Africa Strategic Policy and AI Resource Centre (AFRI-SPARC) Reviews 

The Afri-SPARC Fellowship was a valuable learning experience. It deepened my understanding of AI policy and governance in Africa and gave me the opportunity to work on real policy research. Through the fellowship, I improved my research, policy analysis and critical thinking skills while gaining a better understanding of the gap between AI policy commitments and actual implementation. I am grateful for the knowledge, experience and opportunity to learn from others in the field.

They classes were interactive, and engaging.

It was a wholesome experience with very relevant instructors, feedback and hands on skills experiences. Very very practical. I am grateful for this experience.

What made my experience great in this company is the way we practice how AI should be governed and how we can evaluate AI policies of different countries. What I love the most in my AI policy and research fellowship is how assignments are designed in the way that helps us to engage with AI governance problems and give us a chance of practicing as if we are really working on these problems as a job. That's is really great!

I love the content of the course and the manner of meeting weekly for classes and discussion. I had a bit of difficulty working with my group mates because some were imposing but it also came with the lesson to co-exist and work with diverse groups.

AFRI-SPARC is an interesting initiative focused on artificial intelligence, strategic policy, and responsible AI development in Africa. Its focus on connecting technology, research, and policy has the potential to contribute to Africa’s digital transformation. I look forward to seeing its continued impact in supporting innovation, knowledge sharing, and AI development across the continent.

My experience was exceptional. I got a chance to interact with many AI governance professionally across the continent of Africa.

AFRI-SPARC's adversarial evaluation of SASKI and MyTrusted.ai was one of the most valuable things we've done as a company.

They tested us across 105 criteria and pushed hard where it matters most for AI that talks to people: suicidal language detection and 988 routing, holding boundaries under emotional pressure, multilingual privacy protection, dependency detection and jailbreak resistance. Seeing top of scale results in those areas gave us confidence our deterministic approach works.

Just as important, they found what we missed. Their evaluation surfaced 12 critical output enforcement issues, and we fixed every one in code with passing tests. That's exactly what an independent evaluation should do: not just confirm what works, but make the product safer for the people who rely on it.

If you're building AI for vulnerable users, I'd recommend putting it in front of AFRI-SPARC before it goes in front of the public.

Stephen Calhoun
Founder & CEO, SASKI Institute