Latest / CXOTalk / Global AI Governance and Digital Equity: A UN Perspective | CXOTalk #862
Transcript
- 0:00Welcome to CXO Talk Episode 862. I'm Michael Krigsman, and we are
- 0:07exploring perspectives on AI from the United Nations with
- 0:12Bernardo Mariano Junior. He's Assistant Secretary General
- 0:17and Chief Information Technology Officer at the UN.
- 0:20What is core to the United Nations?
- 0:22Right from the United Nations Charter from from 1945, which is
- 0:28really safe succeeding generations look at human rights
- 0:34and social progress as well as better standards of life.
- 0:39So when you take that the core United Nations Charter and then
- 0:47you overlay that with the technology and the the digital
- 0:51transformation, including AI. So you want these
- 0:56transformation, this technology to really accelerate those core
- 1:00values. I mean, we have the sustainable
- 1:02development goals, the 17 goals, we want AI to accelerate those,
- 1:08the achievement of those goals. You we have of course then the
- 1:13the, the social progress, we want to make sure that it
- 1:16accelerates that we want to make sure that AI is used for good.
- 1:21We want to make sure that, of course, all associated risks
- 1:25that it brings are mitigated, to make sure that humanity and the
- 1:30people of the world fully benefits from the artificial
- 1:35intelligence and the whole transformation it's bringing in
- 1:38our today's world. The goals that you're describing
- 1:41are very noble goals, but the question is how does the UN turn
- 1:49those goals into reality on the ground?
- 1:53Difficult conversation. It's our everyday life in United
- 1:57Nations. And if you want a job that
- 2:03everybody agrees that's not United Nations.
- 2:07Our, our, our role is to really bring together the different
- 2:10interests, different conflicting interests to make sure that we
- 2:14charted the way to extract the maximum benefit of that.
- 2:18So we have the, the, the, the pack of the future that was
- 2:22recently adopted in the, in the, in the, in the General Assembly
- 2:28of, of, of, of this year in October.
- 2:31And, and within the pack of the future, we have the Global
- 2:34Digital Compact. We do have initiatives such as
- 2:37artificial intelligence for good led by the International
- 2:41Communication Union. We do have, for instance, the
- 2:45ethical use of AI that was published by by UNESCO.
- 2:49So we do have an instrument that are we put together to make sure
- 2:55that as we go through this transformation, those who have
- 3:00and those who have not are not, they don't go far apart.
- 3:07And if you think of, if you look at, for instance, the global
- 3:09digital compact, the, the five commitment actions, including
- 3:15exactly that. So to, to make sure that the
- 3:18digital divide is not exacerbated, to make sure that
- 3:21we, we use the, the AI and the frontier technology in a secure
- 3:27way to make sure that the governance of data is for the
- 3:31betterment of human humanity, but also to ensure that we have
- 3:35some sort of international governance on AI.
- 3:38So those are hinged in the, in those, in those principles.
- 3:41And, but also, let me say for the very first time in the
- 3:45United Nations General Assembly approved document such as the
- 3:49pack of the future, there is 229 instances of the word digital.
- 3:55There's 770 instances of the word technology and there's
- 4:00almost 40 instance of the of the word artificial intelligence and
- 4:05over 50 heads of states in the last General Assembly, they
- 4:09mention the word artificial intelligence as one of the key
- 4:13areas of concern that they want to address.
- 4:16So all to say that we have a role to play.
- 4:19We have all to say that we need to ensure this the drive, the
- 4:26transformation that is very, you're moving at the very fast
- 4:30pace. Pace benefits humanity in a way
- 4:33that we can all extract those benefits.
- 4:35So those instruments are there for that.
- 4:38Can you describe the role that the General Assembly and that
- 4:44you believe that technology, digital and artificial
- 4:49artificial intelligence can play?
- 4:52Where is the role of these technologies?
- 4:55In the General Assembly and the role of AI in that, in within
- 4:59that context is to make sure that that we as humanity, we
- 5:05ripe all the benefits, but we as humanity, we mitigate all the
- 5:08risks and we as humanity do not create this huge divide where
- 5:14those who have and those who have not, they go far apart.
- 5:19And and let me just dive in a little bit on this.
- 5:21Those who have, if you go back 20 years, the technologies 20
- 5:32years back was, was driving basically was a physical
- 5:38element, either because you have a a device or you have ACD ROM
- 5:45that contains a system or you have basically something that
- 5:51the national legislation of a country could hang on and say,
- 5:55look, I'm going because you don't comply with these these
- 6:00national regulations, I'm not going to allow you to bring
- 6:03these item into my country. Today that is gone because the
- 6:12borders, these physical borders that would allow countries to
- 6:17somehow protect themselves against some of the negative
- 6:21impact of technology. They're no longer there.
- 6:24Which means that the only instrument that we have to
- 6:28address those risks is really the international instruments
- 6:32where the United Nation plays a role to make sure that we have
- 6:36some sort of an international instrument.
