The UN Sustainable Development Goals were set to be achieved by 2030—yet only 17 percent of SDG targets are on track. In Asia, research has shown that the region will only achieve the SDGs by 2062 if it carries on its present trajectory. Can AI play a meaningful role in accelerating Asia–Pacific’s initiatives to achieve the goals and improve people’s lives while protecting biodiversity? In this episode of the Future of Asia Podcast, McKinsey’s Angela Buensuceso talks to three experts to explore the question, look at the challenges, and give insights into how to collaboratively develop ethical AI solutions aimed at social good. An edited version of their conversation follows.
Angela Buensuceso: Hello, and welcome to the Future of Asia Podcast. My name is Angela Buensuceso and I’ll be your host today. In this episode, we explore the transformative power of AI and its potential to address the social challenges defined by the UN Sustainable Development Goals. Drawing from a recent McKinsey report, AI for social good: Improving lives and protecting the planet, we examine how AI can help make a positive impact on society and the environment, particularly in Asia. I am joined by three distinguished guests: Ahmed, Ankit, and Jennifer, please introduce yourselves to our listeners.
Jennifer Echevarria: Hi, I’m Jenny Echevarria. I lead the Enterprise Data and Strategic Services for Globe Telecom. Globe is the leading telecoms provider in the Philippines, and we serve over 60 million customers, with close to two million on broadband. We cover both B2C and B2B, so we also serve small, medium, and large enterprises. And, while we’re a telecom company, we also have investments in fintech, adtech, and healthtech.
Ahmed El Saeed: Hi, Ahmed El Saeed here. I present the United Nations Global Pulse and I’m the regional head for Global Pulse in Asia–Pacific. Global Pulse is the UN Secretary General’s Innovation Lab where we work at the intersection of innovation and human sciences. We work across the Sustainable Development Goals (SDGs) utilizing new, emerging technologies and innovation capabilities to accelerate delivery against the SDGs and to support our member states.
Ankit Bisht: Hi, I’m Ankit Bisht. I’m a partner in McKinsey’s Digital Practice where I lead our Noble Intelligence initiative. This is where we apply our QuantumBlack data and AI capabilities toward social causes. I’ve recently led McKinsey’s extensive research on how AI can be applied by the private sector, government, and civil society to positively impact the SDGs.
Angela Buensuceso: How is Asia advancing on the 17 SDGs for 2030 and what impact is AI having on this progress?
Ahmed El Saeed: The world is significantly offtrack in achieving the SDGs by 2030. A recent report showcases that globally only 17 percent of the SDG targets are on track or have been met. Unfortunately, also 17 percent of them have regressed. That is an alarming statistic. In Asia specifically, reports have shown that the region would require at least 32 additional years beyond 2030 to achieve the SDGs, indicating that they could not be achieved before 2062.
The progress that has been made in Asia is both uneven and inadequate. The uneven progress is very clear, because there is a clear country divide within Asia. Countries are divided between developed countries, middle-income countries—especially upper-middle income—and least-developed countries that are either landlocked or small island developing states within the Asia–Pacific region. There is also another form of divide and inequality, which is urban versus rural. Unfortunately, in many countries, we see that in urban settings, the quality of life is advancing, AI is being utilized, and smart city concepts are being introduced, while in rural settings, access to and the quality of those basic services are not guaranteed.
Because of that, Asia is not progressing well. We need to pay more attention, especially in terms of climate action as this is the most relevant SDG [SDG 13] for the Asia–Pacific region. AI can have significant impact across SDG themes, from supporting climate action, optimizing energy consumption, improving access to and coverage of quality education, and monitoring biodiversity—to mention a few.
Angela Buensuceso: Ankit, what does McKinsey’s most recent report say about this?
Ankit Bisht: There are many great applications that we’re already seeing from AI, whether it’s modeling proteins, screening drugs, or designing new vaccines (for example, during COVID-19 there was a significant acceleration in the design of vaccines), or in targeting aid to the most vulnerable populations, as well as addressing problems around climate change, such as immediate, early warnings for natural disasters. There are many AI applications that are already out there today. Generative AI (gen AI) could significantly increase and democratize access to those new capabilities as well.
