Breakthrough Days: Food Revolution + Evolution of AI for Good
12:02PM Sep 24, 2020
AI Good afternoon. Good evening and welcome to the AI for Good Global Summit all here always online. My name is senior frontend from the ITU, the International Telecommunication Union. And I have the privilege of introducing day three of the AI for Good breakthrough days. ITU is the United Nations specialized agency for information and communication technologies. And we are also the organizer of the AI for Good Global Summit, alongside Express foundation and in partnership with 36 un sister agencies, ACM and Switzerland, our co convener. The goal of the summit is to identify practical applications of AI to advance the Sustainable Development Goals, and scale those solutions for global impact. Like the most of the world, the AI for Good Summit has gone digital with a weekly programming allowing us to reach even more people across the globe. The heart of the summit has been always organized around the breakthrough days, and this year is no exception. We are moving forward with breakthrough days presenting a series of all star keynotes, interactive workshops, proposals and key announcements. And of course, we are counting on your participation to create very engaging discussions. Before I introduce today's moderator, let me go over quickly some housekeeping rules. If you wish to ask a question, please use the q&a tab, the moderator will select and read out the questions to the audience. And I have the first challenge for you. Could you please let us know where you are calling from which country or which city and simply use the chat function to communicate and make sure to enable it to everyone open panelists and attendees. And let me do first so I'm calling from Geneva. And we have participants calling from Los Angeles Lismore, Geneva to nice when Osiris, Iraq, India, Hong Kong Mexico. Fantastic. Welcome, everyone. Now I'm pleased to introduce today's moderator and Ubuntu for me who is Chief Innovation Officer at Express and the AI for Good program committee chair. And you're welcome and the floor is yours.
Thank you very much semia for this presentation, and it's my pleasure to welcome all of you to the year for two days. We have a very interesting program for you today. And it's my pleasure to introduce the first keynote speakers of today. Let me introduce and share with you the biography of many of Harbor was our first keynote. Emmanuel Faber is the chairman and CEO of Don, under his vision leadership that is today a global food company and the world's leader in dairy and plant based foods. Then in North America, is the B Corporation in the world today. Amir feber is strongly committed to primitive and inclusive business model in 2005. He supervised the first social business experiment acted in Bangladesh with Grameen Bank, and the creation of done on committees in close collaboration with the 2006 Nobel Prize winner Muhammad Yunus. Since 2019, Emily feber spearheads the business of inclusive growth called before he launched in August 2019. In connection with the g7 leaders before it is a first of its kind coalition of 40 leading global companies and partners committed to tackling inequalities, and promoted inclusive growth sponsored by the French President Emmanuel Macron. And coordinated with the OECD. In parallel, and many of feber activity contributes to building the global business movement for biodiversity as such, launched in 2019, at the United Nations Climate Action summit in New York, the one planet business for biodiversity coalition, which he call is with the world Business Council for sustainable development, gathers 19 leading companies joining forces to step up out of alternative farming practices and protect biodiversity for the benefit of people and planet. Please join me to welcome him in your favour. Good afternoon, Mr. Ferber. We're very pleased to meet you and this is a true pleasure to welcome you today. I learned that we have probably grown in the same city of Granada. So that was a happy and interesting fact for me when I was preparing again, we have been following Dan's work for some time. To give some perspective. Mr. Ferber, we have been working towards meeting the Sustainable Development Goals by 2030. And during this summit, we have been designing and building the summit about the impact that technologies such as artificial intelligence can help us achieving the goals. So while we work on all SDGs, we have consistently been focusing on three of them as digital number two, which is fighting poverty SDG number three, good health and well being and SDG 12, responsible consumption and production, and not forgetting and most importantly, the collaboration for the goals. We have this year specifically with a specific emphasis on future food, gender equality, and collective use of intelligence, to build resilience. I'm sure these topics are top of mind for you, as well. So coming to the future of food today, you have a very unique perspective and the global food system. You're often cited for new Food Revolution. Let me maybe to start with your keynote on this topic and understand better what Food Revolution means. And maybe open up a conversation as we take questions as well to evaluate how we can all participate in a collective goal to meet those assistance level goals for people in plant.
Thank you for the nice words. And it's good to hear the fact that you are in Grenoble as well, which continues to be for me, a source of inspiration when it comes to my my relationship to mountains and and the verticals there. Thank you also for the kind words, I'd like to really start by commending both XPrize and the ITU for organizing these, these events, I think they they are taking place at the moment where they are fundamentally needed. Bringing thought leadership and collective thinking and creativeness around the topics that I understand you focused on for the current breakthrough days, which are about the pandemic food and gender. What struck me when I, you know, was proposed to participate was actually that, for me, I cannot answer about food without actually also answering about pandemic and gender. And so I think that that triangle of topics that nourish the SDGs, as, as you just said, are fundamental for the discussion. So maybe let me elaborate. And you will see through this, what I think we mean, when we speak about the Food Revolution, which certainly has been accelerated by the current pandemic, I'll start by saying that when it comes to the pandemic, there are many ways to to look at it, basically, you can say, you know, that's a crisis, we are going to constrain supply through confinement, but also to constrain demand. And the day we will burn, it will restart the game. And there are many, you know, political leaders, obviously, that are eager to make sure that economy restarts and to make sure that, you know, the social crisis, the democracy crisis, will be avoided by a quick restart. I think unfortunately, that it's, it's unlikely to be and to create resilience, because there is a I think, another way of looking at it. The The reason this virus has been in touch with us is also related to the way we live and the way we've been producing our food for so many years. You know, you could think about, you know, thousands of years ago, probably, and hundred years ago, actually were some local habitats would be containing viruses that our species is not prepared to face, but they would remain somewhere in the forest or in the deep wildness, where we are not. But two things happen. First of all, we cut trees to grow massive monocropping agriculture all around the world, that has basically reduced the natural habitus for these forms of life that we are not used to and accustomed to. The second is that we have gathered in cities. So today 60% of humankind, leaving only 2% of the surface of the planet, which means that when something happened there, it happens to a lot of people Because of the density. So that's the first reason why we were in contact with this virus, and that it became a problem locally. But then it became a bigger problem and actually a global problem. Because the virus traveled with our technology, it troubles with our ways of life, taking the trains, and you know, the suburbs, and suddenly the ships, and you know, the airplanes that we're using ourselves. So this pandemic is really something that is related to the way we live, the way we organize ourselves, basically not considering nature as being fundamentally an ally, but either neutral to it, or destroying it. I think, if you think about this, and now turn to the second point, which is central to our discussion, today, food, the same applies, you know, we are only relying now for the whole of humankind on six species of plants for 75% of our total calorie intake.
And that's a huge dependency on only a super small fraction of what nature has to offer to us. And we did that. Because we wanted super big effects of scale, we wanted to focus on the simplest more effective short term plants, the more the less costly plants to grow, to bring that food to as many people as possible in an affordable manner. And yes, that's true. That's what happened. But we didn't see all the unintended collateral consequences, which is this total collapse of not only actually the wild biodiversity, but the cultivated, and the domesticated biodiversity. There are studies in the US, that show that within the whole, I think nine or 10 million population of cows, if you look at their DNA, they all track down 99% of them track down to only two individuals, not even speaking about species here. But individuals that were super strong bulls of the 70s. And through artificial insemination, they gave birth, not to traditionally what was like 100, but hundred thousands of descendants. So that now the lack of diversity is incredibly strong in that herd. And the same applies all across the world, it's very clear that if you only raise chicken, that are the best chickens to cut them into pieces, and, you know, makes the right thing that we find in our supermarket for barbecues. And you raise hundreds of thousands of them into very limited spaces, and they're all the same species. Well, when there is a virus coming, that's a pandemic. If you think about the lack of biodiversity in the wild nature, it's all only only a reflection of the lack of biodiversity in agriculture. And so there is a deep need to re inject biodiversity because this is the only way that we will restore soil health monocropping. And this model focus has ruined soil health all over the place, which means over irrigation, overuse of insurance changes in many things, but certainly not something which where we are building with nature. So it's absolutely fundamental that we access to that. And it's not a matter of food safety anymore. It's for me a matter of food security, the resilience of the food system.
Now, let me link this with gender.