- 6:38Of course, we have now the European Union AI Act, we have
- 6:44them EUUKUS the AI by a treaty. We we have some instruments at
- 6:53the regional level, but there's not address the fact that the
- 6:59marketplace for those who have those that control the
- 7:04technology that are creators of the technology is no longer, I
- 7:10mean the marketplace is no longer the national border of
- 7:13the country where they operate is the whole world.
- 7:16So therefore these equitable access, these digital divide has
- 7:23to be done, has to be framed within the context of
- 7:28international framework. Tell us about the global digital
- 7:32compact, Bernardo. Let me talk about these five
- 7:36commitment actions of the deep global digital Compact, which is
- 7:39very important. And those five commitment
- 7:42actions are, are commitment actions that we want countries,
- 7:46the 193 member states of United Nations, all, all countries in
- 7:51the world to, to, to adopt. The first commitment action is
- 7:55close the digital divide, meaning that within the national
- 8:00borders, you have people that have, have the access and people
- 8:03that don't have. But also within the
- 8:06international context, you have the owners of those algorithms
- 8:11of artificial intelligence that have the computing power for
- 8:13the, that, that, that, that, that virtual intelligence and,
- 8:16and, and, and, and those who just consume it, let alone the
- 8:19impact on environment and other sustainable development goals
- 8:23such as carbon emissions and, and all the rest.
- 8:26So that's the first commitment, close the digital divide.
- 8:30The second one is to expand the inclusion, which is really is by
- 8:33closing the the digital divide you want to also expand that
- 8:38particular inclusion. The third is to foster that
- 8:42secure environment. We, we know that today
- 8:47technology including social media is impacting lives of
- 8:52people. I mean we saw in Australia a
- 8:56young girl committed suicide because of bullying in the
- 9:00social media. We see elections being changing
- 9:05directions because of misinformation.
- 9:08So we have to be aware of these impacts and then create a secure
- 9:15environment and and then good interpretability and also human
- 9:19rights protection within the text of the whole technology.
- 9:23I mean the Fort is really this interoperability.
- 9:27We are in a world where impact actions in one national board or
- 9:34national jurisdiction does not mean no impact on a different
- 9:38jurisdiction today. That's that's not applicable.
- 9:41So whatever happens in whatever place, be it in Timbuktu, by the
- 9:45way I used, I mean I did work in Timbuktu or in New York.
- 9:50There is an impact on, on things that happen in different places.
- 9:54And the the 4th and the 5th sorry, is really the
- 9:58establishment of that international governance or AI.
- 10:01So those are some of the elements the five commitment
- 10:04action on the Global Digital Compact, which we we want
- 10:10countries, companies and the whole world to to to embrace and
- 10:17adopt within the national context, within the regional
- 10:20context and as well as the global context.
- 10:25Subscribe to our newsletter, Check out cxotalk.com, Subscribe
- 10:30to our YouTube channel. Well, we have some interesting
- 10:34questions that are coming in. And this is from Arsalan Khan on
- 10:39Twitter. And he says that we know the UN
- 10:42has great global goals, but what about nations that don't want to
- 10:48agree to those goals and use and therefore use AI for nefarious
- 10:54purposes? How does the UN address that?
- 10:58Let's create a scenario that there's no international
- 11:02framework. So imagine the impact of the
- 11:05same scenario. Just explain of a country that
- 11:09did not adopt or or did not ratify that a global framework.
- 11:17I mean, if there is no global framework, the impact will be
- 11:20345 times worse because there's totally no, no, no, no global
- 11:25framework to, to operate. And United Nations is composed
- 11:30by 193 member states. So we, we've, we, I, I, I, I, I
- 11:35believe and I'm sure it is the majority of the world.
- 11:39So it's true that there are few countries that are not, they may
- 11:43not adopt, they may not ratify, but that that doesn't prevent us
- 11:48from having that framework, doesn't prevent us to continue
- 11:52to work with those countries to adopt, doesn't prevent us to do
- 11:56that. Focusing necessary to make sure
- 11:58that those countries they adopt or at least acknowledge or
- 12:02increase or domesticate in their national laws.
- 12:05Even if they don't adopt, they don't ratify International
- 12:08Convention, but in their national legislations they adopt
- 12:14some of the principles. Of course, the owner is not only
- 12:18on the government, the owner is also on the people of the
- 12:22country, the people of the world.
- 12:24To make sure that they also, especially the civil society,
- 12:29put uses its own power to drive that change within the national
- 12:36context. Michelle Clark on LinkedIn says
- 12:41Are you aware of any of the top AI companies having embraced the
- 12:49Global Digital Compact? The Global Digital Compact was
- 12:53adopted by General Assembly in October.
- 12:55So we just, I mean, a month and 1/2 into it, but they, they,
- 13:02they, the process to which the Global Digital Compact took
- 13:06place, it included a number of private sectors, basically
- 13:11started with a call from the secretary general to have an
- 13:14Advisory Board on AI and they created a produced a report
- 13:20which is titled AI for humanity. I think you can, you can Google
- 13:25and download that. And within the AI for humanity,
- 13:28they were, they had like 77 recommendations.
