Reflections from the Conference on the State of AI in Africa in Nairobi, Kenya.

Every now and then, I attend a conference that not only fills my notebook but also fills my spirit. The kind that ignites my mind, shakes my perspective and leaves me buzzing with ideas. When that happens, I decide not just to write a post but to do a proper write-up to share the insights gained. The State of Artificial Intelligence in Africa Conference (COSAA) was one such event.

In the first week of June, I found myself marinating, yes, marinating in the brilliance of some of Africa’s most disruptive and delightful minds in the AI space. And before I dive into the flood of insights, I must pause to give flowers where they’re due. A massive shoutout to the CIPIT (Centre for Intellectual Property and Information Technology Law) not just for organizing this event but for hosting us with such hospitality and 5-star treatment at the conference.

The Conference: I wish I could summarise everything each speaker and panelist shared because every contribution added such depth to the conversation but unfortunately, I can’t capture it all. Allow me to briefly highlight a few insights from the keynote speeches and touch on a few from the panel discussions.

Welcome Remarks

So the conference opened with welcome remarks from the brilliant yet humble Dr. Melissa Omino (LLD) , who set the tone for the conference by redefining Africa’s place in the AI landscape. She reminded us that we are not merely consumers of algorithms designed in the global North but we are currently the architects, builders and most importantly, custodians of our digital future. She spotlighted several initiatives across the continent.The NOODL license was an interesting one to learn about. To briefly explain the NOODL License is a tiered open and equitable license that addresses the differences in the needs and contexts of users of African datasets. You can learn more about it here. It was clear from her speech that AI is not just a luxury in Africa but a lifeline that can address our most pressing and critical needs.

The Keynote Speeches

• Prof Caroline B Ncube keynote took us deeper into the intricate world of AI governance with her keynote titled "Forging an effective AI governance, Policy and Regulatory Toolkit for the Africa we want." She painted the governance landscape with precision whereby she noted that creating rules, strategies and policies at both national and continental levels means managing a chessboard of competing visions. I quote “Stakeholders arrive at the negotiating table armed with different interests,” she said. “And our job is to reconcile those into instruments that serve the greater goal.” She stressed that we need to fashion AI policies that can work seamlessly across borders in Africa. She reminded us that Africa’s AI future won’t be built by accident. It must be constructed deliberately, inclusively and wisely. She also shared the book below that is quite insightful. Please get your copy here and you will learn alot.

• The next keynote was one I enjoyed and learnt a lot from Kathleen Siminyu . Her presentation was titled “The Data Behind the Dialogue,” was refreshing to me. She peeled back the layers on one of AI’s most overlooked frontiers: language diversity in AI research. With clarity and charm, she walked us through how African voices are too often missing from the datasets that shape global machine intelligence. She offered a striking example of how there has been an attempt to use religious texts in some AI models for the purpose of getting diverse languages in AI systems. As religious texts are often written in multiple languages like kiswahili. While creative, she went ahead to show how there were challenges in using religious texts. She underscored that domain diversity matters in data used for training models. Religious text alone cannot represent the fullness of African linguistic and cultural nuance as the words are normally used to reflect more religious setting. She also hinted that data is the most transparent part of any AI system and therefore the easiest to interrogate. She said “This is a train that’s already moving,” “Whether we choose to be part of the representation or not, we are being represented.” I can say her call was simple but urgent: care about the data. Interrogate it. Diversify it. Own it. Because the quality of our future AI systems will only be as good as the stories we feed into them. This made me think that African stories, slang, proverbs, unique terms, arguments and even jokes deserve a place in the dataset which is something we Africans have to make a reality.

• Leonida Mutuku gave a keynote titled “Harmonising Progress: Striking the Balance Between AI Innovation and Regulation through Data Governance,” which was a case study in how ethics meets practice in data. Drawing from her work at the Local Development Research Institute , she took us through the painstaking but powerful journey of building AI ready datasets with farmers for over 3 years and across seasons. Her team ensured that they did not just collect the data but went ahead to educate farmers on data protection, obtained informed consent, built trust with the community and gave power to the farmers. This served as a reminder that we cannot always scrape data and believe it will be fit for our use. We have to get our hands dirty to create clean, complete and quality dataset. Importantly, part of the data they built was open sourced which enables others to build solutions grounded in real and contextual African data. In a field often criticized for data extraction without return their approach felt heroic.

