We plan to examine data from our recent EEF randomised control trial in the UK involving over 5,000 students in 63 primary schools to determine whether  learning outcomes, engagement and retention differed significantly by sex.

Maths-Whizz caters to the needs of marginalised or disadvantaged child groups by being adaptive, inclusive, and non-discriminative. As an AI-led virtual tutor, it personalises instruction solely on the basis of each learner’s ability and progress, not background or circumstance. This ensures every child receives tailored support to close gaps or extend their learning, regardless of gender, ethnicity, or socio-economic status. Features such as voiced-over lessons and tests, scaffolded help, and gamified activities further support, for example to learners with low literacy, or those at risk of disengagement. Teachers benefit from real-time reporting dashboards that highlight individual needs, including those from disadvantaged groups, enabling timely intervention.

At scale, Maths-Whizz has proven impact in marginalised contexts; for example, the iMlango project in Kenya reached over 100,000 students, many of them marginalised girls. Similar localised projects in the DRC and Mexico have adapted the product to cultural and linguistic contexts, ensuring accessibility and relevance. By being non-discriminative, adaptive, and globally adaptable, Maths-Whizz helps level the playing field so that all children can benefit from high-quality maths education.

• Non-Discriminative by Design: The AI-led Tutor adapts to a learner’s ability and progress, not to demographic factors such as gender, ethnicity, or socio-economic status. This ensures equitable learning opportunities for all.
• Accessibility & Universal Design: Features such as voiced-over lessons/tests, clear visual cues, scaffolded support, and gamified engagement align with WCAG 2.1 principles, making the product accessible for students.
• Cultural & Contextual Relevance: Content is reviewed with local experts when entering new regions, ensuring alignment to national curricula and adaptation of imagery, voiceovers, and examples to reflect local cultural values.
• Teacher Empowerment: Teachers receive real-time data dashboards to identify and address gaps, ensuring they remain central to the learning process. The tool augments rather than replaces teacher expertise, supporting professional capacity.
• Inclusive User Interface: The UI is intuitive, child-friendly, and designed for learners with diverse abilities. Features such as numbered questions help children who need clearer task boundaries, e.g., those with ASD.
• Diverse & Collaborative Development: Development involves international education experts, local partners, and classroom feedback, ensuring diverse perspectives shape the product. Collaboration with international partners, and researchers helps to ensure disadvantaged groups are represented.

The Maths-Whizz Tutor promotes AI equity through its rules-based design, which makes learning pathways transparent, consistent, and free from hidden algorithmic bias. By continually reviewing content, translations, and pedagogy with educators and experts, we ensure that every child, regardless of background, benefits fairly. This commitment to equity and accountability underpins our mission to deliver inclusive, trusted AI-driven education worldwide.