Pathways to Impact: Miguel Luengo-Oroz  

Conversations with data science for social impact leaders on their career journeys

miguel-luengo
miguel-luengo

How did you find yourself in this field, and your current role?  

I grew up in Asturias in the north of Spain and then followed my academic interests to engineering degrees in both Spain and France. My initial focus was on math and signal processing. My curiosity then branched out to humans — I pursued a Master’s degree in cognitive science, which was essentially psychology and neuroscience. Around that time, I was fascinated by exploring artificial intelligence (AI) and creativity and created an AI that could write poetry in 2003. After that, my interests shifted again — I became intrigued by biology and the elements of life itself.

It was an interesting trajectory: science to humans and then to what’s going on inside humans! My Ph.D. work was on mathematical models applied to biology and genetics. Essentially, we did something like Google Earth of the first 1,000 cells of a vertebrate embryo.

In addition to my academic background, I spent some time in Silicon Valley. It was there that I realized that I wanted to have an impact beyond research.

I have always admired the United Nations (UN) as a place to have that kind of impact. Every country has a seat so the organization represents, as much as possible, all of humanity. In 2011, I applied to a position they were opening in a new unit called the United Nations Global Pulse (UNGP), and I became the first data scientist working for any international organization.

2011 was a big year for the awareness and practice of data science. These were early days for this work, particularly at the United Nations. A precipitating factor for the UN position was the aftermath of the global financial crisis: in mere weeks or months, many countries lost the development progress that had taken them years to build. And the UN — and often the governments of those countries — had limited visibility into these losses. The UN was looking for new ways to measure, in real-time, the well-being of people worldwide. This led to the creation of UNGP allowing the broader organization to innovate and experiment to benefit from the use of data and AI. The organization has now evolved to maximize the impact of digital innovation to address emerging global challenges and crises, minimize the risk of harm, and create pathways to scale and sustainability.

Having worked as a researcher in several areas, I have benefited from multidisciplinarity; for me, it’s always about trying to see the connections. In my career, that’s meant drawing analogies like, “Look, the same way we can see how cells move in an embryo might help us understand where population moves after an earthquake.” I know there are many analogies you could draw with other fields of science that would apply to social impact.

What part of the impact challenge does your work solve? What are you working on right now?  

At UNGP, I’ve been working on so many challenges: poverty, food security, refugees and migrants, conflict prevention, human rights, gender issues, and climate. Our team always works closely with other UN agencies that have a specific mandate and issue expertise. For instance, during the peak of the COVID pandemic, our team created the epidemic models for the biggest refugee camp in the world: Cox’s Bazar in Bangladesh. This was a joint project with the UN Refugee Agency (UNHCR), the World Health Organization (WHO), academic institutions, and private sector partners. Together, we created a digital twin for this camp, to aid in decision-making when crafting health policies for the camp. More broadly, we mobilized a community around challenges and opportunities for epidemic modeling for people that have been forced to flee their homes.

With the Ukrainian crisis, we are supporting UN colleagues from multiple agencies to estimate challenges like numbers of affected people, either refugees or internally displaced populations, using AI to speed up satellite image analysis or understanding future network effects of broader socioeconomic impacts.

What is clear is that to build an equitable future, across all these communities and ecosystems, we need to foster diversity and inclusion, and to make sure that no voice is left behind.

miguel-luengo Dr. Miguel Luengo-Oroz Senior Advisor on Frontier Technologies United Nations Global Pulse (UNGP)

Another interesting area we work in is not focused on what we can do or how we can do it — but on what we shouldn’t do. We work on guidelines and instruments to assess the potential risks and harms of data science projects themselves. When the risks are too great and cannot be mitigated, there are things we must not do.

We also work on the governance of new and emerging technologies. Consider neurotechnology, for example. Imagine something can read and write the neurons of your brain directly, what are the mechanisms for consent? We will need to tackle issues like mental privacy. Therefore, it is important to identify early pathways for the governance of emerging technologies.

Were there any blockers bringing this data science competency into the UN, creating this team? What were some of the early challenges for you?

We’ve been fortunate in a number of ways. First of all, the nature of work at the United Nations is already aligned with social impact. Secondly, since we launched, we have had the support of leadership, including the UN Secretary-General.           

There was a lot of work that needed to be done creating those first data science teams: the terms of reference simply didn’t exist. We had to create all those terms of reference and job descriptions: data scientist, data engineer, data visualization specialist, etc. And when we first started to open those positions, we did not get a high volume or a diverse set of candidates.