There are, I think 1.7 billion women that live in rural areas, they produce 50% of the worldwide food production. In Asia, actually, these numbers are even bigger. 60% in Africa, in particular sub Saharan Africa, it's 80% of the food that these rural women are producing. They only own 2% of the overall land
that is used by agriculture 2%.
And they only get today 5% less than 5% of the public aid to agriculture and training. Now, if you think about food security, there are obvious studies that show that in Sub Saharan African countries, if you would swap The role of a man and the role of a woman on an arable land on a farm, you would increase productivity by 20%. If you would reduce the amount of time that women are cooking, or going out to find wood, because of deforestation more and more time that they need, like three, four hours a day, the same to go and fetch water, and instead reduce this and bring them into the field, you'd increase the overall rural household ability to generate 50% more
And we shouldn't think about these 1.7 billion women, as in any way going to be reduced, as in OECD countries where, you know, agriculture workforce has reduced to be only a fraction of the total population, because in Sub Saharan Africa, as well as in India, there will be hundreds of million more rural population in 20 years from now than today, out of the 2 billion additional inhabitants that will come from these two sub continents, several hundreds more of rural inhabitants. So they are here to stay. And there is no way that we can obtain food security at the global level, without focusing on family farming on these areas, and in particular, on women. So why all of this is relevant to an AI for Good event? Well, last year, my colleague Cecile, on behalf of Diana was participating in the event and saying, AI has to be a common good. And I think for the unknown, it's very clear. AI can be the best or the worst for humankind, basically. And I think you guys know that much better than I do. So how do we ensure that there is a governance of data governance of AI, that is empowering people and not enslaving people? If we do that, then AI can obviously do wonders for the food security topic that I just mentioned, reducing food waste, which we would reduce deforestation, support regenerative agriculture, for instance, then honors decided that co2 was not an externality. For us. It's the resilience of agriculture, because it's not only the emissions, but it's actually what agriculture can sink in the soil. Because every culture has this unique ability to be a carbon sinking activity. If we do it right with regenerative agriculture. To do this, we need to measure the co2 footprint on the land used by the farmers upstream our activities in a number of cases in our livelihoods fund that is working on regenerative agriculture, upstream our activities, we are using technology, we are using drones to measure the size of the trees the size of the mangrove the size of the greens that are producing the co2 footprint or reducing the co2 footprint of these rural areas. And that are then being used to transfer and allow us to report our co2 footprint as a company. And by the way, for the first time ever. Last year, we published a an earning per share post carbon footprint, the whole cycle of carbon around and not only our direct emissions, but all the emission coming from agriculture. Interestingly enough, we have picked our carbon emissions, absolute emissions, five years in advance of our plan. And now it means that with this carbon charge accounting charge going down, instead of going up, the earning per share of done on post carbon will grow faster than the earning per share of the non According to the International gaps. So for this, we need technology. One aspect to this is that that needs to be critically worked on is the availability of that technology in rural areas. And frankly, I'm happy to have IQ here with us because when not, you know spending quite a bit of time in particular in Africa over the last several years but you mentioned Bangladesh and India for instance. It's very clear that there is a lack of connectivity, the the lack of infrastructure, we will never get the data You know, if you think about what what tech and AI in particular could help when it comes to index insurance on weather, weather forecasts, market price dynamics for farmers that would avoid them to, you know, go and travel for two days to the market and not being able to realize the price that they want me to realize
the data on regenerative agriculture, that could be used even for creating public policies on those topics, they need an infrastructure that can capture this data. And today, still, in particular, in Sub Saharan Africa, there is a big missing backbone of connectivity, whether these are the optic fibers or the satellite systems, I think there is an urgency that the in particular the broadband commission, which where there is itu and UNESCO together for sustainable development, is actually ensuring that funds are being directed to create the infrastructure in these old rural areas where these family farmers are producing 50% of the world production of food. And they need the connectivity to have access to the data that will bring in and nourish the AI systems. So that's clearly for me, one of the breakthroughs that need to happen if you if you look at these numbers in Africa, it's very clear that you know in terms of bits per second, or in terms of the cost of accessing internet for users in in Africa, compared to what it is in, in OECD countries, for instance, that has a huge gap, that and that creates a barrier for accessing this very important part of the world where when we are speaking about food security that derives issues like migration that derives issues like democracies, and therefore the rise the issues of the whole resilience of how we recover from the current crisis, but also from the decades where we've been creating a food system that, in many ways is broken, is not able to even address the climate change adaptation that it needs to address the super fast. Among the topics I'd like to maybe share is that I, I think one thing that AI can really do for us is, by the simple brute force of its data power. And the ability to treat so many data in the same time is to help us on the biodiversity and the support of biodiversity. When I speak biodiversity, I'm talking here about agriculture, biodiversity, for nutritious food in a number of ways, we have grown food, where the cost of calorie has decreased. But the content of calorie, its qualitative content has decreased as well. There is less calcium today, in in the milk than 50 years ago, there is less vitamin C in oranges than 50 years ago, because we've chosen species that grow faster, because it's more efficient, it's less costly, it goes faster to the market, but it doesn't take the sun as it did before. And it doesn't charge itself within us written in C. So in many juices of orange, you add on artificial vitamin C. So the importance is, how sustainable is the agricultural model? And how sustainable is the nutrition model that go together. That's the core of the one planet, One Health framework of action of them. And I think that is the fundamental response to this COVID crisis when we see that unfortunately, there are huge commodity BDT factors of obesity and diabetes against the COVID virus. weakness in terms of health conditions, and the statistically bigger morbidity for people that suffer these non communicable diseases, which are related to their diets. So I'll just use one example before I conclude and then we can go for questions, commands, challenges, and many things. We, we are super happy to be partners of a startup, which I just want to say you know a few words about, that's a company called bright seed in the west part of, of the US. What they do with their model is that instead of Focusing on these few species that I mentioned, they observed the reality, which is that we are only we only know 1% of the overall plant kingdom. So their system, which they call for adjure, for good reasons, is basically exploring super fast, super deep the 99% that we don't know.
On the other side, that's the planet side. On the other side, on the health side, they have scanned with data, entirely new ways of understanding the metabolics of human health. And the relation between the human condition health condition and diets. And where the real power of thing is, is when you connect both, because these are they are able to find new applications, new, nutritious, and benefits to plants that exist. And they are also able to map on the needs nutrition needs, plants that we have no idea about in terms of agriculture, and that are therefore discovered as a potential solution for the future. So a small startup like this one is leveraging the power of AI to create the biodiversity in the fields that we need to address tomorrow, the challenges of food, which for me are not about quantities. It's really about making sure that the right food is at the right place. I don't think and if I'm back for one second, in my conclusion about AI, I don't think we should look at AI from the top. You know, we we I don't know who is we but we the elite control AI and AI basically, you know, giving it or selling it or renting it to the people for the good. We feed the world. No, I don't think large multinational companies like mine, feed the world. I think people feed themselves. And we are here to serve them. And that's the whole theme of the Food Revolution. The Food Revolution is that people realize that in many ways, they have delegated to large brands, large banners, hypermarkets, the responsibility to provide them safe, diverse, nutritious, affordable food, and suddenly discover that this model has limits that we are hurting some of its limits, climate, agriculture, irrigation, many things. And therefore people are back to, I want to be suffering, I want my food sovereignty. That is what they're claiming, in a way. They're claiming that what the UN is called the right to food, they want that. And the choice is for us to serve that revolution, or to resist it. No, no, no, I want to stay in control, or Okay, I'm going to see how I can serve you. That's a totally different attitude. And it leads me to my very final word, which I think that it's high time to read localize the food system. There are many pressures for that, whether they are political, they will be tariff wars, there will be economic wars, there will be trade wars, there will be you know, government sovereignty decisions. There will be environmental decisions, climate change, adaptation, all of this, and People's Choice will be to re localize. And I think we need to go there, because one of the aspects today of the food system is that it's overly reliant on a few very complex supply chains. And if there is one problem there, there will be a problem everywhere. So the food sovereignty, and resilience will come from the fact that local food traditions should be seems the diets, the recipes, the ways of cooking in all these different regions of the worlds locally as a treasure that we should continue to nourish, because it is creating the diversity that will make the system resistant. Because if there is a problem somewhere, there won't be a problem everywhere. And I'm back to square one with this pandemic. And the fact that it is local will mean as well, that it will be numeration backed by an agricultural model that will precisely fit what environmentally, the planet is able to deliver locally. So it's not the end of exchange. It's not the end of the trade, but it's restarting from the base. That local is fundamental. And I think the same should apply to AI in my mind, and I'm ready to discuss that with you. Thank you for your attention. Thanks for inviting me.
