Embedding AI Where It Matters: Teacher Support and Instructional Practice in Sub-Saharan Africa
Across Sub-Saharan Africa, education systems face an urgent challenge: how to improve literacy and math skills for students while strengthening the instructional support teachers receive. At the same time, rapid advances in artificial intelligence are reshaping what is possible across education systems globally. While AI is often framed as a transformative force, where and how it should be embedded to meaningfully improve learning remains an open question.
In response, Renaissance Philanthropy, in partnership with the Gates Foundation, is building on their broader work across Sub-Saharan Africa by supporting early, evidence-building efforts to explore where AI can most effectively strengthen teaching and learning, particularly by helping teachers improve their instruction. Rather than treating AI as a standalone solution, this work focuses on how AI might augment instructional support, help teachers improve day-to-day practice, and inform future system-level improvements.
Despite expanded access to schooling, both learning and the quality of instruction are poor. Nearly 90% of 10-year-olds in Sub-Saharan Africa cannot read and understand a simple text. At the same time, fewer than 1 in 10 teachers demonstrate full mastery of the primary curriculum, and most receive limited ongoing coaching or feedback. This gulf between evidence-based instructional approaches and the support teachers need to implement them consistently is a major barrier to improving outcomes at scale.
Instructional Support as the Missing Link
A growing body of evidence underscores the importance of clear instructional expectations paired with ongoing support for teachers. Many education systems have adopted approaches to align lesson plans, teacher guides, and assessments around proven teaching practices. Yet curriculum and materials alone are not enough.
Instructional quality depends on whether teachers receive reinforcement, feedback, and coaching to support effective practice. Without that support, gains in learning are difficult to sustain.
The Coaching Gap Holding Back Impact
For instructional improvement efforts to succeed, teachers need ongoing support, not just initial training. Research shows that pairing classroom observation with personalized coaching can improve student learning. Yet across Sub-Saharan Africa, systems struggle to deliver this support consistently.
High supervisor-to-teacher ratios, limited infrastructure, and uneven pedagogical expertise mean that coaching is often infrequent, delayed, or generic. As a result, follow-up support is often minimal creating gaps between intended instructional practice and what happens in classrooms day-to-day.
Challenges in teacher subject-matter knowledge further compound these issues. Many primary school teachers haven’t mastered the literacy and numeracy content they are expected to teach, reflecting not only limitations in pre-service training, but a systemic absence of sustained pedagogical reinforcement.
Why AI Changes What’s Possible
Recent advances in AI create a meaningful opportunity to rethink how instructional support is delivered. AI-enabled tools can analyze classroom interactions, instructional materials, and their alignment with lesson plans, to generate timely, personalized, and actionable feedback for teachers. Rather than relying solely on sporadic in-person visits, teachers can receive more continuous guidance tied directly to their instructional practice.
At the system level, AI-generated insights can provide leaders with greater visibility into instructional quality and implementation patterns. This enables more targeted professional development and more evidence-informed decision-making. For example, the University of Colorado Boulder’s Hybrid Human-Agent Tutoring (HAT) system analyzes small-group tutoring sessions to provide coaches and tutors with actionable feedback on improving instructional practice. Research showed that when coaches and tutors acted on this AI-generated feedback, they made measurable improvements in key tutoring practices that are directly linked to improved math learning outcomes for over 1,000 students.
While systems such as HAT will need to be adapted to the unique instructional environments in Sub-Saharan Africa, these early indicators suggest the promise of applying the underlying mechanisms within local education systems. In this way, AI has the potential to make instructional support more scalable, consistent, and cost-effective, addressing a persistent barrier to improving foundational learning. Importantly, this is not about automating teaching. The promise of AI lies in augmenting existing instructional support systems and extending their reach, particularly where coaching capacity is constrained.
Building Capacity, Not Just Tools
Realizing this promise requires more than deploying new technology. Many low- and middle-income countries face constraints in local technical talent, infrastructure, and the resources needed for AI-enabled solutions. Without intentional investment in local capacity, even promising tools risk being difficult to maintain and scale.
Pairing AI innovation with the development of local technical and institutional capacity is therefore essential. The Gates Foundation, through projects like the upcoming STEP-AI program (Strengthening Teacher Effectiveness & Practice with AI), is advancing this approach by prioritizing learning, evidence generation, and sustainability alongside innovation. By supporting early-stage exploration focused on real instructional challenges, the Foundation is helping to shape how AI is responsibly explored in service of stronger teaching and learning across Sub-Saharan Africa.
Looking Ahead
Embedding AI-driven feedback and coaching into instructional support systems promises to strengthen teacher practice and accelerate foundational learning. When integrated into existing instruction and paired with investments in local capacity, these approaches offer a responsive, data-informed complement to traditional coaching models.
The new Gates Foundation-funded STEP-AI program is one example of how this agenda is beginning to take shape. STEP-AI will explore how AI-enabled coaching can be embedded within instructional systems, building technical capacity, and generating evidence and public goods to inform the broader field.
These early efforts point toward a future in which AI is embedded where it matters most: supporting teachers, strengthening instructional practice, and helping ensure that more children develop the foundational skills they need to thrive.