Dr. Rukmini Banerji is one of the world's most influential figures in grassroots education. Trained as an economist — a Rhodes Scholar at Oxford and a PhD from the University of Chicago — she chose classroom-level work over academia, returning to India in 1996 to join Pratham, one of the world's largest education NGOs, where she served for over a decade as CEO before recently stepping into the role of Senior Advisor.

In 2005 she launched the Annual Status of Education Report (ASER), the citizen-led household survey that has demonstrated, year after year, a hard truth: going to school is not the same as learning. She also pioneered Pratham's Teaching at the Right Level (TaRL) approach — grouping children by what they can actually do rather than their age — now used across India and adapted in parts of Africa and Asia. For this body of work she was awarded the 2021 Yidan Prize for Education Development, the world's highest education accolade.

Today she is bringing that same rigor to AI. Pratham has partnered with Anthropic to deploy a Claude-powered "Anytime Testing Machine" (ATM) — an assessment tool that grades handwritten student work and generates curriculum-aligned questions, with teachers reviewing AI feedback before it reaches learners. First adapted for over 5,000 women in Pratham's Second Chance program, it is now piloting across 20 schools with plans to reach 100 by the end of 2026.

In this episode, Rukmini talks with Svenia Busson about:

• Why she left economics for the classroom — and how "big ambition and not much money" shaped Pratham's community-driven model
• The origin of Teaching at the Right Level — teach the child where they are, not where the curriculum says they should be
• What ASER has revealed every year since 2005 — that enrollment and learning are two very different things
• Pratham's partnership with Anthropic and the Claude-powered "Anytime Testing Machine" — and why human oversight is non-negotiable
• The gap between what edtech companies want to sell and what the ground actually needs — "technology rains on you, then you're expected to use it"
• Assessing learning in the age of AI — why the process (the questions a student asks) matters more than the final essay
• Why middle school (ages 10–14) is education's next great frontier — and why the 15-year-old of 2031 must look different from the one of 2026

Mentioned in this episode:

Education Futures Fellowship

A six-week programme starting on September 22nd, where seven world-leading experts in developmental psychology, neuroscience, and AI child safety will share everything they know with you: educationfutures.ai/fellowship.