The rapid rise of Generative Artificial Intelligence (GenAI) has sparked one of the biggest conversations in education in recent years. While some educators worry that tools such as ChatGPT and other Large Language Models (LLMs) could undermine academic integrity or diminish critical thinking, others see an unprecedented opportunity to redefine how teaching and learning take place. 

Seeking to contribute to this growing pool of knowledge, Dr. Benson Kituku, a Senior Lecturer in the Department of Computer Science and Dean, School of Computer Science and Information Technology at Dedan Kimathi University of Technology (DeKUT), embarked on a research project titled, Supporting Active Learning and Creativity in Computer Science Using a Large Language Model. The project was supported by a Ksh. 1.5 million KENET Education Technology (EdTech) Innovation Grant awarded in the 2023/2024 Financial Year.

The funding facilitated the acquisition of technological infrastructure required to access AI tools, involvement of Master’s students, engagement of a software developer who accelerated system development, publication of research findings, and stakeholder workshops to disseminate the findings. The grant also provided modest facilitation for lecturers who participated in the classroom experiments, acknowledging their contribution beyond their normal teaching responsibilities.

Apart from financial support, Dr. Kituku noted that the project opened valuable opportunities for collaboration through KENET's research ecosystem. Interactions with fellow researchers enriched the study, while invitations to present the project's findings in different workshops, including a Vice Chancellors' meeting, created opportunities to contribute to national conversations on the responsible adoption of AI in higher education.

Dr. Benson Kituku, Dean, School of Computer Science and Information Technology at Dedan Kimathi University of Technology (DeKUT).

According to the lecturer, traditional teaching approaches in many universities remain largely content-driven, where lecturers prepare notes, deliver them during class, and assess students at the end of the learning process. While this model has served higher education for decades, it often leaves students as passive recipients of information despite the availability of emerging technologies that can promote participation, creativity and deeper understanding.

"Our focus was to establish how lecturers can use Generative AI to facilitate active learning instead of passive learning," noted Dr. Kituku. "The lecturer must remain at the centre of the learning process, but AI can become a powerful assistant in helping students engage more actively with the content."

After securing the grant, the research team began by identifying practical classroom applications of Generative AI. They developed 19 use cases, which were grouped into three broad areas: instruction and content generation, teaching and classroom delivery, and assessment. From each category, representative use cases were selected and tested in real Computer Science classrooms to determine their effectiveness in promoting active learning and creativity.

One of the first interventions focused on instruction and content generation, where Generative AI was used to develop lecture materials and improve presentation slides. "We were able to use the material we teach to generate better slides using GenAI, but the emphasis was on the human in the loop because the lecturer remains active. You generate the slides, carefully examine them because AI can make mistakes, and then use them in class," explained Dr. Kituku.

To make classroom teaching more interactive, the team integrated the Formative platform into lessons, enabling lecturers to embed multiple-choice questions directly into presentation slides. As students responded to the questions in real time, lecturers could immediately gauge their understanding and adjust their teaching accordingly. "If at least 75 per cent of the class answered correctly, we proceeded with the lesson. If they scored below that threshold, we revisited the concept before moving on." At the end of each lesson, the platform also highlighted questions that students had found most challenging, which formed a recap at the beginning of the next class.

The research also explored how Generative AI could support assessment. The team successfully generated multiple-choice examination questions that were subjected to the same quality assurance and external moderation processes as lecturer-developed examinations. Building on these findings, they developed an AI-powered tool for generating multiple-choice questions, which has since been integrated into DeKUT's Learning Management System (LMS).

Building on the success of the project, Dr. Kituku and his team are now working to enhance the AI-powered assessment tool by expanding its capabilities to generate open-ended questions and support automated marking. The advancement is expected to provide lecturers with more comprehensive assessment options and reduce the time spent on preparing and grading examinations.