We recently conducted extensive research on this topic. We spoke to dozens of experts across countries and continents and analyzed over 1,000 AI grants and investments into 20,000 AI-focused companies. There were three key findings.
The first is that the momentum is exciting. We found evidence of the deployment of over 600 AI-enabled use cases across the 17 SDGs. When we did similar research five years ago in 2018, the number was 170. This is approximately a threefold increase. In addition, when we looked at the 20,000 plus private sector AI investments, we saw that 40 percent of the investments were linked to improving one or more of the SDG areas.
However, the second finding is that a lot of these deployments are currently stuck at pilot stage and are not scaling to impact the millions of people in the way that they could. Most grants and efforts are happening at what we consider the seed stage rather than the scale-up stage of creating impact.
The third finding is that there is inequity by markets. The developing markets are not seeing a fair share of AI investment in the SDG causes. In Asia, for example, while China and some of the developed markets are seeing good investments into AI, the developing countries are seeing a far smaller share than what is required, based on their GDP and per capita normalization.
Jennifer Echevarria: To put it in the Philippine context, some of the biggest sustainable development issues for our country—and we’re likely one of those that will take three decades to reach the targets—are poverty, hunger, education, and health. For example, the poverty incidence in the Philippines increased from 18 percent in 2021 to 22 percent in 2023, which means that 25 million Filipinos are poor. And looking at the recent Q1 survey of Social Weather Stations (SWS), 14 percent of Filipinos have experienced involuntary hunger.
The challenge in the Philippines is that there’s not just one definition of who is poor. If we don’t have an agreed definition, the scale at which we have to solve the problem will be a challenge. When I said that 25 million people are living on or below the poverty line, I meant that is a person who earns less than $220 a month for a family of five. But other groups say it is $500 a month—if we use that amount as a threshold, then 70 percent of the people we interviewed are borderline poor.
The reality in the Philippines is that, while the government has made strides in trying to reduce poverty, income is not high enough because of high inflation and climate change, specifically El Niño. We’ve had farmers tell us that they had to move their planting cycle by a month—imagine the impact on their income.
The same is true for education and healthcare. The Philippines is one of the youngest nations, with an average age of 26. We sit on a demographic dividend, but we always rank low in international assessments. The irony of it is that the spending for both education and healthcare is not very far from the Philippine’s neighboring countries, but outcomes are very different.
That is why I think AI could really be powerful to help us address the sustainability goals. Imagine how AI could help improve resource allocation; how it could help farmers predict crop yields? Or in education, how could we personalize the learnings? How could we help teachers identify students who are falling behind and make the whole learning experience more meaningful? There’s so much that we can do; the challenge is how to do it at scale.
Angela Buensuceso: What is working in terms of the application of AI for social good? Can you share some concrete examples of how organizations are deploying AI to address social challenges?
Ankit Bisht: In our research, we found that the experts concurred roughly with what we’ve found in terms of deployments and where there’s highest potential. We found five SDGs with highest potential: good health and well-being, quality education, climate action, affordable and clean energy, and sustainable cities. We also found that 60 percent of nonprofit AI deployments were focused in these areas. However, there are other areas that could benefit from significant uplift, where there is potential but there isn’t enough deployment. For example, more could happen in the SDGs that cover no poverty [SDG 1], zero hunger [SDG 2], and even peace, justice, and strong institutions [SDG 16] by reducing misinformation.
Let me give you an example in terms of being able to support farmers in increasing yield. Outgrow is an agricultural farmer and engagement platform in India that connects farmers with high-value products and services, including automated soil labs and microclimatic weather devices. Using this AI-based platform, farmers can predict and detect crop diseases in real time. They can learn how and when to irrigate. They can also connect with agricultural experts in multiple Indian languages, which are auto translated using neurolinguistic programming (NLP) and AI. Outgrow’s platform is currently supporting more than 18 crop types and about 200,000 farmers. But the question is, how do we make sure such platforms are deployed across all the countries in Asia that can benefit from it—from Indonesia to the Philippines to Vietnam? How do we create that kind of scale to impact millions of lives?
Angela Buensuceso: Jennifer, did you have any examples from Globe?
