Jean-Claude Brizard's Recent LinkedIn Posts

Two new randomized controlled trials — one from Stanford University, one from the World Bank Group: Digital & AI — arrived within weeks of each other and together tell a story that every funder, policymaker, and district leader investing in #AI for education should read carefully. The World Bank's From Chalkboards to Chatbots study tested Microsoft Copilot as an English language tutor for secondary students in Nigeria over six weeks, with teacher guidance and deliberate design. The results showed a 0.31 standard deviation improvement on a comprehensive assessment, with learning gains equivalent to 1.5 to 2 years of schooling — placing it among the most cost-effective programs to improve learning outcomes documented in the research. Critically, the largest gains went to female students, and the findings held across the baseline ability distribution — evidence that when AI tutoring is designed and used properly, it can be transformative even in low-resource settings. Then Stanford's National Student Support Accelerator released a complementary study — this one showing what happens without those conditions. In two randomized controlled trials, nearly half of elementary students given dedicated time to use an AI literacy platform independently never logged on at all. Those who did averaged just 2 to 5 minutes per week. Adding a human tutor focused on engagement increased usage by 71 to 80% — but didn't move reading achievement. The researchers' conclusion: access alone is insufficient; implementation and human support matter for promoting meaningful engagement with AI-based learning tools. These studies are not in tension. They are two sides of the same argument. AI can produce extraordinary gains for students — including the most historically underserved — when it is grounded in learning science, paired with meaningful teacher guidance, and implemented with structured support. Without those conditions, students don't use it at all. This is what Digital Promise's work is built on. In our own pilots across 14 Texas districts, the determining variable wasn't the tool — it was whether educators had structured implementation support integrated into their instructional priorities. Our four-part framework — learning science R&D, educator capacity and implementation, state infrastructure, and shared measurement — exists precisely because the Nigeria study and the Stanford study are both true at the same time. The potential is real. The gap is implementation. That's where the investment needs to go. World Bank study: https://lnkd.in/gEvAwkar Stanford/NSSA study: https://lnkd.in/gZcJ6qe4

While #AI holds incredible potential to drive powerful learning experiences, we must balance that promise with the reality on the ground—ensuring we aren’t just adding burdens to teachers without practical guidance. To move from reactive to intentional, we need to ground approaches in real classroom experiences. That’s why I'm excited to hear directly from Digital Promise and BrainPOP researchers during our upcoming webinar. They will be sharing insights about what AI literacy looks like in classrooms today, what's working, what's not, and what educators need most right now. Join us on July 15 for “What Does AI Literacy Actually Look Like in Classrooms.” Register for the webinar here: https://brnpop.co/4xz8wdp

We scanned Hugging Face, GitHub, and Digital Object Identifier (DOI) repositories for the open building blocks of education AI. We found 1,500+ data-bearing records — datasets, benchmarks, curriculum artifacts, assessment resources — contributed globally, from early childhood to adult learners. The surprise isn't scarcity. It's friction. ~80% have some documentation, but inconsistent metadata holds the ecosystem back: 61% don't even specify a grade span, so you can't tell if a dataset is for kindergartners or college students. The unlock is unglamorous but powerful — standardized metadata, grade-span tags, and machine-readable, data-file-level licensing funders can require. That's how scattered public goods become public infrastructure. This is the work of Digital Promise's K-12 AI Infrastructure Program — and DrivenData just launched a new platform to put these datasets, models, and benchmarks to work. Read the scan and explore it here: https://lnkd.in/gAJrfgfa #AIinEducation #EdTech #DigitalPublicGoods #OpenData Jeremy Roschelle, PhD, Xin Wei, PhD, Rebecca Griffiths, PhD.

Learning technology is never a "silver bullet"—it is a powerful tool to teach with, not a substitute for a human instructor. Introductory statistics is a notorious gateway course that frequently acts as a major structural barrier to college degree completion, disproportionately impacting low-income students and learners of color. To tackle this challenge, the Gates Foundation funded Lumen Learning to build an equity-centered courseware product called Lumen One. At Digital Promise, our post-secondary research team conducted independent, rigorous research across 63 introductory statistics classes and 20 higher education institutions to analyze how instructors integrated this tool. The data confirms our core theory of change: technology shines brightest when deliberately paired with human-centric, Evidence-Based Teaching (EBT) practices. When instructors leveraged EBT strategies—such as instructional transparency, reducing statistics anxiety, and fostering an inclusive sense of belonging—the impact was undeniable: - Stronger Affective Engagement: A massive correlation (r=0.61) with students' confidence, interest, and success expectations in statistics. - Measurable Learning Gains: Higher conceptual understanding and performance on our cross-institutional standardized assessment. - Higher Achievement: A statistically significant increase in final course grades. The report outlines 7 practical recommendations for educators looking to effectively weave digital courseware into active, peer-collaborative learning routines. Let's shift our focus from what technology can automate to how it can free us to build deeper connection, empathy, and mastery in our classrooms. Dive into the data and read the full report here: https://lnkd.in/gz-Nve6r Barbara Means, Ph.D., Julie Neisler, Ph.D. Daniel A. Parker, Ed.D. Vanessa Peters Hinton, PhD, @Pressler, Ph.D.

Proud that Digital Promise is a part of this important work.