Over a decade later, we’re in a more mature market: data scientist is a regular job in the United Nations and many other international and humanitarian organizations. Now it’s possible to get hundreds of applicants representing a broader base of diversity and backgrounds. For example, we saw positive change recently when our most recent call for Fellows drew 50% female candidates. Back when we started, it was a lot of work to create and fill these roles; we also had to lay the groundwork for their success which required navigating strategy, politics, and many other things beyond data science. Still, bureaucracy is not always easy and there is much more to do.

You’ve built an innovation network of excellence that can partner across the UN with different agencies, that bring in issues like refugee resettlement, poverty, or epidemiology. Was there any resistance to bringing along area specialists into the practice of using data and data science?  

One of the ways we addressed this challenge is by acting as a partner that can provide a safe space for innovation. At first, the requests were about showing our UN colleagues the potential of data science through proofs-of-concept, then we saw an increase in the appetite for more projects but also an understanding of the risks and limitations. Then we were asked to support during emerging crises. These days, requests are more holistic including anticipating possible futures, building digital innovations that are more inclusive, and looking to mobilize communities to address global challenges.

Our team worked to launch innovation labs around the world, in New York at the UN Headquarters, in Indonesia, Uganda, and Finland. Creating and nurturing those local ecosystems was important, and equally important was using these spaces to bring the data, tools, and expertise together with the different agencies. Like introducing any type of innovation, progress is not always straightforward or linear: you need to look for local champions, communities, and the enthusiastic early adopters, while also ensuring you have buy-in from leadership and the critical stakeholders.

And of course, it has not always worked. Failure is also part of the process and the journey of organizational change. In any context, it’s hard to change, even if you create the safest place for innovation. Not every organization is going to go there, and some may be reluctant to change. Sometimes it’s been slower than expected. More work needs to be done to scale the impact of the innovations, which is something the UN and many organizations like data.org are looking to accelerate.

But I think now we are in a better position, and the digital revolution is here. Data roles are everywhere, especially in the private sector. The next step will be when we have a new generation of senior managers and leaders that also have data backgrounds framed by strong ethics and values. We’ll pave the way for more effective leadership and change when we are routinely promoting people who bring a background in data science, on-the-ground fieldwork, and a deep commitment and understanding of policy-making, human rights, and humanitarian aid.

Part of the Pathways to Impact series

Curated conversations with data science for social impact leaders on their career journeys

See all Pathways to Impact

What other community of people or resources supports your work? Are there specific groups you engage with?  

When I look at my career trajectory, I’ve tried to have an impact on research and methodology, and then to have an impact on operations and operational impact. Last but not least is to have an ecosystem impact on communities.

I believe that science has a lot to offer, and I want to bring academics closer to social impact. Data science or AI — however you refer to it — is an enormous field. The strategy for me has been to engage with different sub-communities. For example, you have communities of practice around complex systems, deep learning, human-computer interaction, or computational social sciences, each with its own groups and conferences. I try to send signals to all those communities that they have something to offer, and maybe they don’t know it yet. We are also approaching communities that a priori might even look further away, like behavioral sciences or futures and foresight communities. There is great potential there.

My other communities are made up of development, humanitarian and policy experts working on key issues for the UN: people, this planet, its peace, and its prosperity. I can draw from communities of colleagues working for children, refugees, food security, gender equality, human rights, and health. And I’ve been really lucky to be exposed to the ins and outs of all these topics through these amazing experts.

What is clear is that to build an equitable future, across all these communities and ecosystems, we need to foster diversity and inclusion, and make sure that no voice is left behind.  

Your background in mathematics, cognitive science, biology, and data science all enable your contribution. But is there any non-data science skill set that has been beneficial? What are some of the skills that you’ve developed in the job, or as a leader, that has helped you? 

Empathy and learning to listen. Understanding where people are coming from, listening, and then translating between domains. I think it’s being able to translate between data science and human rights, between data science and food security, and between data science and policy making or diplomatic issues. These translational skills have been critical, because context matters.

When we work on a project, of course, part of building out the project is about the technology, but most of the time it is even more about people and cultures. Here’s an example of how context matters: when you do AI for satellite image analysis, imagine you are counting shelters. This typically is done by someone by hand. And it’s done by someone by hand because the count has to be perfect. Based on this count, someone will then assign a number, like X tons of food or whatever we are delivering. So essentially, when you do an AI algorithm, you have to focus on the most important thing. It’s not important to optimize the performance: we don’t care how good the algorithm is across all dimensions. We care that the algorithm will give a timely response when it is used by the human expert embedded in existing processes for emergency response.