Thank you very much for this insightful ideas and also the vision that you portrayed. I'd like to talk about three different aspects of what you shared with us. One is the we talk about sovereignity of food. In your mind, this sovereignty is it to be left to governments, to organizations or to the individual. And if he has to be left to the videos, what are the ingredients and the mechanics of the frameworks that needs to be in place so that everyone is well informed, able to participate in the decisions around the food system, and the flow of nutrition from one plus two, or benefit from the global economy, and also not be left out with social injustice related to everything that you mentioned? So is there a framework or a set of guidelines that you have in mind that could help bring sovereignty to the lowest granule level that we can afford to have?
Thanks for the question. And I think it's got to be at the individual level. And basically, that is what you see, frankly, all around the world, the Food Revolution I'm talking about is not only the one from an elite that can afford buying organic food in your natural channels. And, yes, that is there. But it's way, way beyond that, it's interesting to see that there are the same patterns for the million years and even more the gen Zed generation, this the pattern when it comes to food topics, is much more common in Indonesia, and Mexico, Russia, the US and some European countries than it is with in each of these countries, with their parents and their grandparents. But what happens is with these young people is that as they're masters of social media, they create social norms. So when we transition, and we have to transition out of it to too much poverty, I mean, animal protein based diets overall, in the in particular, in the developed, quote, unquote, world, we have to transition out of that and be more flexible, we know that for the planet boundaries for the planetary diets, they are driving this because they speak to their friends and their families. And they convince and they are militants. So the changing the mix, things happen. Without the government without the large companies, they operate on their own, and they choose what they want to choose. If you think about these apps that are all over the place to guide consumers in their choice, they're not ruled by governments. And actually, that's one of the topics By the way, which is that there are you know, a bit of, frankly, garbage in garbage out, because not all these apps are fully reliable in particular, when it comes to facts and science and everything else. So we are still at the nation, part of this industry, which is a very important industry, which is how AI how tech is supporting consumers choice, exactly what you say educating people. So we believe it's a it's got to be a function of both. Making sure that the front of the back is clear that there is access on data for people who choose their products, easily. But also that we create a level playing field. And that's a work that is being done by government. So for instance, in Europe, we have a system called Nutri score, it was started and piloted by the French government, then on was the first company to you know, fresh dairy and plant based business to start it with alpo and Don brands and a number of others, then we extended it to water. And then now we are extending it to the rest of Europe as Nutri score is going to be the system in Europe. So I think you have to have both and one needs to guide the other. At the end of the day. All of this has to be collective. We can't succeed on our own. I'd like to take another example maybe if I can for a few second. We've been working in this coalition that you mentioned that we launched exactly a year ago at the UN one planet business biodiversity which is a 25 of the largest company that are using products extracted from the soil. So it can be textile, cosmetics, food, hygiene, household etc. The we've been working as part of this coalition to reintroduce biodiversity in agriculture, with a startup called How good How good is super interesting because they are basically linking, again, the formulation and the nutrition that is derived from using such and such ingredients, but also the social topics that you are talking about, you know, I've, I've said already that I thought that food was a fundamental right, and that market economy could not be resilient, if it does not produce social justice. So I'm entirely on this as well. And the fact is that how good is also tracking human rights in the, in the supply chains, an issue that are related to any given specific ingredient. So they're able to map that and with data to shake a profile of a product. So when you scan it on the shelf, you basically have your trade offs, you know, some would be better for the planet, not as good on the social side. So what you know, what do I want to do as as a consumer, and therefore, all of these, which are going to help people to ask themselves the question is fundamental, because, again, the old model has disconnected people from their food and from agriculture.
vision of a global traceable food system with enough data that would allow making those trade offs, how do you balance this global system with the concept of being local, and nurturing local ecosystems?
Yes, I think that it's like for us at done on, we are a company that we relies on local. So I don't I think 95% of what we produce is actually sold in the same country. A few things are being exported, but most of it is locally produced, locally consume, it does not mean that we don't have one company language. So we we are aiming at having one data system, because that is the language. And so as long as we're not talking about or tasik systems, but autonomous systems, it's like, you know, a neural network, you have these cells, but they speak all together, and they need to speak the same language. And that's where the data is for me. So the data interconnection should not be a barrier to have a locally routed system. But it's, it would be a way to elevate the ability of being more knowledgeable, transparent, and overall more efficient, in the way we speak about our food and the way we design it.
Um, I would like to ask you questions about an important player in the chain, which is a local farmer, the small farmer, and talking about AI and machine learning how to see technology be an enabler in the service of the small farmer, and could be about inclusion, social impact, economic contribution, or generally speaking their place in the overall chain, how this small farmers can be empowered with, with the advances of machine learning and AI, and have another question is about how you can promote more AI contributions. But this this small farmer, I think, is a is an important player in the chain.
It's a huge question. I mean, I mean, if I had the the silver bullet answer, um, you know, no one has the answer to that. It's a very complex and very fundamental question. And when you when you speak about, you know, small family farming in many ways, the truth exists, I mean, that exists as well in OECD countries. So, you know, as far as that is concerned, we were collecting our milk directly from farmers all around the world. So, and their family farms, some of them are farming with us for two, three generations, some more recently. And it's like, you know, a rotation every week or doing twice a week were visiting them. So we we know, in our own industry, there are so many others, but we we know the reality of the family farming, whether that's in Africa, or again, in Ukraine, or Russia or UK, whatever, us and Argentina, etc. The, the The fact is that there is evidence that these people are incredibly, incredibly, I insist, knowledgeable, willing to learn, change, and improve. I've been so impressed. But then the question is not so much the technology I remember a discussion I had. And I think with Muhammad Yunus who who started Grameen Bank, and we've been pioneering bioneers, together in Bangladesh, and on this small locally produced fortified yogurt that was sold in the villages with a super small factory that we had built in the northern part of Bangladesh. The, you know, when when the mobile phone technology was launched, I think in 2000, or something in, in Bangladesh, people immediately mentioned Oh, well, that's a great technology for the elite in Dhaka, and Chittagong and a couple of other big cities. But you know, scammin, they said, well, give me the license, give me one license, and I will make it work for people that are in the villages, the sub, the 60,000, villages of Bangladesh. And people often said, ha, they don't even know how to read and count. So when he said, you know, if, if they're living, you know, will rely on the fact that they are able to read the 10 figures, the 10 digits, that are going to be there, I can tell you in one week, they will learn the digits, and they will know how to put the numbers etc. And they started grabbing form like this, by having Grameen ladies being ranting second, by second minute by minute, these phones to the rest of their villagers with connected with a solar panels for recharging the battery. And indeed, these illiterate women became business people that were just trading time on the phones. And I've seen that with my own eyes visiting all these villages. And that changed the lives of these people. And I was, you know, talking about the connectivity in Africa, because I saw the same there, you know, you all know, m pesa, the, you know, the money transfer system on mobile in, in Kenya that was started by DFID, by Vodafone by, you know, many people Safari calm and the government and others. It's incredible, because it just changed the way people were using money and made things so much safer. And some, you know, easier for farmers to be connected to their bank instead of having to travel when were there other to the Midtown, etc, etc. So the, we should never think about this farmers are being people that need our help. They I mean, frankly, I don't think they do, you know, we saw in araku Valley in India, you know, people just by reorganizing their, the footprint of their one actor, you know, growing trees under cover and plants under cover and starting new things they generated for themselves just by the hard work and being prepared to change their agricultural practices to bring significant cost down and revenue up. The question is, do we get
infrastructures multilateral aid, government money for these farmers or not? I was mentioning the fact that women would only get 5%, not even 5%. Whereas there are like 80% of the workforce in Africa have the public aid for agriculture. But the reality is that there is very little public aid for farming, family farming in Africa, you find public aid, I mean, international aid from the financial ifd. Okay. grieving in Germany, DFID, USAID, for bigger things. But when it really comes about going deep in the communities and offering training, it's a whole theory of change that is so difficult to track that no one is actually putting the money where it should be. And so I think, again, I don't have any magic thing, because it really starts from considering that these people are autonomous, they know what then they are prepared to chance to improve themselves. They need support. I, I can't say more than this. I mean, I could be much longer on this I can unconscious are probably too long. So I'll stop here.