Jennifer Echevarria: We use a lot of data and AI at Globe. That’s primarily driven by the fact that the Philippines is a prepaid country—97 percent of cellphone users are prepaid, so we don’t know much about them. As a result, we have to use AI to look at our data and design models to be able to segment our base and deliver products and offers. For example, we had to build AI models to know the life stage and affluence of our customers, because life stage will dictate what they need, and affluence will show how much they are willing to pay. We use data for our models to be able to determine whether they use the internet less or more, because that also spells a big difference.
The way we try to create products and serve the next best offer is through AI. In fact, we’re now going into gen AI because it’s relevant and we’re able to cover even the vulnerable segments of our base. A good use case would be GCash in terms of addressing financial literacy and inclusion in the country. Today, more than 90 million Filipinos use GCash. We’ve also used data and AI to build GScore, which has allowed us to extend credit to about 4 million borrowers. If you compare that to two or three years ago, the unbanked number of people has dropped significantly. It used to be 70 percent and is now less than 50 percent.
We have deployed AI internally to improve our operations and how we serve our customers. We have also used AI for GClimate, which provides us with a five-day forecast for cell site disruptions due to weather. It therefore enables our field teams to reduce service interruptions from typhoons—this is relevant because the Philippines is hit by roughly 20 typhoons a year. We are one of the countries that is most at risk for natural disasters; they can set us back by around 1.7 percent in our GDP.
These are just example use cases, but the long-term vision is to be able to expand these efforts to multiple domains, such as healthcare, urban planning, food, and agriculture. This ambition, however, hinges on collaboration. If we want to achieve scale and success, we actively need to seek partnerships with data providers, model builders, private institutions, and even local government units across the Philippines that share the same commitment to do social good. We feel we cannot do it alone as a private institution. We want to pool resources and expertise to create impactful solutions that will address the Filipinos’ problems.
Ahmed El Saeed: Across the region, we’ve seen how AI helps in providing information to citizens on available services, in streamlining access to quality services, and in improving the targeting and expediting processes through automation. One concrete example here is an excellent model of collaboration that we have between the United Nations and the private sector, where McKinsey plays a key role, and also a partnership with Globe. And further, an initiative called DISHA—Data Insights for Social and Humanitarian Action—that looks at population mobility postcrisis to inform and target aid. This can help better understand the most pressing needs of those populations in partnership with telco providers like Globe, which are developing unique collaborative models in the region.
In addition, we have another concrete use case that looks at assessing damage postcrisis—for example, with earthquakes, looking at high-quality satellite imagery, automating the analysis for those images, and understanding better the magnitude of damage and the locations to expedite aid going to those areas. It’s a significant effort that really helps save lives on the ground. These are just a couple of examples under the DISHA initiative. We are looking at scaling it across more countries globally and building in more AI applications to see how we can amplify its impact.
This is a real showcase of how the private, public, and social sectors can come together to truly deliver impact that no one individually can. Our aspiration in supporting initiatives such as DISHA is exactly that—how can we together create impact at scale across geographies? This is also a good example of how you can use telco data or satellite imagery to create benefits for society.
However, we also need to think of how we can use granular data from banks, grocery stores, and all kinds of consumer companies (which is where all the data sits) to the benefit of society. This is an open call for everyone to think about how they can partner and create coalitions similar to DISHA and connect to create something that is beyond what they can deliver individually as an institution.
Angela Buensuceso: Let’s now look to the challenges that we face with AI for social good. Can you share more on the biggest challenges with harnessing this technology, and what strategies and unlocks are needed to scale it in Asia?
Ahmed El Saeed: Scaling for AI innovation is challenging. With AI and other types of emerging technologies, we need to realize that scaling is a multiplayer game. You cannot do it alone; you need to partner. Those kinds of partnerships are critical, and we need to continue to do more of them if we want to enhance the potential impact that AI can bring in the advancement of sustainable development.
Other challenges can include the scarcity of talent and capabilities. In the social sector, this continues to be difficult—attracting the right caliber of person, finding individuals who have the right capacity and who are able to support AI for social good. Cost implications need to be considered as well. We want to build AI solutions that have sustainable business models, factoring in computing power, requirements for AI models, and connectivity issues. In many areas where the United Nations works globally, only two-thirds of the people are online; that is one in three people. That is also something that we need to consider.