- 13:31And some of those recommendations are the ones
- 13:36that actually in the being built in in those five commitment
- 13:40actions in the global digital compact.
- 13:44All to say that this was not the global digital compact was not
- 13:49just the creation of United Nations is a Co creation between
- 13:56basically government, private sector civil society members
- 14:01that the secretary general called them to create to
- 14:04establish this Advisory Board and based on their report this
- 14:08this was was great. So with the way there were tough
- 14:11discussions in the process and there were companies involved
- 14:15in, in, in, in the discussion to, to, to establish that the
- 14:18global digital compact. But let me give you one example
- 14:22that really I mean that I, I, I enjoyed very much by that
- 14:28interaction. I met young professionals, young
- 14:33innovators that attended. I think they met in the AI for
- 14:38good in Geneva, hosted by ITU and they create basically they
- 14:43are promoting this adoption of AI principles by private sector.
- 14:49So, so we see this energy of the, the global digital compact
- 14:54in the private sector with innovators to say, let's make
- 14:57sure we push that the responsible use of AI.
- 15:01So, so, so, so kudos to these young folks that I, I had the
- 15:05privilege to meet them here in New York.
- 15:07But I, but we want more and we want these adoptions to, to
- 15:11actually not just be something that is driven by United Nation,
- 15:15but it's something that we all embrace.
- 15:17And we all in both those who who have big profits with the, with
- 15:24the, with the, with the technology, but those who are
- 15:27also startups to really embrace these principles that AI has to
- 15:31be used for good. Would it be fair to say that
- 15:35what you're describing falls under the category of
- 15:40governance? Is that the right term, or would
- 15:42you describe it in some other way?
- 15:44It falls in two areas. One is governance, but also the
- 15:47other one is advocacy, right? So, so, so the, the governance
- 15:52is, is what we want to establish to make sure that that there are
- 15:55some, I would say, principles that we, we embrace, adopt and,
- 16:03and, and, and, and we use as we developed AI.
- 16:06So that's that's the open use. So that's one area, but the,
- 16:12that advocacy requires all of us require all of us that have
- 16:19access, that have an audience, that have the ability to
- 16:25influence that audience to, to make sure that we push towards
- 16:29that direction. Because the negative impact is
- 16:34it's, it's of great concern. It's not a small impact that
- 16:41will just impact my neighbor or myself or my little community.
- 16:46It's a global impact. So we really need the power of
- 16:51advocacy of the whole world to make sure that we come out of
- 16:56these and we leave to the next generation something that we can
- 17:00be proud of. That's decided to say that in
- 17:02the transformation of artificial intelligence, we the human
- 17:06beings that we are in in in the, in the thick of it be from the
- 17:11those that we created, those that innovative, those that are
- 17:14using, those that are are making profit out of it, those that
- 17:17are, are using it to improve productivity and create new
- 17:23business streams. We all come out of it proud in a
- 17:27way that the next generations will look back and say kudos to
- 17:31this generation for having done a great job.
- 17:34One could say that the advocacy is a set of goals and
- 17:41objectives, aspirations and the governance is the set of rules
- 17:47that we agree to adhere to in order to implement or execute
- 17:52against those objectives. Is that is that a correct way of
- 17:55looking at it? Perfect.
- 17:57Yes, you, you said it better than me.
- 17:59Thank you, Michael. And on the topic of governance,
- 18:06we have a question from LinkedIn, again from Michelle
- 18:09Clark, who says how do you envision governance working?
- 18:15Let's take telecommunication right back in, in the 60s,
- 18:20seventies, or even earlier, I mean, when the telex was the,
- 18:24the mean for, of the world to communicate, the need to, to
- 18:28have standards and to have some principles around it emerged
- 18:32because otherwise, I mean, it would be, wouldn't be possible
- 18:35to communicate. And the International
- 18:37Telecommunication Union is, there is one of the United
- 18:40Nation agencies that really brings the private sector, the
- 18:43public sector and, and, and the governments to, to discuss the
- 18:48regulations around telecommunication, right.
- 18:52So, so we have a model, we have a model that works today I'm
- 18:57able to call my mother that is sitting in a little village in
- 18:59Mozambique. Thanks to that O, we want these
- 19:03to evolve into AI and the whole digital transformation, which
- 19:09has a different set of rules compare that does not involve
- 19:15the same level of physical infrastructure that
- 19:18telecommunication networks or telecommunication area involves.
- 19:23Because normally in the telecommunication you have the
- 19:25network operator in the country with either cell towers and
- 19:29cables and whatnot. And in those two countries, they
- 19:33need to communicate today on AI. You if you take an open AI
- 19:38example, they have the infrastructure in a couple of
- 19:41locations. They don't need infrastructure
- 19:43in all the countries to operate. And they have yet the impact of
- 19:47open AI, it's across the whole world regardless.
- 19:52So, so therefore the rules of engagement, the rules of the
- 19:57regulations and and governance around.
- 20:00AI is very different from the in governance around telecom that
- 20:06because you normally needed a national company in the national
- 20:09border with the national infrastructure to operate.
