Education and AI in Africa: An Exclusive Interview with Alhassan Mohammed Nuruddin | Festina Lente - Your leading source of AI news | Turtles AI

1. Could you please introduce yourself briefly and tell us a bit about your background?

I’m Alhassan Mohammed Nuruddin, a machine learning engineer with expertise in developing AI solutions across healthcare, education, and agriculture. I am currently working at minoHealth AI Labs, where we have developed the world’s first multimodal generalist AI doctor. My work focuses on leveraging AI to create innovative and impactful solutions that address real-world challenges and improve lives.

2. Can you provide an overview of the current status of general education in Africa? What are some of the key challenges and opportunities?

The current state of education in Africa is marked by a mix of progress and challenges. On one hand, there has been significant improvement in enrollment rates and literacy levels across the continent. Challenges persist, such as inadequate infrastructure, a shortage of qualified teachers, and limited access to quality educational materials. There are also disparities between urban and rural areas, with rural communities often having less access to quality education.

While I can’t speak for the entire continent, my experience in Ghana suggests that education in Africa is evolving rapidly. We’re seeing increased adoption of technology in classrooms, but there are still significant challenges such as resource limitations and uneven access to quality education across different regions. However, there’s a growing recognition of education’s importance in driving economic development.

3. How would you describe the status of infrastructure in Africa, particularly in relation to supporting educational initiatives?

Infrastructure in Africa is varied, with some regions experiencing rapid development while others lag behind. Urban areas often have better access to facilities like electricity, internet, and transportation, which are crucial for educational initiatives. However, rural areas often struggle with unreliable power supply and limited internet access, which can hinder educational efforts.

Investments in infrastructure, particularly in ICT, are crucial to support educational initiatives. Governments and private sectors are increasingly investing in improving infrastructure, but there is still a significant way to go to ensure equitable access across the continent.

4. What is the current state of research in AI within Africa, and how is it evolving?

AI research in Africa is growing, with several universities and research institutions establishing dedicated AI centers. There is a strong focus on applying AI to address local challenges, such as healthcare delivery, agriculture, and language processing for African languages. The African AI ecosystem is supported by organizations like MinoHealth AI Labs, Data Science Africa, KaraAgro, Deep Learning Indaba, and various AI labs and hubs across the continent.

Collaborations with international organizations and the growth of local tech talent are contributing to the evolution of AI research in Africa. However, challenges remain, including limited funding, access to data, and computational resources. Despite these challenges, there is a vibrant community of researchers and practitioners driving innovation in AI across Africa.

5. Can you tell us about your experience at the 2nd Workshop on AI in Education in Africa (AfricAIED 2024)? What were the key highlights for you?

The AfricAIED 2024 workshop was an incredible experience, offering a platform to connect with other innovators and thought leaders focused on transforming education in Africa through AI. One of the key highlights for me was showcasing our project, the NSMQ Assistant, which aims to support students preparing for the National Science & Maths Quiz. It was inspiring to see the enthusiasm and commitment of various stakeholders in driving educational change across the continent.

The workshop emphasized the importance of creating AI solutions that are contextually relevant and accessible. There were insightful presentations on the integration of AI in educational systems and panel discussions with experts like Professor Elsie Effah Kaufmann on policy frameworks. Additionally, the opportunity to network with other participants and learn from their experiences was invaluable.

6. Your team placed second in the AfricAIED hackathon. Can you describe the solution you presented and what inspired its development?

Our team developed the NSMQ Assistant, an AI-powered tutor designed to help students excel in the National Science & Maths Quiz (NSMQ). Our solution also included a multimodal large language model (mLLM). The assistant provides personalized, adaptive support tailored to each student’s unique strengths and areas for improvement. It utilizes advanced language models to deliver accurate, succinct, and insightful responses based on past quiz content.

The inspiration for the NSMQ Assistant came from the need to address the challenges faced by students preparing for the NSMQ, such as limited access to quality resources, lack of personalized feedback, and inadequate simulation of the quiz environment. We wanted to create a solution that provides continuous support, tailored feedback, and high-quality resources, enabling students to maximize their potential and succeed in the competition.

7. What specific problem in education does your AI-powered tool aim to address, and how does it work to solve it?

The NSMQ Assistant aims to address several challenges faced by students preparing for the National Science & Maths Quiz. These challenges include limited access to quality resources, lack of personalized feedback, and inefficient study methods. Our solution provides a comprehensive and intelligent tutoring system that offers personalized, adaptive, and interactive learning experiences.