And we'll come back to that. We have a lot of questions. So I would like to take a few questions from the q&a section to be able to at least get get some of these interesting questions presented to you. And we'll take short questions, short answers and demo them.
So let's just read some of you.
The first one is that do you see but and you answered that somehow a little bit earlier. But do you see potential for blockchain solutions in traceability of foods, organic pesticides free etc and do consumers want more information on where the food comes from, and work processes used.
So the super short answer to these questions is yes. And yes, I can elaborate more. But that's that's what it is. I think that blockchain will take a lot of time. But, you know, who knows the how technology can actually here, be hyperbolic. But yes, it's definitely part of what transparency can bring, and people will want and will need, and we all need more transparency.
Another question is that you consider and also is done in looking into AI and robotics and vertical farming to reduce strain on natural environment and localize the food supply chain?
Yes, we are. We we have set five years ago now. We've created a venture capital fund, called Denon manifesto ventures, which is investing in and supporting some of the companies I just described. Actually, I just mentioned the startups. And you know, it doesn't have a kind of startups, including agriculture, including vertical agriculture, pandemic, or agriculture, urban agriculture, because we believe there are bricks of solutions for tomorrow. They may not bring I mean, they won't be the solution. But for me, they are clearly one of the solutions that we need to develop indeed, yes.
Thank you. Um, another interesting question, which is, what is your personal view about the dairy industry's impact on global emissions, currently? And what is your framework for seeking solutions?
Yep. So a
couple of things here, and I'll try to be very short one is agriculture accounts for the same co2 emissions net than the whole of industry. The difference is that agriculture can sink carbon back into soil, and we need to come up with solutions like this. The second is that, of course, dairy has a carbon footprint that is high. But there are many ways to improve it in particular nutrition. Cows are ruminants and if you give them the compounds of soil that have become the totally non biodiverse kind of nutrition for cows, today, they emit a lot of methane. If you given linseed and grass, you can divide by five, the carbon emissions or the methane emissions of cows simply because they are built to eat grass, not grams of compounds of soya and maize, etc. So the nutrition is critical. And we are back to the question of diversity. The third aspect is that we believe that people will need to become flexitarians. And we are pushing hard. We are the largest plant based alternative to dairy business in the world, we acquired white wave silk in the US alpro in in Europe and a number of other brands four years ago, it's been a mega deal for us. But we know that we we need to shift and support people shifting their diets. Is that going to be the end of dairy? My simple answer is no. Because there is no way in the next 30 years that there will be farming without raising animals for many, many reasons, that does not fly. So it's got to be a totally different kind of raising animals and taking care of animals. But there will be animal farming for a long, long time. That will need to be reinvented the way we deal with that.
super interesting. And I think we don't necessarily have the vision that you have the knowledge about the length of time that it takes to transform habits or farming or agriculture. Talking about agriculture, there's a questions about in your opinion. And also it seems that one of the biggest challenges of using a agriculture is the lack of technological training and digital skills. And while we don't necessarily have to have everybody be trained, how can we overcome that challenge? Well, Krishna is clear.
it's it's a it's a complex decision for a guy that does yogurt. I'm not I'm not an expert in AI and digital. In many ways, yes. I think it's, it's back to what I said about training. I mean, I don't think that, I mean, put it this way. No one would be using AI based applications. I mean, I'm using 45, you know, and Spotify is bringing me new lists of music every week that are better because I'm training without even knowing it this, this this app. And so I don't think that the farmers need to be, you know, super involved in AI and tech, the need to have apps that are allowing them to be the right interconnection with the big AI system that's behind. But I'm sorry, it's like a general answer.
When I think that's an important one, because from the investor perspective, you want to simplify usage of technology, not making it super difficult. And we have been using this term of problem owners versus problem solvers. And and if the problems are well identified, and to the core, it's easier to align engineers, scientists and technology providers to help with those problems. If we don't always have this, this this wireframe doubt. I'm going to go to two more questions briefly. The first question is that, are there any initiatives done on is undertaking to support women's health and economic empowerment? So they can get financial backing credit and support from investors to increase their chance of being truly sustainable entrepreneurs and not subsidized? You mentioned the the work that you've done with Muhammad Yunus, which we had the guest talking about the global data pledge, but borrowing from that experience and your overall experience, how do you see empowering woman? And specifically woman's in, in the farming industry or in the food system?
Yes. Well, I'll give you an answer to this in the sense that we have created 10 years ago, an endowment fund taking profits from the business with the support of our shareholders to work on capacity building, for the most vulnerable members of the broader drug ecosystem have done so around the not not our employees, but the people in the farms, the scavengers in the land fields, the street vendors in Mexico or you know, Britain is IRS to down as well as the farmers as I mentioned before with the funds in, in tropical zone, Africa, Indonesia, Bangladesh, India, Guatemala to dump all of them are focused on women. And the the experiments with Muhammad Yunus in Grameen was focused on women is focused on women, the people that are distributing the product I've been in the in the villages have been trained about the nutritious aspect of this small yogurt, and they are going door to door as entrepreneurs to sell these nutritious food to the families surrounding in the villages. so empowering women, to bring them economic autonomy has been the driver behind the effort that for the last 1015 years, we've been doing to work on the food ecosystem around a rundown on
let me go to one more question. This one is about the role of harmful pesticides on agriculture, not only in developed nations, but also developing countries that have been known to be cancer causing, and how is the model taking that into account if we really want to focus on food security, since they're all interlinked and have tremendous impact on the food chain trust environment? You did mention that in your in your presentation. Is there any short takeaway, or quick soundbite answer that we can have a good understanding of this topic?
Yes, it's a complex topic in terms of transition. But let me say the following. We are only focused on a few plants and within a few plants, a few species, varieties and species because we felt they were bringing the best yields efficiently. That led to as I just said, mono cropping all over the place. So we have basically decided that we would not look at nature based solution. There are many, many, many ways of getting away from most pesticides. But by changing the seeds that you're using, by mixing what we call toxic co ecology and socio ecology Which is that? If you put plants together, they are protecting each other against pests. It's very clear, between, for instance, carrot and leek, put them together, they protect each other in the soil and out of the soil. So that's a very simple example. But there are thousands of such examples. So what we need, what needs to happen is that governments need to stop subsidizing the use of chemical insurance and start subsidizing agricultural practices that are regenerating, that are recreating soil health, because the plants would be more powerful if there were more new natural nutrients in the soil and forever not because you bring another cycle of nitrogen with your artificial nitrogen being in involved there, because we've been ruining the soil by taking the carbon out of it. So it's a full system. If you go to agro ecological systems and regenerative agriculture, you will not need by far the quantity of pesticides that are being used today. Can we go entirely out of pesticides? I you know, I don't know, but so much that we can, you know, avoid that. I think that's absolutely where we should be going.
Thank you. Thank you very, very clear answer. I like to ask an important question. I'm trying to be clear, when I asked this question. You mentioned you have a founder and endowment and you investing to not only start up, but also initiatives, if you had the power today to or imagine that you would be investing in not only technology, AI is one of them for the common good. And balance, common good and economic growth. And knowing that the best investments are the one that are supported by other investors and governments to make them sustainable and continued. Right, systemically, what would you invest into that?
