AI, Learning and Equity

Group Submission Type: Formal Panel Session

Proposal

Artificial intelligence is transforming education globally, yet it often bypasses marginalized learners—the very students who need support most. This symposium explores how artificial intelligence (AI) is reshaping education systems across diverse contexts, with particular attention to equity, inclusion, and teacher agency. The four papers span rigorous field trials, global policy frameworks, inclusive design initiatives, and regional evidence, offering a comprehensive view of how AI can both advance and challenge education for all. The discussion will highlight frameworks for creating AI with educators, grounding solutions in evidence, and embedding practices in policies that promote learning equity.

The first paper - From Chalkboards to Chatbots: Evaluating the Impact of Generative AI on Learning Outcomes in Nigeria” - presents results from one of the first randomized controlled trials of generative AI in African classrooms. Conducted in Nigeria, the study deployed Microsoft Copilot (GPT-4) as a virtual tutor for senior secondary students in English. Over six weeks, students receiving AI support improved by 0.31 standard deviations overall and 0.23 in English—the primary outcome—equivalent to 1.5–2 years of typical learning. Cost-effectiveness analysis positioned the intervention among the most impactful globally. The largest benefits accrued to female students and those with higher initial performance, highlighting both the potential of AI tutoring and the importance of monitoring distributional effects. The findings, presented by a senior researcher of the World Bank demonstrate how generative AI can provide transformative support in low-resource contexts while raising critical questions about sustainability and learning equity.

The second paper is entitled: UNESCO’s AI Competency Framework: Implications for Teachers, Learning and Equity.” Presented by a senior UNESCO expert, this paper first describes the AI competency framework responds to the urgent need for teacher-centered guidance in navigating AI. The paper emphasizes that teachers must remain in control of technology use, empowered to critically assess when AI tools add value and when they risk undermining pedagogy, ethics, or equity. Risks include erosion of teacher and student agency, privacy breaches, surveillance, bias, and disinformation. The framework proposes curriculum goals and activities for embedding AI competencies into teacher training, equipping educators with both technical skills and ethical awareness. It calls on governments to update ICT and education policies with AI-specific principles, ensuring regulation is trustworthy, safe, and human-centered. The central message is that AI should empower—not displace—teachers, who remain the agents of equity and resilience in education systems.

In the third paper, a senior UNICEF expert will present on the following topic “The Beyond Digital as Usual: Inclusive EdTech and AI for Marginalized Learners” This paper highlights UNICEF’s global initiatives to ensure AI benefits extend to marginalized learners. The EdTech for Good Framework, co-developed with governments and industry partners, offers a comprehensive structure to evaluate whether AI-powered tools are effective, safe, and equitable. Complementing this evaluative approach, the Accessible Digital Textbooks initiative embeds inclusive AI features such as text-to-speech, dyslexia-friendly formats, offline access, and national language availability. Together, these efforts illustrate a dual strategy: build trust in AI through global evaluation principles, while ensuring practical, accessible tools dismantle barriers for learners with disabilities and those in low-connectivity contexts. The paper argues that inclusion must be embedded in both design and policy to prevent AI from reinforcing divides.

The fourth paper entitled “The Use of AI in Support of Learning Equity: Evidence from Latin America” situates AI adoption within long-standing struggles for educational equity in Latin America, in line with the Universal Declaration of Human Rights. Drawing on the author’s work on a 2025 regional study with UNESCO-IIEP Buenos Aires, it examines how AI and AI-adjacent approaches are being deployed to address persistent inequities linked to poverty and sociocultural exclusion. Case evidence shows AI being used for adaptive learning, resource targeting, and support for marginalized groups. The paper stresses that while AI offers promise, its benefits must be guided by equity-centered research and inclusive policy frameworks. Lessons from Latin America carry broader implications for how AI can advance peace, human development, and fairness in education globally.

A senior official from the UK FCDO will serve as Discussant for this panel.


Together, the four papers demonstrate AI’s transformative potential when anchored in ethics, inclusion, and teacher empowerment. From Nigeria’s classroom evidence to UNESCO’s global framework, UNICEF’s inclusive initiatives, and Latin America’s regional perspective, the symposium underscores a shared imperative: AI in education must be designed and governed to advance learning and equity for all.

Sub Unit

• General Pool

Chair

• Daniel A Wagner, University of Pennsylvania

Individual Presentations

• Beyond Digital as Usual: Inclusive EdTech and AI for Marginalized Learners - Frank van Cappelle, UNICEF

• The use of AI in Support of Learning Equity: Evidence from Latin America - Nathan M. Castillo, University of Illinois

• From Chalkboards to Chatbots : Evaluating the Impact of Generative AI on Learning Outcomes in Nigeria - Martin De Simone, The World Bank Group

Discussant

• Ben Gomes, Google