Addressing risks and opportunities of AI from a gender perspective

“AI should be made much safer through policies and regulations, and specific types of AI, such as autonomous weapons systems, should be forbidden.”

Eleonore Fournier-Tombs

A medical doctor, a foreign correspondent, a dancer. Eleonore Fournier-Tombs wanted to be many things when she was a child, but, by age 13, she was already attending Model UN conferences and aiming at the path she eventually followed: working at the United Nations.

“In a way, now, being a researcher at the UN, I have the opportunity to work for the organisation but also be creative and conduct my own research, which is wonderful”, the data scientist tells us.

Eleonore Fournier-Tombs is a global affairs researcher, with a specialisation in technology, gender, and international organisations. She recently worked as a Senior Researcher at UNU Macau, where she led research on gender and Artificial Intelligence (AI).

When it comes to AI, she explains, “there are both risks and opportunities from a gender perspective”. Among the risks are discrimination, “when an AI system has a different, erroneous output, based only on gender”, and stereotyping, “when an AI system propagates the ideas about women’s sexualisation and inferiority”. Other risks, she adds, are exclusion, “when the use or development of an AI system is restricted to women”, and insecurity, “when an AI system can cause safety issues for women”.

However, she notes, if these risks are addressed, “AI can provide important socioeconomic opportunities for women, such as supporting their businesses, providing solutions to women’s health issues, and reducing gender-based violence and trafficking”.

One way of addressing the risks, she says, is “combining different government policies and regulations, along with technical standards that ensure that technologies are safe for women before they are deployed”.


Gender sensitive AI Policy in Southeast Asia


One of the research projects Eleonore Fournier-Tombs led at UNU Macau was “Enhancing The Development of Standards and Frameworks for Critical Technologies in Southeast Asia”, for the International Telecommunication Union (ITU) Regional Office for Asia and the Pacific.

In this project, she and her team investigated gaps and opportunities in gender and other societal biases of AI in Thailand, Malaysia, Indonesia and the Philippines.

The project provided the policy report  “Gender Sensitive Artificial Intelligence (AI) Policy in Southeast Asia ”, as well as “a series of trainings to policymakers in the region, with the objective of informing the development of policies, standards and networks that would mitigate gender risks in these technologies and increase their opportunities”, she explains.

“We found that policymakers in these countries were very receptive to the notion of gender-safe AI, but needed some clear guidelines as to how to think about them and address them in AI policy”, she says. “Generally, gender-sensitive AI policy is not extensively addressed globally, so Southeast Asian countries doing this would be pioneering in the domain.”

Eleonore Fournier-Tombs says that, in this research project, she learned a lot “about the intersectionality of gender in these countries, such as the difference in experiences for rural women, indigenous women, and elderly women”.

There were also different considerations from a religious perspective, she adds. “For example, Muslim women in Indonesia and Malaysia may have different realities and different uses of AI than women of different religions. There are also many migrants coming to and from these countries, and AI in a context of migration is a rising area of concern.”

The research team was also composed of JeongHyun Lee and Arthit Suriyawongkul, UNU Macau; Preeti Raghunath, University of Monash in Malaysia; Matthew Dailey, Joyee S. Chatterjee, Philippe Doneys, Wanchanok Sunthorn and Sirayuth Thongprasert, Asia Institute of Technology (AIT); Kris Villanueva, Government of the Philippines; and Febroza Belda, Indonesia.

While in UNU Macau, she also led the research project on “Gendered implications of Artificial Intelligence on the implementation of the Women, Peace and Security Agenda in Southeast Asia”.

But AI technologies pose other challenges as well, and most of the issues on gender are observed with any group that has traditionally been marginalised in society, the researcher points out. “So, we see similar issues with people of colour, linguistic minorities, indigenous peoples, and so on.”

Also, she adds, “because AI technologies are used so broadly across all sectors, they can have a number of adverse effects, including on the environment, as many models require specialised hardware and extensive energy resources to run”.

But the researcher notes that just because there are many risks in using AI, it doesn’t mean that we should never use it. “Rather, it means that it should be made much safer through policies and regulations, and that specific types of AI, such as autonomous weapons systems, should be forbidden.”


Deliberative democracy


Eleonore Fournier-Tombs is also interested in research on deliberative democracy. During her PhD at the University of Geneva and postdoc at McGill, she developed a tool called DelibAnalysis, which helps researchers predict the quality of political deliberations.

“It was developed in order to have better comparative metrics for deliberative venues, to support better deliberations”, she explains. “For example, it was used originally to compare online and offline venues for deliberation, to think about specific features of these venues and how helpful they were in consensus-building.”

The tool was then applied “to compare indigenous and partisan parliaments in the Canadian North, to understand what features of each were conducive to good decisions from parliamentarians”, she says. Now, she explains, “it is used around the world by different researchers” and she plans “to redevelop it in the next few years to make it easier to use and upgrade the technology”.

Eleonore Fournier-Tombs has also worked as a data scientist at the Centre for Humanitarian Data (OCHA). She started working there at the beginning of the pandemic “to develop models that would predict the spread and the severity of COVID-19 in countries that were already experiencing humanitarian crisis, such as South Sudan, Sudan, Haiti, Yemen and Central African Republic”.

She was then asked to contribute to similar work at the World Bank, “to look at the compound effect of natural disasters on COVID-19”. “We studied Jamaica, Kenya, and Indonesia”, she says.

“Overall, it was very interesting for me to see the impact of these events on the virus spread, and to be able to add features indicating vulnerability to COVID-19 as they varied in different countries and contexts”, she notes.
This work also led to her participation in a grant from the International Development Research Centre (IDRC), received at the University of Ottawa, which aimed to localise the OCHA model to Senegal and Mali.

Today, Eleonore Fournier-Tombs is Head of Anticipatory Action and Innovation at the UNU Centre for Policy Research (UNU-CPR), in New York.

The teenager who took part in many UN mock conferences has grown up to be a researcher who enjoys “digging deep into an issue and providing conceptual and methodological framework for people to think about complex topics”. And she enjoys doing this “from a global perspective”, so she tends to research “issues of interest to the United Nations and member states”.

“It’s also very important to me to contribute to technical empowerment of people, so I like to do multidisciplinary work where social scientists or policymakers will gain additional technical skills and experience”, she says.

A mother of two young children, there is one thing she cannot work without: “childcare”. She recalls “this really was a big issue during COVID-19”, when she had a five-year-old and a nine-month-old at home and was “trying to concentrate to do modelling and research” – “I did a lot of work at night”. “Now, I really appreciate each day that my children are fine doing their activities, so that I can write.” The science world, we guess, appreciates it too.