Another example, we worked on a water delivery system project in a refugee camp in Jordan during the Syrian war, where we had to align stakeholders in the camp. There were many stakeholders, including organizations providing clean water services, truck drivers delivering the water, and of course the citizens in the camp. Here the first proposed solution did not work. The technology was not the most difficult part, so we focused on alignment. Once stakeholders were aligned, we made an alternative technical solution, and the project succeeded.

So really, context matters, not only the technology on its own. Being able to have empathy, translate, and understand what everybody is “really” saying, is critical for success. We always have to consider the human first, and then the technology. And then, you have to make sure that the tool works as it should!

What advice do you have for someone new to the field who’s interested in doing data for social impact work? That could be a student entering the workforce, or someone mid-career with great skills looking to apply them.  

The most important message is that we need all these people, the early stage and the mid-career and senior entrants!

We need people with hard science skills and social science skills; the future is at their intersection. We must convey to people that data for social impact is a real job. I think that is important to convey because sometimes that job and career have not been so obvious. You can build a career, but the impact you’ll make won’t be as easy to measure by financial metrics.

For someone just getting started, or wanting to begin, I would suggest seeking out a role that gives them a lot of exposure and lets them understand the context, to start to learn to translate. For someone mid-career, I would urge them to keep an open mind. I’ve seen many people enter the sector: astrophysicists, health care professionals, bankers, and intelligence analysts. Keeping an open mind and staying adaptable is key to success — along with being great at your work!

For many mid-career people, it’s also about learning to see the big picture. For example, if you come from a bank where the focus is on optimizing time series forecasting, the metrics for success are very clear and relatively narrow. In the social sector, there are more layers: human rights, environmental impact, and advocacy, Looking back, I think the people who have made the transition best are the ones that can adapt and adjust to seeing the multiple layers — and scales — in the big picture context.

What’s coming next? I know you don’t have a crystal ball, but what are your predictions for data science for social impact?

Beyond the obvious challenge, the climate emergency, I see big-picture public and private opportunities.

The first is a movement for broader creation of and support for digital public goods. I am hopeful that these public goods will continue to grow and be shared as a kind of digital infrastructure for humanity.

On the private sector side, I believe we will benefit from having more social entrepreneurs, like those in one network I support, Ashoka. If more social entrepreneurs manage to sustain successful businesses around social impact, that might change many things. For that, we need more investors in the intersection of technology — and deep tech — and social impact. This intersection does not yet exist, but I believe there is opportunity there. 

Beyond digital public goods and social entrepreneurship, I hope to see mainstream concepts around the responsible design of technology. We need to start asking upfront what we won’t do. Not just that, but also anticipate potential governance and regulatory pathways, especially for newer technologies. Where are the limits? Where are the red lines in the digital world? How do we design for good and also ensure that what we are putting out in the world does not have negative or unforeseen consequences after it has been released?

Finally, the last point on the future is about trust. Working with the WHO on infodemics, we have witnessed how mis- and disinformation can have a real impact on people’s health. Not just that, in politics, conflict, and climate — it is clear that we’re at an inflection point with objectivity and we need to find a way to restore trust in this post-factual world.

What’s your don’t miss daily or weekly read? What do you read to stay sane and happy and informed on a daily or weekly basis?

I listen to a lot of podcasts. Every week, I listen to Nature and Science magazine podcasts to understand what humanity is doing around science. One other I like is called Exponential View.

I also go down deep dives into specific topics. One week I might decide to go deep into the latest IPCC climate report and explore a bunch of different podcasts and articles. Another topic might be public speaking, so I listen to experts on how to make a persuasive and compelling talk. I enjoy deep dives into very different things and authors.

Beyond podcasts — and doing sport while I listen to them — I also experiment with concepts I learn about, like quantified self activities. For example, conducting ten antibody measurements in my blood over the last year has allowed me to model my immune response and show that every ~40 days, my antibodies for COVID divide by two. Just an “n” of 1. Small data gives me useful information and I find it fascinating!

About the Author

Series

Pathways to Impact

This data.org series interviews leaders in Data Science for Social Impact with a lens of how they got there, as well as the skills and experiences that have fueled their career progression.

See all Pathways to Impact

data.org In Your Inbox

Do you like this post?

Sign up for our newsletter and we’ll send you more content like this every month.

By submitting your information and clicking “Submit”, you agree to the data.org Privacy Policy and Terms and Conditions, and to receive email communications from data.org.