AI for Gender Equality and Inclusion Innovation Research Network

Machine learning technologies like Automatic Speech Recognition systems (ASR), language models, machine translation systems are developed to work in environments that have gender and other forms of biases and discrimination present. These biases are present in the datasets that are used for model training and the choices around the ML models. The developed ML models end up reinforcing biases that are embedded in training data. This research will focus on a data-driven research approach and carry out a scoping study to provide a deeper understanding of gender biases and inclusivity of AI tools on the African continent. The results will be used to build a framework and guidelines for mitigation of gender biases in ASR systems through equity in data collection and model creation. Finally, this work will develop a model governance framework to address gender and ethical aspects around the data and algorithms from which ASR systems are built. The main objective of this research is to focus on a data-driven approach to understand gender biases in building Artificial Intelligence models in the African context.

This research aims to explore gender inclusion in AI across the countries identified within the East African community that is Anglophone east africa(Tanzania,kenya and Uganda) and Francophone East Africa(Rwanda,Burundi and DRC). The research project will examine power in the AI ecosystem from a gender perspective,study on barriers to creation and usage of gender disaggregated data for AI,the gender gaps and biases in AI and explorer how power in this instance can be challenged by studying existing AI enabled systems and usage across Anglophone and Francophone east africa.This will include the structural and systematic barriers to promoting gender equality and inclusion in the AI sector and Testing approaches to addressing the challenges posed by AI in gender. The main objective of the research project is to examine power in the AI ecosystem from a gender perspective, study on barriers to creation and usage of gender disaggregated data for AI, the gender gaps and biases in AI and explore how power in this instance can be challenged to enable a gender.

Artificial intelligence (AI) technologies are being explored in Africa for increasing food security and improving the efficiency of agri-food systems value chains. In an environment where distinct gender inequality dynamics are already evident, women form the largest proportion of agri-food systems actors in the informal sector, particularly in the production, value addition, and marketing of food. While digitization and the digitalization of agriculture may have socioeconomic benefits for women, there is a likelihood these women will be excluded from the deployment of AI technology. This poses a risk of further widening a gender gap, preventing them from equally benefiting from the digital transformation. This study investigates how inclusion of women in Kenya’s informal economy contributes to inclusive digitization. The project will identify and emphasize the obstacles to women’s digital inclusion and consider ways to articulate women’s perspectives in the development and application of agricultural AI systems, as well as associated policy processes. The goal of this project is to run three distinct but related scoping studies to investigate the status of women’s inclusion in the design, adoption, and skilling of AI in the agricultural sector of Kenya. This will be with an aim of informing interventions that could improve the inclusion of women in agricultural informal spaces, thus enhancing their ability to tap into the benefits that come with AI technologies and their deployment in agriculture in Kenya.

Food waste is a systemic problem that happens along every step of the food supply chain, from farm and field to fork. This not only poses a problem of entrenching food insecurity, poverty, and inequality in South Africa (SA) but also exacerbates climate change, pollution and biodiversity loss. There are proposals that deployment of data-driven systems such as artificial intelligence (AI) can boost resource efficiency (RE) and support the transition to systems that minimize food waste at different points in the food supply chain. But most of these proposals are only sociotechnical imaginaries from the global North and fail to elaborate the risks and harms of deploying data-driven technologies in the global South and/or in existing inequitable ecosystems that discriminate against women. By actively incorporating an intersectional gender inclusion lens, this case study will explore AI use to reduce food waste at different points of SA food supply chains. This project calls for the development of better evidence on AI deployment, gender, and waste in food systems as a necessary first step in the path towards gender-sensitive policy making and overall gender equality to resolve the triple challenges associated with food waste

Aquaculture plays a very indispensable role in sustaining the world population at large ranging from its various levels of production to significant contributions to the economy. However, there exists a wide gap between gender compositions in the Aquaculture value chain, and the involvement of women is more evident in post-harvest activities, which are generally lower-profit activities. While men tend to dominate in fishing activities, which tend to be higher profit roles. The adoption of AI techniques in Aquaculture will ensure equity, diversity, and gender inclusion while minimizing unintended bias and providing abundant opportunities for women is essential. Thus, the project will focus on the assessment and analysis of AI techniques applications in aquaculture systems, and critical gender analysis in the use of AI in aquaculture systems, through extensive data collection, scoping studies, gap analysis, and communication with stakeholders and policymakers to develop innovative recommendations. The general objective of this research is to review the application of artificial intelligence techniques and gender inclusiveness in aquaculture.

Despite current gender educational regulations and other interventions aimed at ensuring gender equity, equality and inclusion, it still remains a major issue in Kenya. STEM education is intended to equip students with the knowledge, skills, attitudes, and behaviors required to thrive in an inclusive and sustainable society. A recent study by UNESCO shows that only 35% of all students enrolled in STEM related subjects at higher learning institutions are female students. This study intends to examine the participation of female students’ in STEM courses by investigating how gender inequality affects student placement and students’ perceptions of AI in STEM programs and eventual work placement. In addition, the study will investigate the ethical purpose and societal benefit of AI application use. The general objective of this research project is to determine STEM gender inequality using a Machine Learning Model in tertiary learning institutions and industry.

Against the background of the worsening change in climate predisposing African farmers to adverse effects of the vagaries of the weather occasioned by their heavy dependence on rainfed agriculture, the study seeks to propose an AI driven solution to addressing the menace, especially as it affects the women farmers who play prominent roles of production and reproduction in the rural households of sub-Saharan Africa. Specifically, the study seeks to espouse existing AI driven solutions to climate related problems, especially, erratic weather conditions faced by women farmers and make data driven policy recommendations on how women farmers can be empowered to best utilize appropriate and responsible AI driven solutions. The project will be of immense benefit to empower women farmers as renowned drivers of rural household socio-economic status and contribute towards attainment of SDG goals promoting gender equality, no poverty and zero hunger (SDGs 5, 1 and 2, respectively). The general objective of the project is to employ/deploy AI driven solutions for climate smart farming as a pathway towards economic empowerment of women farmers in West Africa.

Artificial intelligence has greatly transformed the way businesses operate across all industries, and natural language processing plays a crucial role in representing identity, observations, and ideas from high-level theories to the machine’s everyday functions. Many successful digital organizations have achieved their success through the collection of data in written, verbal, and visual forms. However, most African local languages are under-resourced, which has hindered Africans from taking advantage of AI opportunities. To ensure Africa’s participation and benefits in the global AI ecosystem, decisive action is needed from policymakers, human resource development, private sector involvement, and fostering innovation. This study aims to examine the current state of African local language representation in the global AI ecosystem, identify challenges, and explore potential opportunities for AI in Africa. The general objective of the study is to conduct a Systematic Literature Review (SLR) to identify the challenges, opportunities, and possible interventions for the development and application of AI for African local languages.

AI systems copy and reinforce existing social biases, a problem now widely recognized and studied. But current research on gender bias in natural language processing (NLP) definitely points to a solution. Gender biases in AI applications such as machine translation (MT) are common and should be carefully considered when developing such applications. In this project, we propose to develop an MT prototype capable of detecting and mitigating gender bias. Since the target languages Amharic, Ge’ez and Awign fall into the category of resource-limited languages, we will use transfer learning to take advantage of available resource-rich languages such as English. The project is relevant and current in addressing the challenges of the language barrier between different local communities as well as the issue of gender bias in MT datasets. Successful implementation of the project could have the potential to address these challenges inside and outside Ethiopia, particularly in sub-Saharan Africa.