February 11, 2025

Key takeaways from our first Gender, AI and MERL Working Group meeting

The Natural Language Processing Community of Practice (NLP-CoP) hosted our first Gender, AI and MERL Working Group on January 29. In this blog post, Gender, AI and MERL Working Group co-leads, Savita Bailur and Allison Sambo, are sharing some of the key takeaways from our gathering, emerging threads of discussion and a glimpse of what’s next for this working group.

Key Takeaways from our first meeting

While around 60% of the participants self-identified as “beginners in AI”, the depth and breadth of work being conducted by the group in Gender, AI and MERL was incredible – there’s clearly a great deal of work going on. We can be a powerful network for connection, support and growth in learning, capacities, and advocacy for a gender lens on AI and MERL.
Five broad themes emerged: 1) gendered impacts of AI, 2) ethics and accountability, 3) a feminist and decolonial lens, 4) AI skilling and gender-intentional AI user interfaces, 5) and finally, the intersection of AI, gender and climate.
Given the attendance of almost 300 participants from all around the world and our intention to move towards more participatory sessions, going forward we will meet quarterly as a whole group and invite others to help us co-lead additional meetings based on timezones to be sure everyone can participate fully.

What emerged from the meeting

A clear need for a group focusing on Gender, AI and MERL

The first meeting of the Gender, AI and MERL working group convened on January 29th, topping out at almost 300 participants, highlighting an exceptional interest in the convergence of these issues amongst researchers, evaluators, organizational leaders, information management professionals, and creatives.

The group included people working in national and international organizations/civil society organizations primarily, with a strong representation of independent consultants and representatives from universities, philanthropy, the private sector, and bilateral/multilateral institutions. We were thrilled to have global representation at the meeting, with more than 40% of participants joining from outside the US/Canada/Europe.

Our goal during the first meeting was to understand specific interests, needs, and opportunities among prospective Gender, AI and MERL Working Group members. We ran a few polls during the meeting to better understand who was in the room:

About 61% rated themselves as beginners, and about 60% also noted they are not currently working on gender-related AI projects.
Those working on Gender and AI projects said they are using AI as a tool in evaluation and research (32%) and evaluating AI as an approach (14%), with a subset building AI tools for evaluation or AI tools for programs (17%).
Overall, the group expressed broad interest and experience around ethics, tools, and methodologies; and, in particular, concerns about discrimination, bias, and safety.
Within the attendees, there was an incredible wealth of experiences and generosity in sharing these experiences with the group, including:
- Creating and deploying chatbots for youth and gender programming;
- Gender-based Violence prevention using LLMs;
- Ethics and governance frameworks, and making the economic case for safety;
- Utilizing AI for qualitative analysis.

Reproduction of poll results conducted during the meeting. We asked participants what they were more curious about when it came to Gender, AI and MERL and they expressed interest in Ethics (124), Evaluation methodologies (114), Tools (109), Use cases (95), and Building tools (59).

“AI in MERL and MERL in AI”

The diversity of experiences and interests shared in this first meeting led us to reflect on what we are working on at the intersection of Gender, AI and MERL. As Linda Raftree (the founder of MTI) put it, it’s about doing MERL on AI as well as using AI for MERL, and both through a gender lens. For safety reasons, this meeting was not recorded and we have not named individuals or organizations in this post.

We saw around five key themes appearing at the intersection of Gender, AI and MERL:

Gendered Impacts and Opportunities of AI: Several participants shared their work on examining how AI systems and algorithms disproportionately affect women and marginalized groups. This includes concerns about gender-based violence (GBV), gendered harassment online, and how AI can perpetuate or mitigate inequalities. Interesting projects that were mentioned included a couple researching how AI is used to silence LGBTQ+ voices and communities on social media. Participants were also exploring how chatbots can help survivors of gender-based violence, for example, AI-powered chatbots for GBV survivors. These efforts focus on both immediate support and long-term prevention.

“I am concerned about gender, civic space closures and potential impact of AI for democracy.”

Ethics and Accountability in AI: There was a strong interest in gender-intentional AI ethics, especially regarding its use in marginalized communities. Issues such as data privacy and responsibility were mentioned. It was noted that data scientists and developers could most benefit from such ethical frameworks for AI in the social sector.

“I am incredibly interested in the role of AI and privacy ethics when it comes to reproductive health access in the US”.

Feminist and Decolonial Perspectives on AI: Several contributors advocated for feminist and decolonial approaches to AI. This includes a call to critically examine how AI, as a product of colonial and patriarchal systems, may reinforce or exacerbate biases. Feminist groups who joined the WG meeting were especially interested in exploring and sharing these perspectives.

“I am researching ethics and governance of AI from a feminist and decolonial perspective.”

AI skilling for women: Both “supply” and “demand” side considerations were raised. There’s a growing focus on incorporating a gender lens into AI design and implementation. On the “supply” side, open-source tools, frameworks and toolkits were mentioned. On the “demand” side, AI curricula resources were shared. Both ensure that gender considerations are integral to technology creation and use. Another area of interest for researchers was the use of AI in qualitative data analysis, what biases that might include and how we can be better equipped to recognize this.

“We are working on an AI skills curriculum, including for low literacy women.”

“I would like to understand the use of AI in qualitative data analysis and any gender bias.”

AI and climate, climate and AI: While this was a smaller group of participants, some highlighted their work on AI data gathering to inform climate disaster, which could be critical in helping women, and equally, on the impact of climate change on women due to the water and energy consumed in the processing of data for AI and plans to build an ever greater number of large data centers to that end.

“I am working in the area of AI data gathering to inform climate disaster response – lots being developed in this area of course, with very little guidance for CSOs.”

What’s next for us?

We were blown away by the interest, as well as the breadth and depth of work being conducted by participants… did not seem like a group of beginners! We’d love to feature guest posts from the Group on gender, AI and MERL on the MERL Tech website. Get in touch with the Working Group leads if you have a topic you’d like to write about!
We plan to host a casual Gender, AI and MERL Working Group meet-up during the CSW in March: if you signed up to join the NLP-Cop and the working group, keep an eye out for invitations in your inbox. Relatedly, if you are involved in any CSW sessions related to Gender, AI and MERL, let us know so we can help spread the word!
Please get in touch with us if you’d like to volunteer to get involved in co-hosting a smaller (online) meeting. We want to be sure we’re available in various time zones and are looking for folks who might want to lead smaller virtual meetings. We can help with planning and logistics as needed.
Going forward we propose meeting quarterly as a whole group. We invite you to help us co-lead additional meetings based on timezones to be sure everyone can participate. If you’d like to volunteer to co-lead a sub-group in a specific timezone, let us know!

Don’t forget to join the NLP-CoP (on Slack and LinkedIn) and Gender, AI and MERL Working Group, where you can share your own gender, AI and MERL-related work, ideas, events and writing and learn about our next meetings and events.

For those reading this on 11 Feb: today the NLP-Cop is hosting an event focusing on Technocolonialism!