- 20:13In this case you don't. So therefore, those rules need
- 20:16to change, the governance model needs to change to make sure
- 20:21that the impact if then those negative impacts are, are
- 20:25mitigated in such a way that that that we can reap the
- 20:29benefits. Now, let me give you an example
- 20:32of, for instance, if I take misinformation as one example,
- 20:36if I from I'm, I'm from a country A and country B is some
- 20:42company, the country B is interfering in my election, in
- 20:46my national election. And then I'll and then using AI
- 20:49to generate all sorts of misinformation.
- 20:52So country A will write to country B say you you country B
- 20:57you I mean, you are inferring creating inference in my
- 21:00election. But that country B is not the
- 21:03government that owns that particular company and that
- 21:06particular company is operating within the legal framework of
- 21:10that legislation. So it's not infringing any law
- 21:14of that national government yet is impact the other country with
- 21:21all the misinformation I go. So you see the absence of that
- 21:25international governance and framework basically makes these
- 21:31negative impacts to go pretty wild, I have to say, and with
- 21:36the impact that we all don't want.
- 21:39I see what you mean when you said earlier that the UN is all
- 21:44about having an environment of conflict and different agendas
- 21:51coming together. Yes.
- 21:53And that's why I say that those difficult conversation of, of,
- 21:59of how to find that right balance that the address the
- 22:05need to profit from the private sector to start with, the need
- 22:11to evolve at the from governments that those companies
- 22:16operate in those legislations, but also the need to protect
- 22:20human rights in those countries that will have the negative
- 22:24impact. That's the the fine balance
- 22:27United Nations needs to play and United Nations needs to be
- 22:31equipped to know what he's talking about in this area,
- 22:34which is very new. Because you can have also
- 22:38situation where an early implemented regulation can stive
- 22:46basically stop innovation or a late implemented regulation of
- 22:51framework will, will will be too late to address the negative
- 22:57impact. So that that balance of time
- 23:01balance of what what is what the regulation addresses balance of
- 23:07the, the, the impact of the, the regulation and the impact of the
- 23:12profitability from the private sector.
- 23:14It's a fine balance that we need to operate.
- 23:16And there is no other entity other than the United Nations to
- 23:20do that, to facilitate that, to host that, to create that
- 23:24environment or to bring this convening power to bring all
- 23:29these players together to really come up with something that we
- 23:32can all be proud of or the next generation can be proud of as
- 23:38they look back into what we did in this particular century.
- 23:41On this topic of innovation, we have a very timely question from
- 23:48LinkedIn from Greg Walters who says how do you plan to
- 23:54encourage both innovation and ethical use worldwide, including
- 24:01among non member organizations that the UN does not directly
- 24:06regulate? It's almost like if you ask me a
- 24:09question, let's say I'm in the private sector, me and my
- 24:14brother, we want to, you know, make some profit.
- 24:19One of us do the legal way, the other one do the illegal way.
- 24:23And we all make profit because but what?
- 24:25One is selling drugs, the other one is selling mobile phones.
- 24:30So what is core between that week?
- 24:33What's the core value that we can push?
- 24:37We can all agree whether one is for profit or whether one is for
- 24:44nonprofit, it is really this ethical value that you just, I
- 24:48just, you just mentioned, meaning that yes, you will have
- 24:51criminals that will try to use the technology for criminal
- 24:55purpose. And if you don't have the
- 24:57regulation that will basically amplify and, and, and promote
- 25:05more criminals to join that particular enterprise OR that
- 25:08particular practice. So we need that regulation to
- 25:11make sure that we it is a deterring fact and the value
- 25:16that we bring and these ethical values that we all got are
- 25:20guided to through is an ethical value that will bring both the
- 25:24nonprofit, the profits and the social civil society to work
- 25:27together. Now we don't have a world where
- 25:31only the people that are have good intentions operate.
- 25:36The whole world has always a segment of population that think
- 25:39that they can use criminal activities to to perform their
- 25:44act. But if if that if there's no
- 25:46regulation that criminalize some of these activities, of course,
- 25:49that becomes a challenge. Now let me talk go about talk
- 25:52about the country itself. Let's say that the one country
- 25:55say, I'm not going to adopt that, but we are in this
- 25:58intertwined and global world, meaning that if if an individual
- 26:05in a particular country is, is committing a crime, that is, is,
- 26:13is basically an international, there's an international law,
- 26:17international regulation that say that is a crime.
- 26:20That individual, yes, maybe in that particular jurisdiction
- 26:24will not be prosecuted, but it can he or she cannot travel out
- 26:30of that jurisdiction. So we need to have something,
- 26:33even though we, we understand that there might be situations
- 26:37where in some geographies or or some jurisdiction that
- 26:43particular country or that particular jurisdiction there
- 26:47is, did did not adopt that. But as I said before, we don't
- 26:51also stop and sit comfortably with, but we also use advocacy.
- 26:58Think about human rights. I mean, human rights was not
- 27:01adopted by the many countries that are adopted today.