• The Keynote by Dr Albert Njoroge Kahira was one that left us in deep thinking & reflections. A few things his speech hinted at are :

1. If Africa is open for innovation, then we need to say it clearly with policy, incentives and national strategies that attract talent, capital and partnerships that suit us.
2. Talent: He stated that we do not have a shortage of Talent but we lack of proper strategy to handle the talent. He stated that if we’re not intentional about our talents we’ll keep exporting our genius and importing their products, paying for what we should be creating. So let’s ask: Where does our talent go? What would it take to have these brilliant minds stay and build at home? It was clear that we need to treat AI talent like national infrastructure with bold, long-term strategies for education, retention and innovative ecosystems.
3. Positioning Africa in the Global AI Value Chain: Too often, we’re either the raw material suppliers or the consumers of finished tech. Rarely are the owners. Rarely the integrators. Rarely are the decision-makers. We need to move up the value chain not just providing data, but shaping the models. Not just mining minerals, but building chips. Not just testing tools, but deploying our own. That’s how you change Africans’ future: by changing our position in the system. We need to bring a proper value to the table.
4. Inclusion: He insisted that our innovation must include our communities. We should not let innovation exacerbate inequalities. We can’t keep creating systems that are smarter than us, but not fairer than the world we come from.
5. INFRASTRUCTURE INFRASTRUCTURE INFRASTRUCTURE: Without infrastructure, there is no innovation. He clearly stated that when we talk about infrastructure, we are talking about sovereignty not just talking about tools. If we are serious about building Africa’s AI future, then infrastructure must be built in Africa, by African institutions and guided by African priorities. This means that for Africa to truly participate in the AI era, we must have a first-class seat not in policy only, but in every technical layer: From energy to chips, from training to deployment. It’s about building infrastructure that reflects Africa’s realities not our donors’ convenience. Infrastructure is not a nice-to-have. It’s the backbone.

In closing the speech, he stated that

" So when we say “The State of AI in Africa” it’s not just a title. It’s a statement of intent. We are not just observing the state. We are defining it. We are not just reacting to trends. We are setting them. We are not just building tools. We are building the future.

And remember nobody’s coming. Nobody’s coming to build for our problems. Nobody’s coming to prioritize our languages. Nobody’s coming. But that’s not something to fear. That’s something to understand. Because when you stop waiting, you start building. You start shaping the future like it belongs to you.

So let me leave you with this: We have the talent. We have the urgency. We have the vision. Africa doesn’t just need innovation. Africa needs ownership."

• Prof. Chijioke Okorie, PhD keynote speech was also quite insightful, she focused on how Intellectual Property (IP) can be used to scale AI innovation such as adopting equitable open access. She also hinted at the need to leverage the IP exclusive right & exceptions (such as Fair Use) to ensure they both work for AI innovation in Africa. It was clear from her speech that as Africans we need to define how IP laws should be fashioned to fit our context in this AI era. Policy discussion and intervention by the government are necessary and required to address the governance of IP from development to use to the enforcement of IP rights.

Presentation: AI practitioners on Responsible AI - AI perception and Operationalisation

Betsy Muriithi-Ochieng' 's presentation i believe, captured the depth and urgency of the conversation on responsible AI. It was refreshing to hear an innovator emphasize the need for regulation. Five key points from her presentation were:

• Ethics is Under Siege in Tech Policy: The word ethics is weaponized in support of deregulation, self-regulation or hands-off governance which is alarming. She highlighted how “ethics” is often co-opted by tech companies to appear responsible while avoiding meaningful accountability or regulation.
• Superficial Compliance Undermines True Responsibility: Responsible AI frameworks risk becoming tick-box exercises, especially when funders require them more for reporting than for actual ethical impact. This point warns against reducing ethical practice to administrative formalities that serve optics rather than meaningful considerations.
• Self-Regulation Has Limits and Real Consequences: As she explained that we’re all seeing the limit of self-regulation from chatbots giving harmful advice to teens, to misinformation being embedded by design or negligence such by GROK. This underlined how unregulated AI can result in serious harms including misinformation and even loss of life which points to the urgent need for legal frameworks.
• Responsible AI Must Be Participatory and Contextual: She hinted that we must define what responsible AI means with users, using value-sensitive design and participatory approaches to embed local values and priorities. This was a clear call for co-designing AI systems with targeted communities rather than imposing external definitions of ethics or responsibility in our local communities.
• Empathy Over Compliance in AI Development: She posed a question that Do we need ethical people building technology, or do we need ethical technology? it is clear that we need ethical people designing technology that is ethical. If you're aware of the harms and still build it, that’s on you as the developer or innovator NOT the technology. This challenges developers or innovators to internalize ethical awareness and act with conscience not just policy adherence especially now that regulations are not in place.

She also shared a link with various toolkist and resources that you access here

Insight from Panel Discussions

The panel session, “Charting the Path: Understanding the Kenya National AI Strategy,” moderated by Dr. Melissa Omino (LLD) , the session brought together leading experts including Leonida Mutuku , linet kwamboka , Mark Irura and kevin mwiti njeru. What I learnt from the panel and Kenya AI Strategy? The inclusivity. Kenya’s AI strategy wasn’t drafted in isolation it was powered by stakeholder participation and multi-sectoral collaboration. I think that’s how you build a strategy. The panel unpacked how the strategy addresses immediate needs like capacity building, digital infrastructure and local innovation while also laying a foundation for ethical AI governance.

The Panel Titled "Contextualising AI Regulatory Approaches in Africa" moderated by Butera Michael tackled some of the interesting issues: Desmond Oriakhogba, PhD clearly stated that “AI is not human. It’s a tool. A legal object not a legal subject.” From an intellectual property standpoint, he stressed that liability must always trace back to the humans that is the developers, deployers or operators behind the system. Globally, from the legal standpoint, there seems to be an understanding that we must focus on the legal persons involved, not the code when liability comes into question for AI systems.

I had the opportunity to be a panelist alongside Ridwan Oloyede during a Fireside Chat titled “National AI Strategies: Policies for Growth and Inclusion – East African Perspectives”, excellently moderated by the amazing FLORENCE OGONJO . I shared insights on the evolving AI policy landscape in Tanzania, where we are in the final stages of completing our National AI Strategy which is set to launch next month in July. Ridwan offered valuable perspectives on Rwanda’s progress in this space and highlighted lessons that other East African countries can learn from. It is clear from the discussion that there are a lot of opportunities for collaborations among EAC in AI development & policy.

The Panel titled Data Governance in Shaping the AI Regulatory Landscape in Africa well moderated by Joshua Kitili was insightful and left us thinking that with AI in play, there is more to be done in Data Governace. One key point from the discussion was on Cross-border data challenges that seem to be a huge hiccup in the region due to varying data localization laws. We all agree we need data to flow within our borders for AFCTA and even the EAC bloc. Nevertheless, there are no proper policies for data sharing and interoperability. Another point hinted on was the varying approaches and standards of our countries in the enforcement of the data protection laws whereby it makes it difficult to harmonise the policies at the regional level.

CIPIT closed the conference on a high note by launching the East African AI Policy Hub. The Hub is designed to advance responsible and ethical AI policymaking across the region. It will serve as a collaborative platform for knowledge sharing, policy dialogue and regional influence in AI governance and development. This is a major milestone and I hope that the EA will lead the way in shaping, developing and implementing responsible AI policy on the continent.

Conclusion

As I boarded my flight back home, my heart was truly full. It was full of ideas, full of hope and most importantly full of gratitude to have been part of a gathering of minds that have a bold visions for the future of AI in Africa. People across different sectors are truly pushing boundaries and making AI a reality in Africa not just in theory, but in action. It’s truly inspiring. COSAA 2025 wasn’t just a great conference. It was a reminder that Africa’s future in AI is not just possible it’s already being built.