🎙️ Q&A with Eleonore Fournier-Tombs

Eleonore Fournier-Tombs, Head of Anticipatory Action and Innovation at UNU Centre for Policy Research, is a leading voice at the intersection of AI governance, gender equality, and sustainable development. In this exclusive interview for the Ethical Tech Digest, she discusses the pressing challenges and opportunities in ensuring Artificial Intelligence (AI) benefits humanity equitably while addressing crucial issues like gender bias, climate change, and democratic participation. 

Sudeshna Mukherjee (SM): What are the most pressing challenges in developing global policies around AI that are both effective and equitable?

Eleonore Fournier-Tombs (EFT): Addressing global inequality is one of the biggest challenges faced by humanity. At the UN, we focus on ensuring countries around the world can equally participate in and benefit from AI development. There's a risk that AI could accentuate power concentration in certain countries and companies.

We need to develop governance mechanisms that protect human rights for everyone. Recognizing and supporting local AI ecosystems in Global South countries like India, Indonesia, South Africa, and Brazil is important. These ecosystems often involve various stakeholders including local companies, foreign tech giants like Google or Microsoft, national governments, civil society organizations, and universities offering degrees in artificial intelligence.

For example, in Southeast Asia, Jakarta has emerged as a major AI development hub. Companies like GoJek, a billion-dollar ride-sharing platform, exemplify the innovation happening in these regions. The ecosystem thrives through collaboration between local startups, multinational tech companies, government support, civil society organizations, and academic institutions and universities offering specialized degrees in AI.

SM: How can AI and data science be leveraged to address complex global challenges, particularly in regions vulnerable to historic injustice, disasters, climate change, and humanitarian crises?

AI and data science can be applied in three key areas related to climate change: prevention, mitigation, and adaptation. For prevention, AI tools can help with energy optimization and a better understanding of how to reduce consumption and carbon footprints. This is especially important as the climate becomes warmer and people use more energy for cooling.

For mitigation, we focus on supporting and understanding the effects on populations that are more vulnerable to climate impacts, whether they're located in coastal areas, small islands, or areas susceptible to drought. In agriculture, there's significant interest in using AI for crop monitoring and distribution of water effectively.

For adaptation, AI can help monitor and better evaluate how different groups might have to migrate and how to build safer buildings as climate disasters increase. Additionally, for humanitarian interventions, predictive modeling is crucial for planning aid and preparing for crises.

SM: In another topic of concern, how do you envision machine learning technology contributing to the improvement of deliberative democracy in the digital age?

EFT: I developed a machine learning method to measure the quality of political deliberations online. The goal is to compare different venues for deliberation and identify characteristics that lead to more inclusive discussions. These include freedom to participate, a respectful environment, and constructive exchanges of ideas rather than positional politics.

This technology can be applied to various contexts, from shaping global objectives like the UN's Sustainable Development Goals to facilitating local political discussions. It's about creating inclusive spaces, both online and offline, where people can contribute to policy-making and discuss emerging crises and events.

While no country has a perfect solution, we are learning through experimentation. Switzerland, for example, is committed to regular referendums. France has experimented with citizen forums on issues like climate change. I’m also interested in indigenous modalities of deliberation, such as those practiced in Nunavut, Canada.

SM: In your book "Gender Reboot," you explore the impact of AI on gender norms and women's rights. How can AI be harnessed to promote gender equality?

EFT: In the book, I discuss three main risks of AI for gender equality: discrimination, where AI tools sometimes produce biased outputs, like rejecting female job candidates based on gender; stereotyping, where generative AI often reinforces traditional gender norms or sexual stereotypes; and exclusion, where despite women's significant contributions to AI development historically, there remains a lack of recognition and support for women in the field today.

Even though women have been quite involved in the history of artificial intelligence development, with numerous women pioneers in programming and coding throughout the years, and the first computer programmers were women, there's been a certain amount of exclusion of women in the development of AI and a lack of recognition for women's advances. For example, a New York Times article published last year named the 10 leading figures in artificial intelligence, and all of them were men.

To promote gender equality, it's crucial to support women-owned AI enterprises, which often receive less venture capital funding than those owned by men. We must encourage innovations that address issues unique to women, such as safety features in ride-sharing applications or AI applications in reproductive health. There's also a need for strong oversight of generative AI tools to prevent the propagation of harmful stereotypes or disinformation that could negatively impact women's rights.

SM: What advice do you have for aspiring professionals looking to work at the intersection of AI and the Sustainable Development Goals?

EFT: I recommend developing a skill set with a T shape - having downward expertise in data science or AI, which is helpful, and then trying to understand sustainable development and global objectives more broadly. This approach allows you to be flexible while still having critical technical skills that are in demand in international organizations.

Think about the data science toolkit: data collection and storage, data retrieval, modeling perspective (understanding machine learning, deep learning, or Natural Language Processing models) and analysis, and visualization communication. Having that toolkit relatively solid is a huge asset for doing predictions and text modeling. It also allows you to speak with more authority in ethics or policy because you understand how the tools are developed.

While AI is rapidly evolving, I don't think the foundational skills will become obsolete. Even though techniques from a few years ago would need to be upgraded, having a foundation in data science is important. Sometimes simple methods are more accurate - it depends on the use case.

Follow Eleonore's work on LinkedIn.

Edited by Estelle Ciesla and Ava Sazanami