- 27:04And we did, we, we didn't say human rights.
- 27:07Those who didn't adopt, we just sit, sit comfortably, do
- 27:10nothing. We say no, we appreciate for
- 27:13those who adopted and, and as part of their, their national
- 27:16legislation, we continue. And that's the work of the
- 27:19office of, of human rights continue to push and to, to, to
- 27:23work with those countries that we didn't adopt to incorporate
- 27:27in the national legislation. So, so we have to they have this
- 27:292 front approach of for appreciating those the early
- 27:34adopters, but also encouraging the late adopters to come on
- 27:38board. And again, that combination of
- 27:41advocacy combined with governance.
- 27:45Exactly. This is an excellent time to
- 27:48subscribe to the CXO Talk newsletter so we can update you
- 27:52on the amazing shows that we have coming up.
- 27:57So this is a question from Arsalan Khan, who says the AI
- 28:03race involves algorithms and data collection.
- 28:09Are there data sets and algorithms that are available
- 28:14from the UN for other nations to use?
- 28:18And is the UN using AI to judge or evaluate other countries in
- 28:24different areas in order to direct resources?
- 28:28So I think he's really, he's really asking these two things.
- 28:301 is about data and algorithms available for the common good
- 28:36and #2 the use of AI internally at the UN.
- 28:41My role is it has two prongs. Right part is to equip United
- 28:47Nations to leverage on on technology and AI to optimize
- 28:55and and, and improve performance and, and, and also support can,
- 28:59can support support countries by basically support programs that
- 29:03support countries on, on, on, on on the use and of AI to make
- 29:10sure that it it, it, it, it accelerates the SDGS and and,
- 29:14and older. So, so that's the internal
- 29:16element. We, we do have internally, we
- 29:19also have a policy we published recently a policy on AI, but
- 29:26also we have over almost more than 700 use cases that are that
- 29:31are that are that are ongoing in different areas.
- 29:34I mean AI use cases. So that's one area.
- 29:36The other area is really how how do we then make sure that the
- 29:46the both because AI, there's three components, right?
- 29:50There's one is the the power computing power.
- 29:55The second is the algorithm and the third is the data and the
- 30:03computing power in it is of course we because the United
- 30:08Nation we have to look at all the things including carbon
- 30:12emission. So we are concerned about the
- 30:16increase of the, the need, the increased need of power and the
- 30:22increased, I would say emission of carbon emissions that these,
- 30:29these will, will, will, will, will, will, will have specially
- 30:33for companies that they had already by committed to, to, to,
- 30:38to, to carbon 0 emission by 20-30.
- 30:40Now they're basically saying they will not be able to meet
- 30:43that because of the investments that they're making in the
- 30:47computing power space. Now on the algorithm, of course,
- 30:52it's this whole innovation around it.
- 30:54There is a number of young professionals that are working
- 31:01very hard to make those algorithms in mind.
- 31:04Our hope is to make those algorithms to address to not be
- 31:09too biases to make sure that the some issues are under the
- 31:13accuracy of those algorithms in terms of generation of the of
- 31:17the response or or elimination of the and hallucination it
- 31:22takes place. And let me deal down to data.
- 31:26And that's where also some concern we have some concerns
- 31:28around that on the data side. One is of course, quality of
- 31:34data. So the so the Global Digital
- 31:36Compact talks about the governance on that.
- 31:41That's, that's 111 concern. The other is to exacerbate some
- 31:45the, the, the exacerbation of the biases these data have,
- 31:50because we have, our data has a number of biases.
- 31:53And, and if we just let the algorithm use that data without
- 31:58the human intervention, that's why the secretary general likes
- 32:00to say the humans are in charge, not the AI, right?
- 32:05To make sure that we address the, the, the, the, the, the,
- 32:08those issues. When, when I, when I was at the
- 32:11World Health Organization, I was CIO there back in, in 2018, we'd
- 32:17already talking about how can we make sure that the trusted
- 32:21source of data is the one that bots and algorithms used to make
- 32:26sure that we don't have negative impact or minimize negative
- 32:29impact. So there are some some work
- 32:32being done in that space to see how we can do with some
- 32:38discussion to say, for instance, we could have a data set that
- 32:40algorithms test. Because you see, if I, if a
- 32:43company in the, let's say, North America creates an algorithm,
- 32:47tests that algorithm with the data from North America and uses
- 32:50that algorithm to generate content for Africa, you might
- 32:55have a lot of issues there, right?
- 32:59So we want to make sure that the geographies and the contextual
- 33:03issues of each geography is addressed as these algorithms
- 33:07generate the different responses in the context of where the
- 33:13question is being asked for the context of the country where the
- 33:16question is being asked. So, so that is something that we
- 33:19still working in terms of how to to to establish that.
- 33:22And then the the proposal of establishing an AI office that
- 33:28which is part of the the the pack of the future is really to
- 33:31address these these these area. But I want to close, I want to
- 33:36also address the issue of equity.
