independent think tanks they are also discussing on the issues related to AI and machine learning in the society particularly there is one think tank called Singapore Internet and Society they are providing different inputs for the policy documents so I think this kind of discussion at the civil society level is happening so my government is one platform which is seeking inputs from the citizens but they're not always publishing the citizens viewpoints on the platform always but we can see that in the public domain different civil society organizations are also putting their opinions views on different public policy documents so this way we can interact with the government thank you thank you and Anupra since you are you have the floor there is a question for you in the chat specifically do you face any difficulty or challenges to get the data sets related to crimes in India this is for the cyber crime methodology AI application sphere actually there is one agency called Nassal can control and they were maintenance the data sets for different states so that researchers can seek the access to the data sets and that is possible so only they are providing to the where authenticated researchers not to any individuals only the researchers they are providing the data sets so we can analyze the data from the NCRB that is important that is possible thank you and so one final question then for all of our panelists I think it's a very it's a question that would merit an entire discussion but the question is around what can be done to bridge the digital divide and exclusion biases to accelerate AI ML adoption to achieve the development goals that we've been discussing so let me go to Monday first what can be done to accelerate the bridge building okay I think what I think can accelerate all this is one of the talk was on there should be transparency there should be fairness and one of that qualities I think if those things are put in place those biases may be reduced I think so thank you Yes I think using the human centered approach could also help to address the the digital divide that they mentioned because people will have more trust on the technology and also on the you can imagine like someone developing an AI solution for healthcare so if if people who are using the technology are not engaged well then they may not even not trust the results from the technology or from the solutions so to us as a rub we think we are trying to put a human centered design at the center and also engaging the participatory approach thank you also very much a question of trust Anup what can be done telegraphic answer if I may ask you Data security policies which are to the European that policy data policy and also national level we can have data security and data protection policy for the end users that's why you can build up other aspects of responsible innovation and responsible research in AI and ML thank you thank you you have the final word what can be done to be very honest it's a topic of another discussion it's a very it can't be summarized in one word but then also better research and better global cooperation you know cooperation and coordination and positive approach and trust like you said thank you for that and I think this cooperation, collaboration, discussion, sharing starts here what we have been doing today what we've been doing for the whole week thank you all thank you to our panelists for bringing in their expertise experience and perspectives thank you to the participants for attending but not just attending but really participating for co-creating space this workshop I've been really struck by how much I've learned obviously but also by how much collaboration, discussion, eagerness to understand it's been really apparent that we've been able to create something to me very very special very very interesting thank you all I would like to ask Teresa first and then Marco to have a final word of goodbye and closure of the workshop Teresa just briefly I would like to thank you for involving me in the organizations and all participants and speakers I've learned a lot so I hope and look forward to other meetings like these and have other opportunities to collaborate and keep in touch thank you Teresa Marco so yeah thank you all it was a great panel thank you Roberto for sharing that and a word of the issue of data, local data has been raised in today's panel and please don't forget that ICTP can help in facilitating the exchange so when you're talking from Tanzania and from Nigeria having the same issues I always think it would be great for people to collaborate once they meet here at the CTP or virtually at the CTP so please let the continue the discussion and let's see if there is space for collaboration thank you very much and hope to see you in the next activity thank you Marco for leading this and ICTP for hosting us it's been really interesting and I very much hope these connections that we have been starting to make will strengthen, will continue so thank you all this ends the event ends the workshop and hopefully we'll have other opportunities to continue this important really interesting discussion thank you thank you thank you so much thank you so much bye bye bye bye