You cannot ask me this question. I mean, it's a it's, it's, it's an impossible question to answer. Because there are so much if and, you know, given that and etc. What I what I would certainly say is that I, yeah, no, I truly believe in the power of collective. That's maybe that's the only thing I should I should say, the power of collective that, because none of the difficult problems that we've been talking for and that you are addressing with these days have been sold by the market, by economy, the private sector, none of them has been sold by government known have been sold by NGOs, otherwise, you would not be discussing them, they will be solved. So it's very clear for me that the solution has got to be a collective Coalition's basically. So that's one and whatever, whatever the topics we've been discussing, none of that can then on do either own. No, absolutely not. We need to work with many others. And the second aspect I would bring is the diversity, the diversity of Coalition's but I would also say for me, I would invest in the diversity of AI systems, and making sure that, you know, maybe it's a challenge for all the community around this, this discussion. The worst thing for me that could happen to humankind is that AI is actually standardizing through data, the languages that we're using, and if you standardize the language, you're sterilizing the thought, because the language creates the thought. So how can you ensure that you design a system that accept diversity that accepts exceptions that accept to be self disrupted? Because that's the only way to maintain it in the service of life? So if I was going to invest in one aspect of it, that's the one because I haven't seen it cracked, and I'm sure it has not. But that's for me an underlying topic that if AI is successful, you know, say 10 years from now, if we do not embed that ability to respect diversity, then we run a big risk of having the sort of the unique thought because everyone's thinking with the same language. So, you know, I, I'll give you that back to you and I'm prepared to invest in systems that that would prevent that from happening.
Thank you very much for this answer. And no, it was not an easy question because it's very large and has multiple lives. into it. But I was hoping that you were talking about a collaboration. We talked about SDG 17, which is the collaboration for the goals? Absolutely. That's one one way and one framework to use. But I think you you highlighted beautifully. The need to be respecting diversity, traditions, languages, and understanding, and connecting all together. And, and I think this is a strong reputation for many to, to listen to this conclusion that he made, which I think is very powerful because it sums it up. And I think the future is about collective and collaborations and Coalition's and joining forces to make a better system of life. And I think we have been very inspired by you today. And thank you so much for your insight here. I will be delighted. Thank you so much. And I think everyone has been extremely happy to be attending this. And by the way, we have been on YouTube and Facebook Live as well. So many other people are watching this live right.
Now. Join me to thank you, Amelia, thank you so much.
Thank you much.
Thank you. Thanks, be well, everyone.
I think you will probably agree that this was a very insightful keynote and talk and conversation. And this is a good foundation for now to invite our dear friends, and one of the pillars of AI for Good and Francesca Rossi, that there has been with us since the beginning of AI for Good Summit since 2017. For those who don't know Francesca Let me introduce her briefly. She's an IBM fellow and the IBM Global Leader. And prior to joining IBM, she was a professor of computer science at University of padova Italy. As far as her research interest goes, Francesca is focusing on AI in general, but specifically on constraint reasoning preferences, multi agent systems, computational social choices and collective decision making. We just talked about corrective with the manual, I think collective decision making is a very important topic. She's also for setting ethical issues in the development and behavior of AI systems and in particular for decision support systems for group decision making. Again, we have this group concept that that it seems to be foundational. on these topics alone, she has published over 200 scientific articles in journals and conferences, proceedings, and book chapters. She's a fellow of many associations is too many to name. Of course, the worldwide Association for AI the European Association of AI, fellow of the future flyff, an advisor on Fisher five Institute, the deputy director of the liberal center, she's Executive Committee of the IEEE global initiative on ethical constitutional AI. And she's also the board member and Director and representing IBM on the protection of AI. And she's an expert for a European Commission, high level expert on AI and john Sharkey, Tripoli 2020 conference, and she has been nominated to be the triple AI or global association of AI president in 2000 22,024. As we talked about the evolution of AI for Good. I know very few people that can talk about the past few years, AI for Good has evolved. And give us a perspective of where we are today. And where we're going. So without further ado, I'd like to invite Francesca to give her perspective.
Thank you, Mister welcome. Thanks. Thank you, Amir, thank you for inviting me to give my perspective of how we got here and what we need to do. To move forward. Let me share my screen.
Can you see the screen in the presentation?
Yes, perfectly. Thank you.
So, when Amir asked me to talk about my perspective on the evolution of AI for Good. I thought that there is an interesting, you know, interaction between these areas, which are definitely overlapping a lot. But they're not really exactly the same. So AI ethics and AI for Good. So, let me start with AI ethics, which I think has a broader scope and includes, of course, the fact that you use the AI for Good social good applications. But then the AI for Good applications also has its own specific issues and charge edges. So let me start with AI ethics. And let's start defining in a very simple term of what AI ethics is about. So yeah, ethics is a very multidisciplinary field of study, not just experts of AI, but also experts in many other disciplines, especially in social sciences, like sociologists, psychologists, philosophers, economists, and even policymakers and so on. So very multidisciplinary. And these people together, try to understand how to design and build systems that in some sense are behaving in a way that we that are framed by our values, the values that we care about. And of course, the overall goal is that we want them to behave according to our values, because we want to optimize maximize the beneficial impact of AI, while addressing the possible negative outcomes. And that by doing that, this field of studies, since again, it is multidisciplinary, there will be some solutions that are technical, so possibly given by AI experts. But some other solutions that are non technical, that have to do with the best practices have to do with the
standards have to do with the
lessons learned, guidelines, and so on. So things that are beyond the technical solutions. And all now what what, how can this field of AI ethics came about? Why why we need AI ethics, and not just the ethics of technology in general. So let me refer this to what is specific to AI in this field of AI ethics. And let me start by defining what AI is right now, given where it started. So this is a very, very brief history of AI, of course, Arise has been around since the 1950s. But at the beginning, it was mostly about the fact building machines by defining the algorithm so that we're able to solve problems in an intelligent way, whatever that means is to find the best way to solve a problem. These algorithms were defined by human beings by the researcher, the developer, and then they were coded into the machines. So from then on, that machine was going to be able to, to to solve those problems, according to the recipes, these algorithms that the receptors are put together. Then in the 80s, another approach came about where a researcher said, okay, but there are some problems. So where maybe I can do better by not telling the machine exactly what to do the algorithm the steps to perform. But I can do better by just telling the machines examples of problems and their solutions. And and then I let the machine I just pulled into the machine and algorithm that allows the machine to learn from these examples, so that they can behave well, according to this example, can generalize to situations that they've never seen before. And recently, the deep learning approach, which is a specialization of machine learning, is that is a specialized way where it used, you know, neural nets with more layers, but the overall idea is similar, at least in the supervised case. So this is a fundamental difference. And of course, we know that this approach, this machine learning approach is very successful in some domains, for example, in problems that are not easily what is not easy to define the problem Exactly. And to define the steps to solve the problem, like in problems related to perception, interpreting an image, or interpreting what's written in a text, and so on. So these
approaches work very well now and are very pervasively used, although the other approaches still continue to be used in conjunction with these or by themselves. And, and the other characteristic of this machine learning approach is that they need Of course, examples, as we said, so they need a lot of data. And so since they need a lot of data, they also need a lot of computing power to handle the data. And so that's why they were defined in the 80s. But they are successfully used now only because in the 80s, we didn't have that amount, that amount of data and computing power. So now let's see whether these definition of AI brings us to AI ethics issues that that are specific to AI. And in some sense, I think they are. Because again, as we say, the AI needs a lot of data. So there are questions around data privacy and governance, because of that, also, sometimes these machine learning approaches is not easy to understand how the output comes from the input. So, we need is kind of a black box. So we need a lot of think about explainability and transparency also, now ai, ai techniques. Now, with this additional perception capabilities that can achieve by themselves can get to decision so out on the most or in form of recommendations to be given to a human being. So there are questions about the weather, by making this decision, the system actually follow our values in but for example, they are fair in the fairness says that according to some fairness definition that is appropriate for this scenario. Also, it is machine learning approaches are based on statistics. And so they always they cannot guarantee correctness 100%. So they always have even very small but a small percentage of errors. So the question is about accountability and responsibility when that error is made. And also AI can in fair buy from this data can infer our preferences and possibly use our preferences in some way. So there are issues about human and moral agency. And finally, ai compared to other technologies is very fast, and being, you know, widely spread and pervasive in our society. So there are issues about the impact of AI, on society and on jobs. So that's why more or less this designs, the whole space of the AI ethics issues, that are the main ones that people consider in this area of AI ethics, that are exactly the ones that I said in my previous slide. So now, how did we get here? How did we get to put together this framework for AI ethics with these multidisciplinary studies and so on. So let me start from where I think this overall and therefore started it from my personal point of view, at least. And this started in 20 2015. In 2016, there was a conference in Puerto Rico in January, organized by the future of Life Institute, and this conference was on AI safety. And the title of the concert was the future of AI opportunities or challenges. And it heard that it was not a big conference at about 80 people, you can see them all here. But it was the first time at least for me to get to an extremely multidisciplinary
convening of people. They were AI people, you know, from academia, from other research center, the word philosophers, the word psychologists with the word lawyers were enterpreneurs. So they were science fiction writers, so extremely interdisciplinary to understand what the future of AI should be in a way that is as beneficial as possible. And that I passed to this conference that was also in a publication of an open letter on researcher issues about the robust and beneficial AI, an open letter that is still on the web, on the website of the future. Luffy. So that be signed by many, many thousands of people, ai experts have none. And also attached this conference was a 10 million grant given by Elon Musk at a time for, for putting together a research program handled by the future of Life Institute to support projects around AI safety and beneficially. Yeah, so a solid, you know, research projects around this topic. So to me, that was the start of this disciplinary frame of mind around AI and AI ethics. And then the next step was also organized by the same entity. This was two years later. But I don't want to say that nothing happens in those two, nothing happened in those two years, many other events. But this was also another milestone in my view. Two years later, again, in January, the future of Life Institute organized another conference, this time in California in a filmer on beneficial AI. This time, there were about 250 people and again, you see them all here, again, very, very multidisciplinary, and they've got and the goal of the conference, of course, the event was to discuss the principles that would be kind of overall and globally accepted as principles to guide the development of the future of AI And of course, we started from many, many principles, more than 100. And then we ended up with 1023 principles that are the ones whose topic are written here, or and all the people at the conference basically could vote which principles they like the most. And these 23 are the ones that achieved almost unanimity for all of the 23. And so that's why what they were selected out of the 100 that we started from. So the principles are how, yes, had ethics and values out of the center. So what are the properties related to, you know, the technologies so that we want to make sure that in developing the technologies and their behavior, the technology and the use of the technology, we are following our values and ethical consideration, but also also principles about how research to be conducted, so collaborative, open, and so on, and also issues about long term, longer term issues about control, artificial general intelligence, human level intelligence, and so on. And these principles are even now you know, cited as one of the main, influential and also one of the first pioneering sets of principles about AI.