- 33:40If we want to make sure that the fact that, I mean, the fact, the
- 33:46fact that the labor force that trained the algorithm is, is in
- 33:52the global S, we do not exploit the value of the work they're
- 33:58doing and the remuneration of that work appropriately.
- 34:02And we have seen some, some, some issues around that,
- 34:05especially the International Labor Organization has
- 34:08highlighted some of the issues around the remuneration for
- 34:12those workers that train algorithms and they, and they're
- 34:17sitting in the Global South and the remuneration is not really
- 34:21appropriate for the work that they're doing and very and
- 34:24versus the profit those companies are making.
- 34:26So, so yes, we have some issues on the data, the, the data side
- 34:31and we and we need to work towards that.
- 34:33We have, we're not where we would want to be at, but we have
- 34:38now the Global Digital Compact and we can chart the way to make
- 34:42sure that those issues are addressed.
- 34:44It's very interesting hearing you talk about issues around
- 34:49equity. You have a more nuanced view
- 34:53than many in the private sector because of the the charter.
- 35:00The goal of the UN is operate for unity and for good, as
- 35:06opposed to the charter or the goal being profit.
- 35:10They are not mutually exclusive. Let me give you an example of
- 35:13the pharmaceutical area. Because I was the CIO at the
- 35:16World Health Organization. I can always talk about that.
- 35:20If I take the hepatite C treatment as one example, right,
- 35:26the treatment of Hepatite C when it, the company that innovative
- 35:32that created that, the cost was around almost close to $100,000
- 35:36per treatment. OK, then I mean, now if you go
- 35:43to India, the cost of that treatment is $1000 for the whole
- 35:46treatment. OK, Then you can think, all
- 35:49right, if as a company I will lose profit, but not really.
- 35:55Because if your market is, if you are making, I don't know, a
- 36:00billion dollar out of 100,000 people, you might actually
- 36:03making $10 billion out of, you know, 5 billion people at a much
- 36:08lower price. When we talk about equity, it's
- 36:11not just equity to lower the profit of the company isn't
- 36:15that's not our intention to bankrupt any private sector.
- 36:20The intention is to say, yes, you can have a client.
- 36:23And now I come back to AI. The AI today, many companies
- 36:28charge somewhere around $20.00 per month per user to use the
- 36:33AI. It beat ChatGPT 4 or other,
- 36:40let's say Microsoft Copilot, maybe Gemini is a little
- 36:44cheaper. But all to say that yes, you can
- 36:48have the $20.00 per month charged to use generative AI and
- 36:55that $20.00 per month, maybe your market space will be let's
- 37:00say 500,000 people that use that.
- 37:03But if you make that price $5 per month or $1.00 per month and
- 37:09you get 2 or or, or or 5 billion people using it as a private
- 37:15sector company, you still make money.
- 37:17But then you also contribute to the access to that.
- 37:22So it's the same concept in the pharmaceutical area where I
- 37:26mean, I think there's a say that you can have one, one $1 million
- 37:30from one person or you can have $1 million from 1,000,000
- 37:34people. The most sustainable $1 million
- 37:37is the one that comes from 1,000,000 people and not the one
- 37:40that comes from one person. So that's really the concept
- 37:42that we're trying to, to say. And I'm trying to, to, to put it
- 37:45forward to say access, equitable access doesn't always mean lower
- 37:52margins of, of profit, but but it means more access to people
- 37:57that can afford to make it affordable, but also create
- 38:02these equitable world where you don't have those privileged ones
- 38:06that have and those unprivileged ones that don't have.
- 38:09I see that you're not making a value judgment, but rather you
- 38:14are trying to align incentives around the goals of each of the
- 38:22participants, whatever, whatever is important to them.
- 38:25You're trying to unify the incentives at the end of the day
- 38:28to align activity around that goal.
- 38:32Exactly. And in United Nations, that's
- 38:34what we do pretty much all the time, to have these different
- 38:40perspectives to come to the table and see how we can extract
- 38:43value. I mean, if you look at the, I
- 38:46mean, the, the example I gave is an example where the whole World
- 38:50Health Organization is very keen and pushes it, be it on, on, on
- 38:56appetite C that I mentioned, be it on HIV, be it on, on COVID
- 39:02vaccines, that when, when, when we had COVID.
- 39:04But on technology, we need to follow the same concept.
- 39:07We need to to make sure that it's not just a price that
- 39:11established for those who have. It's a price that establishes
- 39:16according to geography that allows that access, but but
- 39:21without putting the company at risk of go bankrupt because they
- 39:25cannot generate the profit that they they need to generate.
- 39:29We have a question from Lizbeth Shaw on Twitter, who says how
- 39:35does the digital divide impact the AI divide among countries
- 39:42and how does that impact the ability for countries to carry
- 39:46out the ethical use of AI? The digital divide and AI divide
- 39:53go hand in hand, but we have work being done to, to, to.
- 40:00Minimize that digital divide, I mean in the International
- 40:03Telecommunication Union and other UN organizations are
- 40:06working towards that. I mean access to broadband still
- 40:11is still a problem. You still have a large segment
- 40:15of the population I think if I'm not wrong somewhere on 2 billion
- 40:19that don't have access to, to broadband.
