I’m leading a research project that asks a pretty specific question: what does responsible AI in higher education look like in places that don’t look like the places where most AI-in-education research gets done? Most of what we read about AI in classrooms comes from well-resourced settings — universities with fast internet, students with laptops, institutional licenses for the latest tools. But that’s not where most of the world’s students actually learn.

So I’ve been working with two partners on a pilot study at the University of Kabianga, a public university in semi-rural Kenya. The partners are an EdTech company called Three Dots Education, which builds AI-powered learning tools that run inside WhatsApp, and a faculty member at Kabianga who teaches a course in their teacher education program. Twenty-nine second-year students — future teachers themselves — used the platform for eight weeks as part of their normal coursework.

The Challenge

Four things, in plain terms: Does this kind of AI support actually help students understand the course material? Does it reach everyone equally, or does it quietly leave some students behind? How does the instructor experience handing some teaching tasks to an AI — does it free them up, or does it feel like a loss of control? And — this one matters most to me — how does using AI as a learner change how these future teachers will eventually teach with AI themselves? They’re the next generation of Kenyan classroom teachers, and they’ll be making decisions about AI in their own classrooms within a few years.

The bigger ambition is to add evidence from a context that the field rarely studies. Public universities in the Global South are where a huge share of the world’s teachers are being trained, and what works there matters — both for the students themselves and for how we think about AI in education generally. If the model holds, it suggests a much lower-friction path to AI-supported learning than the one most EdTech assumes.

Impact & Outcomes

The pilot ran. That sounds small but isn’t — running an eight-week, embedded AI pilot in a public university in semi-rural Kenya, in a credit-bearing course, with faculty buy-in and a functioning research apparatus, is itself an outcome. Twenty-nine pre-service teachers used AI-supported learning tools as a structured part of their coursework, most for the first time in a pedagogical — rather than informal “ask ChatGPT for my homework” — frame. That experience is now part of how they’ll think about AI when they enter classrooms of their own.

The baseline has already produced findings worth disseminating, including a counter-finding: students are more AI-familiar than the design assumed. These will anchor the post-pilot manuscript. For the University of Kabianga, this is a documented model of low-infrastructure digital learning. For Three Dots Education, this is the first rigorous external validation of their platform in a public-university teacher-education context. This work was carried out in collaboration with Anaelle Guillen-Meyer and Soufiane Bensaid of Three Dots Education, and Dr. David Ngatia and Dr. Florence Kamonjo in the School of Education at the University of Kabianga.

“This project asks whether AI can broaden access in education or whether it will quietly narrow it — a question that should matter to every CPS program.”

Dan Serig

Connection to CPS

CPS’s professional-studies mission has always had an access dimension — meeting learners where they are, professionally and geographically. This project extends that orientation internationally and across infrastructure constraints. My scholarly trajectory at CPS has been about integrating qualitative methods with questions of equity in education. The Kabianga project is where those commitments meet AI — a topic the field is still figuring out how to study responsibly. The project is shaping the next phase of my research and teaching at CPS, including how I bring AI and equity into the virtual spaces where I teach.

This project sits at the intersection of three things CPS uniquely supports: research grounded in real professional practice, attention to the conditions under which adult and professional learners actually study, and an honest reckoning with what AI is doing to education.

Connect with Dan on LinkedIn.