Then, in the same year,
my company, IBM, together with other five companies that you see here, Amazon, Microsoft, DeepMind, Facebook, and Apple, we were meeting, as researchers, we were meeting in various venues about AI ethics. At some point, we said, well, you know, we think that not one single company can actually understand what to what how to address this AI, etc, or even how to define this idea, this or by itself. So we should get together and think about it together, but not just together as companies, but together with all the other voices, all the other stakeholders. So that's where the partnership of AI idea came about. And, and right now, you know, it was started by the sixth company right now, where there's about 100 partners. Each partner is either a company or a non for profit organization, a University Research Center, a UN agency, a civil society organization, like ACLU and many other And right now, out of 100 partners, only 20 odd companies and everybody else is all these other stakeholders. So and what what the partnership on AI does to build projects that work collaboratively, various stakeholders get together, not necessarily all 100 of them, but in a balanced way between profit and non profit to address one of the issues around the AI ethics, whether it's about fairness or transparency or its vulnerability, or impact on jobs, and whatever. So here are the eight sentences that I wrote here are the eight tenets of the partnership on AI. So again, some sort of principles that the partnership on AI wants to be based on, where every partner that joins the partnership in AI has to sign so has to sign that they will adhere it to these tenets. So there are tenancy as you say, as you see about the more or less, more focused about the how we want the partnership to function, you know, so be fun a feature to make sure that AI is as beneficial as possible for as many people as possible to educate and listen to all the stakeholders to the author, researcher, again, multi stakeholders in the business community, but also in the in the other areas of our life, of course, maximize the beneficial effect and minimize the challenges. And this number six actually is a much longer tenet that I didn't I couldn't with many sub bullets that I could not include here. A lot of emphasis on explainability about the technology and the cooperation and openness.
Okay, so in the same year, it also company started to individually also to come up with grapple with what it meant for them individually as a company to define these principles within their own company because of course, every company is different as his own business model and his own way of operating some companies of low cost Avaya globa, someone's more somerby. So each company started to think about what it means for that specific company and of course, IBM was one of the first one to do that by publishing these principles of trust and prosperity. So the idea is here that, you know, IBM said, For us, the purpose of AI is not to replace human intelligence to augment human intelligence. And this is also because we deliver a UI to other companies. So we are going to deliver a UI to support the decision making of professionals in this other company, then the principle is about data and inside that belong to their creators, and we are not reusing them for other projects or other players. And then about transparency and explainability, which is very important that even if you are supporting human decision, so because of course, to build an effective human machine team, you need transparency and explainability. But this principle also said that the trust or the notion of trust and was introduced, so we said, okay, but what does it mean to trust an AI system? Well, it means we spelled it out in those four pillars, fairness, robustness, sustainability, and transparency. So here, in 2017, we laid out the initial framework or initial frame of mind around the our, you know, approach to AI, it's in the sort of my area, and many other companies also try to understand individually what it meant for them. Another initiative that I was at that I think it was very influential around AI ethics, and also AI for Good. He is the IEEE initiative on the ethics of autonomous and intelligent system. This is still ongoing, it started in 2016. And basically, the believer, this book that you see, the cover here is called ethically aligned design. It involves more than 700 experts globally, it has almost 300 pages. And this principle has many chapters, each one for a specific topic of AI ethics. And he listed a lot of issues in the topic and possible recommendations for engineers, because I Tripoli is the worldwide association of all the electrical engineer, so the developers of these AI systems, so guidance and education and recommendations for these developers on how they should develop AI in a way that have desirable properties that end the whole book, and the whole initiatives with many other initiative, sub parts of the initiative, not just a book, for example, also several standards on AI ethics. And the whole initiatives is based on these general principles that have to do with the human rights, data agency effectiveness of policy, accountability, awareness of issues of competence. So you see that it's a slightly different way to say state the principles, but related to what I mentioned earlier as a set of principles. Now, so here, you know, you you saw, you know, multi stakeholder initiatives, you saw a multidisciplinary convening of various stakeholders, individual companies, the developers community, and then also governments came about and they say, yo, you know, we also want to understand what the yacht experience in our regional in, in in our own region of the world. So here, I give you the example of the high level expert group on AI that was put together by the European Commission, to understand what these AI ethics was going to mean, for Europe, for example, this is the group of 52 people that worked for two years, from 2018 to from June 2018, to June 2020, and the laborer, the these four main documents that you see here, to which the most influential one was about the spelling out the requirements and
requirements for what we call the trustworthy. Yeah. So again, the notion of trust enters here, as you know, as a central topic, so what are these? You know, how do we get how did we get to these seven key requirements, we started with the human rights, then we get to principles, and then we got to these requirements that have to do again, with more or less the same topics that I mentioned before, around the what is the prop? What are the properties of the technology that we want once deployed, and what's the ecosystem and framework that we want to put in place around the technology to make sure that is as beneficial as possible. And then the other documents are also about recommendations, of course, for the policymakers for the European Commission, around the policies and investments of for trustworthy AI in Europe. Okay, so the latest The latest initiative that I want to mention now that started very recently in 2020 20, is an initiative where not companies, not one single government, but now several governments decided to get together in a global collaborative way to understand that with the very big community of experts in AI and and others, and other disciplines, to understand what's the best way to collaborate globally, as governments, based on the expert knowledge, to make sure that AI is beneficial, is as beneficial as possible. So there are in this global partnership on AI, there are already 15 countries, it was started as a bilateral agreement between France and Canada. But now it has 15 countries all over the world. And it has four working rules, Responsible AI, data governance, future work, innovation, and commercialization. And again, it is based on some shared principles that are given by the OECD in this case, that have to do with good national and international cooperation. And then also, again, the responsible stewardship, again, of trustworthy AI. So again, trust is part of the picture. So in this is more or less the evolution of the AI ethics field. But and now, I want to show you this, because as you can, as you could see, from what I told you, every initiative, tried to come up with its own view, from its own point of view, you know, of the principles that are needed around the UI. So and there are so many there were so many initiatives that were put together in the last four or four years or five years, that somebody counted them and put them together. So and there are more than 84, I think sets of principles. Here is a visual representation, a very nice visual representation put together by the principles AI project of the bell curve clients cyber Law Clinic at Harvard University, this was done in 2019. And here, it's even impossible to read the first. And if you go to their website, there is a way to zoom but this visual because so that you can see, clearly, you know, who is who who are the stakeholders for publishing those, those principles, and so on. And here are the main is act of siara, private sectors, intergovernmental, multi stakeholder governance, civil societies, all these that I mentioned before, just giving you a few examples, and the end, but they all publish the different sets of principles. But they all have this overlap of chords into these main themes that I met that I wrote here. And as you can see, they're very, very similar to the ones that we started from. So now, of course, why did I want to show you this list of principles? So visualization of the list of richer because, of course, yes, now governments have this principle companies of the principles, multi stakeholder organizations, the principle and so on. Now, it's time to go from the principles to the practice. And in fact, there are a lot a lot that has to be done within each ecosystem. To actually operationalize these principles. You need to put together guidelines, best practice standards, toolkits, educational activities, revision of the developers pipeline for developing AI system, define methodologies for them, make incentivize adoption of that bit large, put together the governance inside the company, that the cabinet oversees all activity. So the this is all, you know, needed within each ecosystem, to make sure that these principles do not remain principles, but are operationalized. And of course, it has to be done again, in a multi stakeholder multidisciplinary approach. Now, let's go to this AI for Good of course, AI for Good is mentioned in every one of these sets of principles. But in some sense, it had the parallel story that was started even much earlier than the AI ethics. So the first time I think the first initiative I would like to mention here is in 2008, that a colleague, a friend of mine at Cornell University, received this NSF big grant a 10 million grant to define these computational sustainability.