Finally, we need to think about the collaboration model and the readiness of public organizations to partner effectively with the private sector, trying to shed light on models that work and disseminating knowledge around how we can partner more effectively in the advancement of AI for social good. Again, scaling is not easy; it takes a long time. We need to make sure that we have the right players, shared value partnerships, and are working toward the same goal.
Jennifer Echevarria: I think sharing information and forming partnerships with constant conversations are important to access the valuable data. One of the major challenges is data availability and quality. If we’re talking about the relevant data sets to address SDG-related topics, they are challenging to create or curate as you need to get them at a high frequency and to cover most of the country. In the Philippines, it’s not easy to get data from less developed regions, while urban regions are highly connected and online all the time. The irony is that the most vulnerable segments of the population live in the less developed regions and are the ones who will benefit the most if we’re able to help them.
Another challenge is that there are a lot of languages in the Philippines—around 170 languages and dialects are spoken. How can the models learn across all these offline demographic groups so that the risks of exclusion won’t happen? We’re always very conscious of this, about how to make sure we don’t exclude the marginalized communities in AI solutions. What we have tried to do is to marry our internal data with on-the-ground information that we have through our market research, because typically, our market research mirrors the national distribution.
The on-the-ground data serves as our training set for building our models and we infer it to the entire population. We also try to make a point to get feedback. Whenever we do things, we get feedback from users from specific segments, so that it always informs and improves the quality of our data, and hopefully gets better results.
Ankit Bisht: This is a great example of the work we’ve done together where we’ve used census data and market research data to create a much more dynamic poverty prediction approach in the Philippines. That can be applied in most countries because, as you know, census and market research typically happen sporadically. Besides what has been mentioned, I have one other challenge, which I believe, if unlocked, can really create scale. In our recent research, we found that there is a lack of sustainable business models in solving a lot of the SDG challenges through AI. There are some areas where the private sector is quite heavy, where there are clear profitable business models—for example, in good health and well-being, especially in more developed markets, and in other areas such as clean energy.
However, there are many SDG themes where there is market failure or not enough scale. Especially in the social sector, it would help if business models can be created that are sustainable. These don’t have to be for-profit; they can be not-for-profit. But at the moment, there is a lot of pressure on grants and commitments, which are well below where they need to be to achieve the SDGs.
One way that you can bridge that gap is for NGOs to have products and services that can create some kind of funding for themselves in a virtuous cycle. For instance, take Thorn, which is a not-for-profit that builds tools to defend children from sexual abuse. They offer both pro bono and fee-based services to a mix of not-for-profits, government agencies, technology platforms, and private companies.
Similarly, there are a lot of big tech companies that are providing either free or affordable pricing to not-for-profits. Through affordable pricing, they still make their costs back and the not-for-profits have a pricing that they can match in terms of their willingness to spend. These kinds of sustainable business models can have ten times the overall impact that any single AI initiative can by giving a sustainable source of financing that can be used to scale across multiple segments, geographies, and markets.
Angela Buensuceso: How are Asian countries addressing the ethical and regulatory challenges to ensure responsible or ethical use of AI?
Jennifer Echevarria: For data-driven organizations such as Globe, it’s very important that we protect sensitive data. It’s a constant balancing act. At the end of the day, it starts with us and how we are using AI in our lives and in our work. So, at least for Globe, we’ve established some governance in terms of the responsible use of AI in the workplace and we try to make sure that AI addresses real-world challenges. We, of course, safeguard sensitive data because at the heart of privacy for us are the values of trust and respect. You realize that it takes years to build trust, but only minutes to lose it.
Ensuring transparency in how we use AI models is also something that we commit to. Further, we regularly review, test, and evaluate our models to make sure that there are no potential biases. Then hopefully, in everything that we do, we are accountable to ourselves, our company, and our customers.
Angela Buensuceso: Ahmed, do you want to share a perspective on the ethics of AI solutions?
Ahmed El Saeed: This is a very important issue that has received a lot of attention globally. At the UN level, there has been some guidance. The Secretary General has convened an AI advisory board that provides some global guidance for countries. At the regional level in Asia, there have been guidelines developed to give practical steps for countries that wish to establish more robust frameworks for AI development or deployment when it comes to ethics and the responsible use of AI. Essentially, the one thing that I would like to highlight is that it is important to make sure that AI models introduce the concept of being inclusive by design. Making sure that the training data is responsive and reflects realities of different groups is very important.
