⚡ Quick Summary
Published by UNESCO under its Women for Ethical AI (W4EAI) platform, this outlook study examines how gender equality is treated through and in artificial intelligence. It builds on the gender chapter (Policy Area 6) of the UNESCO Recommendation on the Ethics of Artificial Intelligence, described in the prologue as a commitment adopted by 194 Member States, and was prepared for discussion at the W4EAI Conference at UNESCO headquarters in Paris.
The study has four sections. Section 1 sets out the motivation and the W4EAI platform's objectives. Section 2 documents the participation gap — women hold 12% of AI research positions, 30% of AI professional roles and 18% of C-suite positions at AI startups — and finds that of the 138 countries assessed by the Global Index on Responsible AI, only 24 have frameworks mentioning gender. It then identifies risks and case studies in healthcare, employment, generative AI, surveillance and online gender-based violence. Section 3 examines ethics-by-design and feminist, gender-transformative approaches alongside the UNESCO Ethical Impact Assessment and Readiness Assessment Methodology. Section 4 presents nine recommendations, each linked to specific provisions of the Recommendation.
🧩 What’s Covered
- Introduction (Section 1): the scale of AI investment (almost $100 billion in global private investment in 2023, $25.2 billion of it in generative AI), UNESCO's definition of an AI system, the seven gender points of Policy Area 6, and the objectives of the W4EAI platform.
- Participation and the gender data gap (Section 2.1): access to AI information and tools, education and training, workforce and leadership, and other barriers such as unpaid care work, with tables of gender-gap statistics and the finding of up to 17% variance between reporting authorities.
- Policy and resources (Section 2.2): the AI policy landscape, the Latin American strategy review that produced five minimum standards, an analysis of 20 GIRAI-identified government frameworks, and an eight-component gender policy agenda for AI.
- Risks and case studies (Sections 2.3–2.4): MIT's AI Risk Repository (27 gender-related entries of 702), Berkeley Haas examples of algorithmic bias, and cases in healthcare, hiring, generative AI stereotypes, facial recognition and surveillance, and online and offline gender-based violence.
- Approaches (Section 3.1): ethics by design across problem definition, data collection, model development, deployment and evaluation; feminist and gender-transformative approaches using a twin-track model of targeted action and mainstreaming; and Box 9 examples such as Incubating Feminist AI and Spain's RADIA programme.
- Tools (Section 3.2): AI impact assessments and audits, the UNESCO Ethical Impact Assessment's scoping and implementation questions, HUDERIA, and the RAM's gender questions across legal, social/cultural and infrastructure dimensions.
- Recommendations (Section 4): nine recommendations with sub-points and the corresponding paragraphs of the UNESCO Recommendation on the Ethics of AI.
💡 Why it matters?
Teams that must show how AI affects women and girls get a consolidated evidence base here: gender-gap statistics, documented bias cases and a repeatable policy analysis method. The audit of the policy landscape shows where gender appears only in appendices or is left undefined, which helps reviewers judge whether a national AI strategy contains measurable gender commitments. The treatment of the UNESCO Ethical Impact Assessment and the Readiness Assessment Methodology connects upstream governance to practical review questions, and the recommendations specify targets, budget lines and reporting duties that policy and compliance functions can adapt into checklists.
❓ What’s Missing
The authors acknowledge that the policy search is not exhaustive, that sources may be out of date, that three identified documents had broken URLs and that the search terms were not exhaustive. Several quantitative gaps the report itself identifies remain: gender-disaggregated AI data are inconsistent, drawn heavily from LinkedIn and Global North sources, and there is no agreed definition of STEM or of AI-relevant fields. The study offers no costings, no model instrument text and no enforcement detail, and its analysis is framed largely around women rather than the broader gender spectrum it advocates.
👥 Best For
Policy advisers drafting or reviewing national AI strategies, gender equality officers in ministries and regulators, and AI ethics researchers needing documented gender-bias cases and critiques of measurement. It also serves audit and compliance teams designing impact assessments or assessing whether existing evaluation tools capture gender-specific harm.
📄 Source Details
Outlook Study on Artificial Intelligence and Gender, published by UNESCO under its Women for Ethical AI platform, developed under the leadership and overall supervision of Gabriela Ramos with contributions from Auxane Boch, Gökce Cobansoy Hizel, Caitlin C. Corrigan, Cecilia Danesi, Sarah Elaine Eaton, Caitlin Kraft-Buchman, Oriana Kraft, Alexander Kriebitz, Eleonora Lamm, Cristina Martínez Pinto, Wanda Muñoz, Paola Ricaurte, Alessandra Sala and Sumiko Shimo. Pre-printed for discussion at the W4EAI Conference, 30 October 2024, UNESCO HQ, Paris; reference SHS/REI/EAI/W4EAI/2024/Outlook. 59 pages, in English; text extraction of all 59 pages was available. No URL for the document itself is printed.