- 40:22So, so that, but the, the progress of that is very, I mean
- 40:26in the last couple of years you see good progress around access
- 40:32to broadband. But we want to make sure that
- 40:35these new, some private sector would call it new customers that
- 40:40embrace broadband. But we, we want to make sure
- 40:44that As for those who are already in that digital
- 40:49ecosystem, we do not exacerbate these divide with those new ones
- 40:55that will come. So we want to address the
- 40:57digital vibe within the ones that already have access.
- 41:01The I mean, making sure that not we want to make sure that the
- 41:05digital divide and the digital, the access to those tools in for
- 41:11those, the segment of the world population that have access to
- 41:14broadband is addressed to ensure that the newcomers, they benefit
- 41:20from it. It's true that we might not be
- 41:22able to change the past, but we can change the future.
- 41:27And that change of the future starts with all of us working
- 41:31together to make sure that the AI is not just for those
- 41:38privileged ones. And those that are not
- 41:41privileged ones, they, they, they, they, they, they become
- 41:45basically the source of data without taking any benefit.
- 41:48When I was the CIO of the World Health Organization, also the
- 41:51director of digital Health, I mentioned, I, I wrote an article
- 41:55about data and the new data and the new, the new, the new money.
- 42:02Because if my data set goes to benefit a company, we have to
- 42:08also ask how much of that benefit comes to me as the
- 42:13source of the data. So those discussions, they're
- 42:15difficult discussion. They need to, to take place.
- 42:18I think there was some, some article I saw about the, you
- 42:20know, the, the, the, the, the writers that wrote books, the
- 42:24professors that wrote articles and the AI in is in use those,
- 42:29those, these datasets to, to ingest into the model and the
- 42:32model generate response. What's the benefit that goes to,
- 42:35to, to, to, to, to, to, to to those we do not want to repeat
- 42:38previous centuries situations where there was that our early
- 42:42adopters, they extract, they extract greater benefits toward
- 42:48to, to to the detriment of those who actually provide the ground
- 42:52groundwork to make to make that particular endeavor a success.
- 42:57So I'm very happy that the United Nation is on the point at
- 43:04the right time at the beginning of it.
- 43:07And I'm very happy the fact with the fact that we are very
- 43:11conscious about the pace, fast pace of, of AI.
- 43:16Of course I'm worried that we might lose the race, but but I,
- 43:19I'm, I'm, I'm confident that with the support of all
- 43:22countries and all members of the United Nations, we will keep at
- 43:26pair with the pace of the technology to ensure that the
- 43:30benefit is extracted for everyone.
- 43:33We have another question from Greg Walters on LinkedIn, who's
- 43:39asking, can you talk about differences between the approach
- 43:43of the UN when it comes to AI and ethics versus the EU, which
- 43:49has a very strict stat or a detailed set of regulations?
- 43:56We work with the old, the different interests that are
- 44:02sometime far apart. And then and then what the
- 44:06result of the adopted or endorsed pact is.
- 44:14It's a compromise between, let's say the most conservative
- 44:17approach to the most progressive approach.
- 44:22I think if you go to Europe that think there's perhaps the, the,
- 44:27the, that gap between the conservative and progressive
- 44:32might be as much narrower compared to the United Nations
- 44:37space where you have these 193 countries, right.
- 44:41So as a result, of course, you might have the EUAI act that
- 44:51that that has a number of details.
- 44:54That's one that that that is more maybe that goes to to the
- 44:59level of detail where the United Nation takes operates a little
- 45:03bit of a higher level to allow these these to allow countries
- 45:08to either navigate towards a progressive approach or towards
- 45:14conservative approach. Within that framework that the
- 45:17the member states agreed that they can safely and they can
- 45:22comfortably operate. That's one one's angle.
- 45:26The other angle is really when do you bring the regulation?
- 45:30Do you bring at the beginning when when the hype is high or do
- 45:33you bring when later on after the hype cycle?
- 45:39You know, one of the, the, the, the, the examples I used to give
- 45:44and I still give, give now is that we all tend to forget that
- 45:48technology has always a hype cycle, a cycle, a period of a
- 45:54lot of promises. And we need to allow that
- 45:57technology to go through that hump to, to, to then stabilize
- 46:01into what it, it can really develop.
- 46:04Because that hype cycle, if you create a regulation during the
- 46:07hype, that regulation will be very strict because the hype
- 46:11will promise things that actually technology doesn't
- 46:14deliver. Now if you create a regulation
- 46:17after that hype cycle, then that regulation will not be as strict
- 46:23because the disillusionment of the hype cycle will take place.
- 46:28And then there will be in that stable space where then what did
- 46:35the technology can really do is not hyped.
- 46:39Therefore the regulation will not tackle and hyped statement
- 46:44or a hyped intention or a hyped promise of that technology.