Key components of the NSMQ Assistant include:

• Vision Model: Utilizes MiniCPM-Llama3-V 2.5 for visual content understanding.
• Question Answering: Provides accurate and contextually relevant answers using a dedicated system prompt and a question-answering chain.
• Document Retrieval and Summarization: Uses FAISS for efficient document retrieval, splitting documents into manageable chunks and vectorizing them for quick and accurate retrieval.
• Chat Interface: Built with Gradio, it allows students to interact with the assistant in real-time, simulating a conversational tutoring experience.
• Language Models: Utilizes state-of-the-art language models like Meta Llama 3 70B and MiniCPM-Llama3-V-2_5 for accurate and contextually relevant responses.

By providing 24/7 availability, personalized feedback, and access to past NSMQ questions, the assistant empowers students to engage in continuous, self-directed learning, ultimately improving their performance in the NSMQ.

8. How did your team prepare for the hackathon, and what were some of the biggest challenges you faced during the competition?

Our team prepared for the hackathon by conducting extensive research on the challenges faced by NSMQ students and brainstorming ideas to address these challenges effectively. We developed a prototype of the NSMQ Assistant, focusing on ensuring its adaptability to various linguistic and cultural contexts across Africa.

One of the biggest challenges was time. We joined the hackathon when it was half way through. Additionally, we had limited resources during the hackathon so we could not, for instance, finetune the mLLM for better visual question answering, but we managed to stay focused and deliver a functional prototype. Overcoming these challenges required effective collaboration and innovative thinking from the team.

9. Can you share more about your teammate, Jeremiah Ayensu, and his contributions to the project?

Jeremiah Ayensu is a talented AI engineer. He is my teammate and fellow AI Engineer at MinoHealth AI Labs. He was instrumental in developing the NSMQ Assistant. His expertise in natural language processing and LLMs complimented my experience in building AI systems. Jeremiah took the lead on the mLLM.

10. What was the overall atmosphere at the workshop and hackathon? How did it feel to connect with so many brilliant minds?

The atmosphere at the AfricAIED workshop was amazing, filled with excitement and a sense of purpose. It was inspiring to be surrounded by individuals who are passionate about using AI to transform education in Africa. Connecting with like-minded professionals and sharing ideas was a highlight of the event.

The workshop provided an invaluable opportunity to engage directly with participants and experts including members of the Kwame AI team, whereas the hackathon, being online, presented its own set of challenges in terms of collaboration and communication. Nonetheless, both events fostered a strong sense of collaboration and a shared commitment to driving positive change. The energy and enthusiasm of the participants were contagious, and it was a valuable opportunity to learn from each other and forge meaningful connections. The experience reinforced the importance of working together to address educational challenges and create innovative solutions.

11. What are some of the most innovative solutions you witnessed at the AfricAIED 2024, aside from your own project?

Aside from our project, I was impressed by several innovative solutions at AfricAIED 2024. One standout was the solution that came 1st position. The solution incorporated speech to text and text to speech and a very interactive interface that could even simulate the quiz process.

12. How do you see AI transforming education in Africa in the next five to ten years?

In the next five to ten years, AI has the potential to significantly transform education in Africa by making it more personalized, accessible, and efficient. AI can enable the development of adaptive learning platforms that cater to individual students’ needs, helping them learn at their own pace and style.

AI can also improve teacher effectiveness by providing tools for automating administrative tasks, allowing teachers to focus more on instruction and student support. Additionally, AI-driven insights can inform policy decisions, ensuring resources are allocated where they are needed most. Overall, AI has the potential to bridge educational gaps and improve outcomes for students across Africa.

13. What role do you believe open-source projects play in democratizing education across Africa?

Open-source projects play a crucial role in democratizing education across Africa by providing accessible and affordable tools and resources. Open-source software can be adapted to local contexts and languages, making it easier to implement in diverse educational settings.

These projects encourage collaboration and knowledge sharing among developers, educators, and policymakers, leading to more innovative and contextually relevant solutions. By lowering the barriers to entry, open-source projects empower educators and learners to access quality educational resources, fostering a more inclusive and equitable educational landscape.

14. Can you elaborate on the significance of using AI models to help reduce domestic differences and improve cooperation and social welfare?