New area, and again, this was the start of thinking how can we can use current existing AI capabilities to solve the problems that are related to social good environmental, economic, or any kind of problem that have to do with sustainable future. So some projects, for example, have to do with, with planning, optimization for wildlife preservation for biodiversity, mapping of poverty, combining learning with inference to accelerate the discovery of new material, for example, and so on. So that is was the start of a whole field, which is now called the computational sustainability that has his own competencies or venues and so on, that is mostly is less in multistakeholder than AI ethics, but is really very practical in trying to understand the how to solve these important problems related to the societal, societal good. So the next initiative that I want to mention that again, started a long time ago around 2010, is an initiative by another colleague of mine mill in Tampa, that was the University of Southern California now is at Harvard, that is really around AI for social good. So his theme is called theme core, and has to do with, again, social work or public health for decision making, or conservation, safety and security. And he uses techniques and now also the COVID-19 pandemic. And he uses techniques that have to do with the core game theory, machine learning, planning the multi agent system, so to address these very important issues around social good. The next one is the IBM science for social goals. Throughout this IBM science for social good, the Avaya within IBM started in 2016. And it's a very good example of something that was done at the beginning to really connect the talent within IBM Research with the new generation of students, PhD students that are interested in AI, to educate them in understanding how AI can be developed in it in a multi stakeholder environment, and for social good purposes. So this is a program that every year at IBM Research, we host, a number of PhD and postdoc students that are mentored by IBM MERS on projects that have to do with social goods, here you see some of them aware of the data. And the definition of the project is not given by IBM is not given by the student, but it's given by a third party. So in a foundation, or solid UN agencies, or some third party that has to do with, with having the data and understanding the problem in depth. So it's really, to me, it's a really learning experience for IBM MERS, of course, but I think it's a very important learning experience for the students that not only learn how to apply some AI techniques to some domain, but also learn to work with the in this multi stakeholder environment. And for for social good purposes.
Of course, all these AI for Good initiatives, compared to the AI ethics initiatives are more focused on the vision of the future. So that's the 17 Sustainable Development Goals are usually taken as the vision of the future where we want to get, and then they try to understand that this initiative, how to build the a trajectory for AI from where we are, so that we get closer to that, to that vision of the future given by the Sustainable Development Goals. The AI at its initiatives are more general, they are not focused on specific applications, although they have these AI for social good, you know, initiative within them. But since they're more general, they they they don't focus on a specific region of the future, although of course, they want a beneficial future in the impact of AI. But they focus more on an incremental view, which is also needed by saying this is the state of AI, this is the concerns that we have now explainability, robustness, fairness and so on. So how do we tackle them? How do we address them so well, in a way that how, as we advanced AI capabilities, we also advance our understanding of how to address those issues. So here I just put the something that yes, that's the vision of the future. But I mean, we have to be conscious that we are not really getting very, very, you know, in a very good trajectory to get closer to that vision. So here there are some examples that you know, we I'm not really in a good track in terms of the poverty, poverty SDG, or the climate crisis, or the sustainability, or food, or violence, or, and so on. So we really is there is a lot of work to be done, even with that vision of the future. And so that's why we need to leverage the best voices and the best talents in all the disciplines. Together, of course, we do AI experts. One thing that we did within the IBM a science for social good product, but I think we can be relevant more generally, is it said, okay, we do all these projects, and they're very happy with all these projects for what we learn what the students learn. But these are all a Docker every time we start from scratch in some sense. And so instead, this is not going to scale. And we need to understand how to scale with it. And we we we try to brainstorm. And we said that yes, to scale, we need the many ingredients, we need the you know, of course, the a platform for collaborating, and infrastructure for the platform where people can actually collaborate, we need to understand whether there is some pattern that gets repeated so that we can leverage a pattern and make it available for the whole community. And then as Amir said, also in the previous question, answer, you know, we need the right business model to make it sustainable, also, economically to scale this social booter programs. So here, for example, is a way to see you know, how we, you know, we looked at several problems, that were all related to some health care issue, you know, like a die car virus, the cognitive diseases, cancer treatment, antibiotics, and recently, of course, the COVID-19. And we say, what is common here? What is the pattern that we can identify, and then once we identify this pattern, how can we use it to make it available to the whole community. So that's why we built the platform infrastructure, for example, that here is related to the generation of AI for Molecular discovery, and that we use the in for making that made it available for the entire community, and to actually in specifically in the last six months for exploring and identifying the promising molecules for the COVID-19 prizes. So of course, we didn't write, we didn't just put in place, the necessary tools, we understood that we also needed to, we needed to support these tools, with a lot of computing power. That's why we injected them into this COVID-19, high performance computing consortium that leverage a lot of computing power for many companies.
And many hardware providers to really make it so that is an example where we understood what the pattern that repeating pattern was. And we leverage that to to build them a tool, and in a new in a platform that then can be used by the whole community into in a framework such as in this case, the high performance computing Consortium for COVID-19. And the business model, as I said, is also very important, we need to understand what is needed for making the AI for Good business model being sustainable. Of course, we need the the funder some of the stakeholders, voices, we need, of course the problem solvers, and the technology experts, but also we need the domain experts. And this is I think it's very much in line with what these these forum, the summit as tried to do over the years, starting in 2017. So the overall goal, as you know, was to come was to connect the as Amir says many times the problem solvers, so the AI experts and the problem owners, so the UN agencies, or those that are on the ground, and understand that really the problems and the features of the problems will be solved in that the data that can help specify the problem. And as you can see here, you know, that the the fact that there has been these incredible growth in attention and in interest for the for the AI for Good Global Summit shows that there was really the need that there was something missing in this connecting these two. And I will say even more than two communities, because again, we want the multi stakeholder approach. Another related thing is the fact that we need also, to have a safe and a safe experimental environment where AI and the other AI experts and the other stakeholders can connect and can try out solutions or possible solutions for certain problems. So how do we share resources and when we share routes, have access to these resources, as well as sandboxes places that we can experiment without really impacting directly? explicitly on real life? And how do we make up the governance of this, and this is the AI commerce idea that Amir again, pushed so much for starting in 2016. And I think he's having a lot of a lot of success, a lot of interest, because again, that was something that was really, really missing. So we are really in a very in a good trap here it with this initiative.