- 46:51So all to say that you have these two perspectives of
- 46:56Progressive Conservative and then the perspective of when the
- 47:00regulation comes. We have another really
- 47:02interesting question from Twitter, again from Arsalan
- 47:04Khan, and I'll ask you to answer this one very quickly because
- 47:08we're simply going to run out of time, he says.
- 47:11How is the UN partnering with organizations outside the UN and
- 47:17are those organizations sharing their algorithms and data with
- 47:21the UN? And very quickly, please.
- 47:23We have a framework that allow in a number of agencies that are
- 47:26framework that allow the private sectors and the society to
- 47:30engage. I mean if I take an
- 47:31International Telecommunication Union as an example or
- 47:35international labor organization as an example, we have those
- 47:39frameworks. Now do they share the algorithm
- 47:41with us? That question the answer is no,
- 47:45but we do we we are as I mentioned before, we are working
- 47:48to have the on the data side to to see would make sure that we
- 47:52create a framework on data governance that allow this those
- 47:58algorithms to lean on in on on data that is accurate, but also
- 48:02on data that that does not have too many biases and things and
- 48:06and and and and some of the ethical principles are are there
- 48:09too. So it's a work in progress, but
- 48:12the no companies shared algorithm with the United Nation
- 48:15because that's part of empirical property.
- 48:17But that takes us to the next question of open source.
- 48:20And we, we are actually we are, we are we convene the open
- 48:24source community to come together because we see a big
- 48:27role of open source community in this, even in this in this space
- 48:31to make sure that the alternative to proprietary
- 48:35systems continues to exist. And we want the open source
- 48:38community to to thrive in this AI era.
- 48:42And I assume that you're trying to exert some influence on the
- 48:48open source community in order to approach their technology
- 48:53work within the framework of UN goals for inclusiveness and
- 49:01being equitable and so on. Yes, and I have to say the open
- 49:04source community is mature on its own without the UN and we
- 49:08just we, we came now to say we, we had actually the second the
- 49:14open source for for good summit that brought all the open source
- 49:18community together to make sure that we we also look at those
- 49:23open source tools that are very important for critical digital
- 49:26public infrastructure that the benefit from from the resources.
- 49:30But the, the Open Source committee has been working very
- 49:33well as as a community, but we want it to go to the next level
- 49:39to especially now that with AI, we want to make sure that the
- 49:44fact that AI providers or AI companies provide a space of
- 49:50free use that does not relegate the open source community to not
- 49:57having to do anything, right. So that's really where where we
- 50:00want to make sure that that community is providing an
- 50:03alternative to many developing countries and we want that to
- 50:07continue to to happen. And of course with the UNS
- 50:10convening power as you described earlier creates the an ideal
- 50:15forum I would think to help infuse these values into what
- 50:20the open source community is doing.
- 50:23Exactly. Exactly.
- 50:24Can you offer advice or lessons to business people based on your
- 50:31experience with AI, ethics, governance and so forth at the
- 50:38UN? Do not give up, no matter how
- 50:41difficult it is that the engagement from from the UN to
- 50:45private sector or from private sector to the UN or to with the
- 50:50government. We should not give up.
- 50:55No matter how bad it looks the whole the OR how huge the task
- 51:05looks ahead. No, we should not give up.
- 51:09So my advice is persistence, perseverance to make to ensure
- 51:14that we all come out of these proud, but also our the next
- 51:19generation come out of it proud from the United Nations.
- 51:23Please be assured that we do not give up no matter how bad the
- 51:28situation. Look, you see them the world and
- 51:30then the number of wars, one might say, look, this looks so
- 51:33bad. We never give up, we keep, we
- 51:38persist in this, in the in the pursuit of a solution that
- 51:43brings peace. And in the case of technology,
- 51:46we pursuit the solution that brings equitable access, that
- 51:51brings creates do not harm component of of AI hinged in
- 51:57this 10 Principles of ethical Use of AI promoted by published
- 52:03by UNESCO that we want to make sure that the world adheres to
- 52:07and we bring extract and and and enjoy the great benefits that
- 52:12artificial intelligence can bring to this world.
- 52:16And with those inspiring words, we are out of time.
- 52:20A huge thank you. Bernardo Mariano Junior is
- 52:24assistant secretary general and chief information technology
- 52:29officer at the UN. Bernardo, thank you for coming
- 52:32back to CXO Talk. Thank you, Michael.
- 52:34Thank you. It's been a pleasure and thank
- 52:36you everyone for this great engagement.
- 52:38Thank you. And thank you to everyone who
- 52:41watched, especially you folks who asked such great questions.
- 52:44You guys are amazing. Now, before you go, subscribe to
- 52:52our newsletter, check out cxotalk.com, subscribe to our
- 52:56YouTube channel. We have extraordinarily
- 52:59extraordinary shows coming up next week.
- 53:01We're talking with someone from the CIA about these issues.
- 53:04That should be interesting. So check it out.
- 53:07And thanks so much, everybody. Again, Bernardo, I'm so grateful
- 53:10to you. Thank you for being here.
- 53:12And we'll see you again next time.
- 53:14Take care, everybody. Bye bye.