Loading Events

Speakers: Manal Jalloul and Nadine Weheba

Moderator: Nagla Rizk

Date: February 8th, 2023

The Access to Knowledge for Development Center (A2K4D) hosted its second webinar as the Middle East and North Africa (MENA) hub of the F<a+i>r Hubs Network titled “Algorithmic Decision Making (ADM), Feminism and Inclusion in MENA: Unpacking Concepts and Call for proposals” on Wednesday February 8th via Zoom. Joining the webinar as a speaker was Dr. Manal Jalloul (Founder of AI Lab and NVIDIA Deep Learning Institute, Certified Instructor and University Ambassador at the American University of Beirut). Dr. Jalloul unpacked concepts of ADM and discussed how it can be harnessed for gender inclusion in MENA. We also discussed the project’s latest call for proposals, titled: “Incubating Feminist AI: Call for Expressions of Interest Algorithmic Decision Making Systems”. The call focuses on harnessing ADM for more equitable outcomes and models deploying more effective and inclusive social protection in MENA.


Key takeaways:

    1. It all goes back to data. Due to historical norms and gender discrimination, there is bias in the data available that is used to create and train ADM systems. Data needs to be more inclusive and representative for ADM systems to work fairly.
    2. Bias exists in the design of the algorithms themselves; algorithms are designed by people with gender-biases. The teams that develop these solutions need to be diverse in order to prevent human biases from impacting the technology.
    3. ADMs have the potential to help meet SDGs such as equal health, gender equality and ending domestic violence.
    4. AI can be used to identify inequalities in data and ADM systems. AI and ADM systems must be tested prior to deployment and before using them on a wider scale. We can use AI to test new technologies to ensure that there are no biases of treatment.
    5. AI policy-making in MENA needs to be data driven in order for policies to inform our unique needs as a region. AI can be used to extract insights from data such as correlations, gaps, inequalities and areas that need attention. We can also use AI to gather data on the impact of policies implemented and make modifications accordingly.
    6. Once datasets are diverse, they can then include the scope of women who are usually outside of national statistics and data sets such as those in the informal economy whose data cannot be found in national statistics.


Link: https://www.youtube.com/watch?v=4k_f5BDm_dg