Findings from a landscape assessment of AI trends in education

• In Event: AI in Education: Evidence, Trends, and Market Insights

Proposal

While India has made significant progress in expanding access to primary education, a severe learning crisis persists. National and international assessments like NAS, ASER, and PISA consistently reveal a large gap between expected grade-level proficiency and actual student performance in public schools. Evidence increasingly suggests that computer-assisted personalized learning solutions could be key to addressing this issue. At the forefront of such Personalized Adaptive Learning (PAL) tools in developing countries is Mindspark, a Personalized Adaptive Learning (PAL) tool designed using insights from millions of data points collected through extensive assessments over several years.

Mindspark identifies individual student learning levels and misconceptions in Mathematics, their vernacular language, and English at a nuanced level through diagnostic tests. It then creates an individualized learning path for each student, providing targeted instruction to enable teaching at the right level at scale. It employs a gamified interface, using questions, activities and challenges to engage learners.

Mindspark program exemplifies innovative use of generative AI to address critical educational needs. It focuses on improving English reading, speaking, and writing skills, particularly in low-cost private and government schools where students have fallen behind. Using generative AI models, specifically GPT-4 and DALL-E 2, it creates three key modules: Story Builder, Feedback Program, and Speaking Program. These modules encourage creativity, provide constructive feedback, and offer immediate speech assessment, respectively.

Some principals that have guided the development of Mindspark throughout the process are:

1. Personalization: AI should adapt to individual learning styles, paces, and needs, creating tailored educational experiences.
2. Enhancement, Not Replacement: AI should serve as a tool to enhance educational outcomes without replacing core teaching functions.
3. Ethical Considerations: Ensuring privacy, data security, and fairness in AI-driven assessments is crucial.
4. Transparency: Clear communication with students and educators about AI usage is essential.
5. Human Integration: AI should complement, not replace, teachers, enhancing their capabilities.
6. Continuous Improvement: Ongoing evaluation and refinement of AI systems based on learning outcomes and user feedback ensure effectiveness.
7. Inclusivity: AI should support diverse learners, including those with disabilities or from different linguistic backgrounds.

This presentation will feature an introduction to Mindspark, demonstrating how it exemplifies innovative use of AI to address critical educational needs. We’ll discuss how its features align with good practices such as personalization, rapid iteration based on user feedback, and promoting inclusivity.

We'll also present data from Mindspark's implementation across 15 Indian states, including improvements in writing quality and student engagement. Attendees will participate in a hands-on demonstration of Mindspark's AI modules, experiencing firsthand how the system adapts to different learning needs.

The session will address critical questions about balancing AI assistance with human instruction, ensuring data privacy and security, and strategies for implementing similar programs in diverse educational contexts.

Author

• Charlie Habershon, Dalberg Advisors