The last thing, which in my view is not the last because I am an AI expert, I am an AI researcher is is this question. I mean, is current AI technology ready to provide us everything we need to address AI for us to address social good problems? Well, I think it's not ready. So of course, it can be successfully applied in many different ways. But it would be even more even more useful if it will have this edition of it. So for example, current AI is very, very good at finding correlations. Hidden correlation in the sense that we you must can't even see in huge amounts of data. But correlations are not everything we need, we need also to understand the causality because correlations allow for predicting, or things that we happen, but it will not allow for intervening, because it does not allow for counterfactual reasoning for what if reasoning. So causality understanding and reasoning is really very important that currently Yeah, is still lacking a lot in that respect. Also, ai needs a huge amounts of data. And sometimes we don't have that huge amounts of data. So we need to understand the how to build a AI system that can learn to solve the problem, but with little data as well. Also, AI is not really able to adapt easily to new environments. When there are some events that disrupt the current trend, AI is kind of lost. And we've seen that also with the COVID-19. Many companies that were relying on AI for the prediction of the demands of the sale, so on, they were in trouble because the old our patterns of behavior of buy things, and so on, were all disrupted by the the current emergency and the crisis are related to COVID-19. So AI does not know how to well adapt to new environments. These are all things that you must know how to do, we can learn really, from data, we can adapt to new environment, we can do counterfactual reasoning, and we can infer causality. So that's why in my latest project, and but it's not, I'm not the only one. But many, many people in the AI community are trying to exploit what we know of you of the human mind, and how what causes our desired capabilities like adaptive adaptability, and so on? What did what are the causes of our capability? And, for example, two theories of human minds that, that we are currently looking at is the Thinking Fast and Slow by Daniel cranham. And also the T the theory of the philosopher, Yuval, Noah Harare. And so by looking at these theories, we can possibly better understand, although they're just theories that we know, we know that we know very little about our mind as well, we can understand how to inject these causes of desire the human capabilities into machines, so that maybe, then we can have machines that have those capabilities as well.
As a summary, I would say that, yes, we have gone a long way in very few years. So so I'm really amazed about all the things that has been done. And we should be proud of everything that we did all the convening and all the discussions, all the principles, all the everything, but now we need a lot of more work to do. First of all, we all need to work in going from principles to practice operationalizing AI ethics principles. So I am personally spending much of my time and many others within IBM in doing that within the company with our internal AI systems. That the governance or the and all the other work streams that we have around that. But I think that this should be done more globally and not within each company but also more globally. Also, we won't we have to work in connect these two approaches, the approach that says, Let's advance AI, and addressed to the challenges that we see on the way. And the other approach that says, let's look at the vision of the future. And let's try to kind of re reverse engineer what how we want the AI to get to that future. The first one, again, understand much better how to make these AI for Good programs to scale. So how to define platforms, partners, platform, business models, and so on to scale, how to share data models and solutions. So the AI Commons is one initial approach, but it we need a lot more work in that space, again, how to make this business model sustainable. And then again, a way is to the last thing that I mentioned how to advance AI to support those capabilities that are really needed in addressing in the best way, the social good the problem, so causality, adaptability, data challenges, and so on. So how do we stop here and welcome any questions?
Thank you so much. Francesca, I think this is a wonderful overview of just five years. Yeah. The whole industry and ecosystem and, and probably there is, is like the iceberg, probably there is a lot more than have happened globally. In many countries that we have not seen. We have a little bit a few more minutes to to basically go to some questions. And I would like to, if you if you agree with that, I have more questions myself, but I want to just physically pay tribute to the audience. And if we can good, very quick answer, so we can get more of them answered as well. So let me read some of them quickly. One question is that could How could trans calculate could you translate the ethics framework? to a specific context? For instance, climate?
Yeah, so uh, so that that is something that, um, I mean, there are some initiatives related to specific social good issues like climate poverty or medical emergencies, health care, and so on. But, but I think that we need to do more about this sectorial. You know, and and problem specific frameworks. Yes, there are initiatives that think specifically about one problem, but I think I agree that we need to do a lot more in defining what are the challenges in that particular issue around social good. So I think that that's another dimension that I should add, no more problem specific framework.
So the same similar questions that came up? And the answer, I think, part of that, which is how do you consider the ethics of AI potentially impact on brother planetary health? But I would like to add, specifically on gender as well, because gender is one of the topics of this year, saw this, these principles, and these AI professional impact on health and gender could be cool. Yeah.
So in fact, I mean, as you know, as part of the, the putting together these gender equity, in tracker at the south at the summit, and I was amazed to see about all the project proposals that we received about all the possible dimensions and kind of solutions around the gender equity, in, in AI and in general in society through AI. And, and I think that, you know, there are so many dimensions, and I saw so many technical solutions, but also non technical solutions, so guidelines, best practices, education. For example, I think there is one project about recommendations and implications for judges in the judicial system around gender equity. Then there are other ones about data sharing about the gender, gender issues. So there are so many dimensions. And I hope that even the project that we did not select because we had to like three of them, that I hope that all of them can be made visible and exposed to the whole community so that they can be picked up or possibly, and also scale that because again, one other issue that we said about those projects that we hope that by exposing them to this AI for Good community they can actually be scaled Much, much more.
I would like to ask two more questions very briefly, because they seem important. One of them comes and the formulation is the following. It looks like there is a quite a bit of room for interpretation, and scope to selectively apply those principles in on the partnership on AI, or the general ethical principles and the tenets, or the principles from the future of life. How do you make sure that they are fit for purpose and drive desired behaviors?
Yeah, so as I said, the principles are usually always these principles are very high level. Okay. And, and that's good, I think, because they get general values, ideas, that and point of view around at very high level, but then we need to operationalize. So for example, at IBM, we said, okay, we have these three very nice principles. But then, and then we couldn't tell the developers now follow these principles and developer Yeah, following the principle, the gap is too big, you need to go much, much with many, many different initiatives, and guidelines and dedication and pipeline revision and adoption and discussion it's on because to go from the principles to actual concrete things that the developers or anybody can understand how to act. So I agree, there is a lot of left to interpretation, but in each context, and then the one interpretation has to be chosen, and then implemented.
One last question,
which is, where do or will the developing nations, specifically those that are almost invisible in the roadmap stand in this AI race? And after the Industrial Revolution, this is supposed to be a chance, it could be a game changer for small developing nations. How did you see that?
Yeah, so I think that
some of the initiative that I mentioned, also include the developing nations exactly because they need to be present is not just multi stakeholder voices in one country, but also multi stakeholder, because we want to hear the voice of and the challenges in every kind of, you know, nation, and especially the developing ones that are behind the others in terms of reaching some goal, some of the goals that we mentioned. So that really, those are important challenges that everybody, even even nations that are not in that at that stage, they need to listen and they need to help solve. So that's why also in the global factory, God Global Partnership on AI, the most recent initiative, there are also states that are that are represented, you know, developing nations. But I think that more and more we need to be global global is very important, different cultures, different stages of the, of the development of a nation, so that we hear all these voices, we hear the voices of women, as we said, for the gender equity, we near the voices in terms of all these other components, and features that you know that I really think that global, and, of course, nations and more divided the goals of which disagree, what we what we always say inside the partnership on AI is that we practice these productive disagreement when we are in a project that we have in the framework of productive disagreement mean that it's okay to disagree. But let's be productive about this bigger agreement and move forward to solve the real problems.
Thank you very much, Francesca. We have way more questions that we don't have time for because we're past our time, I invite all the attendees to join the Slack channel, I think is in the chat. The link is in the chat. And you can continue the conversation asking more questions and discussing on slack. Because we have limited time. Again, I like everyone to join me and thank you, Francesca for this wonderful, bold, inspiring and important blueprints. As our friend Katelyn Craftsman said, the real good blueprints for us to understand where we are today. And we chart a future of fear for good together to all see a precious day again, in the next few days. We'll have the full track with some some results that are coming out of that but also on the 30th we have a few announcement and result of this few days that Francesca will be with us as well to
me, thank you all for giving me the opportunity to address this audience. And I mean, we should Best of luck for the current edition but also for the future of the AI for Good Summit.
Thank you very much and Right. Now I would like to invite Zinnia. Again, churches to help us close the session, but also give us some information about what's coming up next.
Thank you very much Sameer and thank you for all your great questions. And of course our speakers for all your insights. I would like to invite our audience to join us to the workshops that will happen right after the session. And right now on your screens, you can see the schedule so we have three workshops in parallel gender equity workshop, that will take place in five minutes, then we have future food starting in around 20 minutes and we have the pandemic intelligence workshop starting in around 35 minutes. If you would like to join us for this workshops, please register on our website or if you already registered and participated yesterday, your connection link is still available and will be available tomorrow as well. So I would like to thank once again, all our sponsors, partners and Switzerland for your continued support. Thank you everyone, and I hope to see you after this break.