It is also important to ensure that every step, from design, development, and deployment, streamlines important principles, such as transparency, fairness and equity, human-centricity, and accountability. Those can only be introduced in a consistent way through stakeholder interaction and open communication with the different stakeholders and groups. I think that we are yet to see a lot in that space. It is very important to have transparent communication. We also need to assure citizens of the transparency of the development of AI, and how it can respond to everyone’s needs. There are many guidelines, both released and in the pipeline, that will provide more guidance on this overall issue.
Ankit Bisht: There are many countries today that are building a number of different assets and enablers to allow for ethical and responsible AI deployment—whether it’s principles, laws, regulations, or frameworks coming from industry institutions, or even voluntary commitments and declarations and standards that have been put in place by various leading private sector or other institutions. The one thing that we need to think about is how to ensure that there is inclusion in how we help all kinds of organizations to adopt responsible and ethical AI practices—all the way from larger institutions to smaller NGOs and SMEs that need to be able to navigate the relatively complex setup today. A lot of different enablers are being put in place, but they are not necessarily easy to deploy at the frontline and on the ground. We see a lot of these examples, even in Asia, where there are countries that have established some of these practices, for example, Singapore has a strong framework for how to manage AI and governance.
What are the enablers or assets that can help them navigate this, whether it’s from incident databases to tools to be able to prevent biases and malicious use, or to promote security, robustness, transparency of development practices, or explainability? And, most important, to have the resources to train frontline users, create change management that is easy to understand in local languages and is equitable, and to adapt to the moving target of fast-evolving AI.
Doing this well is going to be critical because there are two sides of this equation. Unless we’re able to navigate the risks through these kinds of enablers that we have talked about to manage ethical and responsible use, we will not be able to get to the scale that we need to impact the lives of millions and millions of people in so many different ways and achieve the SDGs.
Ankit Bisht: Jennifer, Ahmed, do you have any final thoughts that you’d like to share?
Jennifer Echevarria: At the end of the day, gen AI could be a force of creative destruction. I would imagine that more and more life-changing decisions will be made with AI, so it’s very important that privacy and explainability are preserved. I think what could help is to start with the human and end with the human. What do I mean by that? It’s as simple as, for example, in any data model or big data that you want to be able to use, you need to make sure that it’s the human stating what the problem is and that the output is checked. That’s very high level. When I say, “Start with a human and end with a human,” it’s also that AI should address real-life issues.
You end with a human in the sense that the technology and models are there to make sure that you are able to do your transformation sustainably. In the end, what really matters is that you made a difference. You improved the life of someone. You were able to help uplift a nation. It has to start and end with the human.
Ahmed El Saeed: I believe that AI really opens the door and creates a lot of opportunities to accelerate and catch up on sustainable development. We can see how AI is driving growth in many countries. There are many positive examples. We should not let the risks outweigh the benefits that could be realized from accelerating deployment and fully utilizing AI at scale. We need all stakeholders to come together, work toward that common goal, with the right capabilities in place. We need to make sure that while doing so, we leave no one behind.
Ankit Bisht: My ask to everybody listening is to join the party. There are many institutions that are thinking about this and want to do things but are stuck, because they’re trying to do it by themselves. Through initiatives such as DISHA, there are channels to come together and collaborate. That is the only way that we can deliver the proportionate impact that’s required to achieve the SDGs.
There is a part to be played by all actors, whether they be from the private sector, public sector, or the social sector. We are excited to hear from various institutions, organizations, and individuals, and for them to come and explore how they can participate and create more impact through AI for their local communities, their beneficiaries, and their users. We’re here to chat—this is an ongoing dialogue, and we look forward to hearing from other institutions in the future.
Angela Buensuceso: Thank you, Ahmed, Ankit, and Jennifer for joining us on the Future of Asia Podcast. To all of our listeners at home or on the road, wherever you are, thank you for joining the Future of Asia Podcast. If you find this discussion insightful and you want to know more about the topics of gen AI and digital and technology in general, you can visit www.mckinsey.com/FutureOfAsia.