AI models can play a significant role in reducing domestic differences and improving cooperation and social welfare by providing data-driven insights and personalized interventions AI can identify disparities in resource allocation, educational outcomes, and access to services, enabling targeted interventions to address these gaps.

AI-driven tools can facilitate better communication and understanding across diverse communities, fostering cooperation and collaboration. By optimizing resource distribution and enhancing decision-making processes, AI can contribute to a more equitable and harmonious society, improving overall social welfare.

15. Beyond education, what other fields in Africa do you see being significantly impacted by AI?

Beyond education, AI has the potential to impact various fields in Africa, including healthcare, agriculture, finance, and transportation. In healthcare, AI can improve diagnostics, optimize treatment plans, and enhance healthcare delivery in remote areas.

At minoHealth AI Labs, we have developed a generalist AI model that aims to provide comprehensive healthcare solutions, making healthcare more accessible and efficient.

In agriculture, AI can significantly impact precision farming, crop monitoring, and pest control, leading to increased yields and improved food security. Our initiatives, such as KaraAgro CADI AI and KaraAgro AI4Cashew, support farmers by acting as early warning systems to quickly identify problems in their cashew farms, helping them keep their crops healthier and more productive. These tools empower farmers with real-time insights and recommendations, enabling them to make informed decisions and improve their farming practices.

The finance sector can also benefit from AI-driven solutions for fraud detection, credit scoring, and financial inclusion, making financial services more accessible and secure. AI can enhance transportation systems by optimizing routes, reducing congestion, and improving logistics, contributing to more efficient and sustainable urban mobility.

Overall, AI has the potential to drive innovation and efficiency across multiple sectors in Africa, contributing to economic growth and development. By leveraging AI technologies, we can address pressing challenges, improve livelihoods, and create a more prosperous and equitable future for the continent.

16. What were the most valuable insights or feedback you received from the judges and other participants regarding your project?

The most valuable feedback we received from the judges and participants was the importance of scalability and adaptability in our solution. They emphasized the need to ensure our AI-powered tool could be easily customized to different linguistic and cultural contexts across Africa.

We also received valuable insights on improving user engagement and accessibility, ensuring that the NSMQ Assistant remains user-friendly and effective for a diverse range of students. 

This feedback will guide the continued development and refinement of our solution to maximize its impact.

17. How important do you think it is for events like AfricAIED to be held in Africa, and what impact do they have on local communities and the broader AI and education sectors?

Events like AfricAIED are incredibly important for the AI and education sectors in Africa. They provide a platform for local innovators to showcase their work, share ideas, and collaborate on solutions to pressing educational challenges. These events help build a vibrant community of AI practitioners and educators dedicated to improving education across the continent.

By bringing together diverse stakeholders, AfricAIED fosters knowledge exchange and networking opportunities that can lead to valuable partnerships and collaborations. The impact of these events extends beyond the participants, inspiring local communities and encouraging the development of homegrown solutions that address unique African challenges.

18. What are the next steps for your project post-hackathon? Are there any plans for further development or deployment?

Post-hackathon, we have several plans to enhance the NSMQ Assistant and expand its impact. Our future plans include, but are not limited to:

1. Add Speech-to-Text (STT) and Text-to-Speech (TTS): We aim to integrate STT and TTS capabilities to make the assistant more accessible and interactive, allowing students to engage with the tool using voice commands and receive spoken feedback across diverse languages.
2. Optimize Current Version: We are focused on optimizing the current version of the NSMQ Assistant to improve its performance, responsiveness, and accuracy in delivering educational content.
3. Fine-Tuning Text and Multimodal LLMs: To enhance the assistant’s capabilities, we plan to fine-tune both text and multimodal large language models for better performance, ensuring more accurate and contextually relevant responses.
4. Add Agentic Features: We plan to introduce agentic features to improve the assistant’s ability to handle complex question-answering tasks, providing students with more comprehensive support.
5. Integrate Virtual Peer-to-Peer Learning Features: We aim to incorporate virtual peer-to-peer learning features, enabling students to collaborate, share knowledge, and learn from one another within the platform.
6. Build Desktop and/or Mobile App: To increase accessibility, we plan to develop desktop and mobile applications, allowing students to use the NSMQ Assistant conveniently on their preferred devices.
7. Partner with Schools and Educational Institutions: We are exploring partnerships with schools and educational institutions to promote wider adoption of the NSMQ Assistant and ensure it aligns with curriculum requirements, providing maximum value to students.