Open Forum #67 Open-source AI as a Catalyst for Africa’s Digital Economy

Summary

This discussion focused on the potential of open-source AI to catalyze Africa’s digital economy, featuring panelists from Meta, GIZ, the African Union, and NEPAD’s High-Level Panel on Emerging Technologies. The moderator began by asking participants to define “open AI,” with most clarifying they meant “open-source AI” rather than the company OpenAI. All panelists agreed that open-source AI represents a significant opportunity for Africa, particularly because it provides free access to AI tools that would otherwise be prohibitively expensive for young innovators and entrepreneurs in low- and middle-income countries.

Kojo Boakye from Meta emphasized that open-source AI could help Africa avoid being “behind the curve” in technological revolutions, noting that Meta’s Llama models have been downloaded over a billion times globally. He highlighted successful African applications like Digital Green for farmers and Jacaranda Health for maternal care. Tobias Thiel from GIZ stressed the importance of open access, infrastructure, and innovation for inclusive digital transformation, while Barbara Glover noted that open-source AI allows young Africans to develop AI applications without waiting for expensive proprietary software or advanced infrastructure.

The discussion addressed significant challenges including limited internet connectivity, lack of digital skills, insufficient access to data and computing power, and the need for appropriate governance frameworks. Panelists emphasized that Africa cannot wait to solve all infrastructure problems before embracing AI, but must pursue parallel development strategies. The African Union’s Adil Suleiman highlighted the importance of African sovereignty in AI development, ensuring that AI solutions align with local cultural values and development priorities while supporting the continent’s Agenda 2063 goals.

Concerns were raised about the balance between enabling innovation and maintaining appropriate oversight, with participants stressing the need for collaborative, inclusive governance frameworks rather than restrictive control mechanisms. The discussion concluded with recognition that open-source AI communities need support and sustainability, and that Africa should position itself prominently in global AI governance discussions, including the upcoming World Summit on Information Society Plus 20 process.

Keypoints

## Major Discussion Points:

– **Defining Open Source AI vs. Closed AI**: The panelists clarified the distinction between open source AI (like Meta’s LLaMA models) and closed/proprietary AI systems, emphasizing how open source provides accessible tools for African entrepreneurs, researchers, and governments without requiring massive capital investments.

– **Infrastructure and Digital Divide Challenges**: Discussion of fundamental barriers including limited internet connectivity, lack of computing power, insufficient digital literacy, and the reality that over half of Africa’s population remains unconnected, creating obstacles to AI adoption and implementation.

– **African Sovereignty and Control in AI Development**: Debate around balancing the need for African control over AI systems to preserve cultural values and sovereignty while avoiding overly restrictive regulation that could stifle innovation and entrepreneurship among young Africans.

– **Practical Applications and Success Stories**: Examples of AI solutions already working in Africa, such as Digital Green’s SMS-based farming information system, Jacaranda Health’s maternal care platform, and malaria elimination efforts in Tanzania, demonstrating real-world impact of open source AI.

– **Policy Frameworks and Continental Coordination**: Discussion of the African Union’s AI strategy, data policy frameworks, and the need for harmonized approaches across the continent, including Africa’s positioning in global AI governance discussions like WSIS+20 and the importance of collective bargaining power.

## Overall Purpose:

The discussion aimed to explore how open source AI can catalyze Africa’s digital economy, examining both opportunities and challenges while addressing policy, infrastructure, and implementation considerations for making AI accessible and beneficial across the continent.

## Overall Tone:

The tone was predominantly optimistic and collaborative throughout, with all panelists expressing enthusiasm for open source AI’s potential in Africa. While acknowledging significant challenges around infrastructure, skills, and governance, the conversation maintained a solution-oriented approach. There was some tension around questions of control and sovereignty, but this was handled constructively with panelists finding common ground on the need for enabling rather than restrictive frameworks.

Speakers

**Speakers from the provided list:**

– **Moderator (Alison Gilwald)** – Research ICT Africa, moderating the session on Open AI to Catalyze Digital Economy in Africa

– **Kojo Boakye** – VP of Public Policy with Africa Middle East and Turkey, META

– **Barbara Glover** – Programme Officer of the African Union’s High-Level Panel on Emerging Technologies, participated online from Johannesburg, centrally involved with AI policy development on the continent

– **Tobias Thiel** – Director for the GIZ for the African Union

– **Speaker (Adil Suleiman)** – African Union, Senior Policy Advisor in the Information Society Division 2

– **Audience** – Various audience members asking questions, including Osei Keja from Ghana

– **Christelle Onana** – Online moderator handling questions from online participants

**Additional speakers:**

– **Stéphane Langlois** – Mentioned as joining online for content moderation (though no direct quotes captured in transcript)

# Open-Source AI to Catalyse Digital Economy in Africa: A Comprehensive Discussion Report

## Introduction and Context

This discussion, moderated by Alison Gilwald from Research ICT Africa, brought together key stakeholders to examine the potential of open-source artificial intelligence to catalyse Africa’s digital economy. The panel featured Kojo Boakye from Meta, Tobias Thiel from the German Agency for International Cooperation (GIZ), Barbara Glover from the African Union’s High-Level Panel on Emerging Technologies, and Adil Suleiman, Senior Policy Advisor in the Information Society Division from the African Union.

The session began with a crucial clarification from Kojo Boakye, who distinguished between “open AI” (referring to the company OpenAI) and “open-source AI,” noting: “I would reframe it more as open source AI rather than open AI… My competitors are open AI. would be delighted that I’m going to wax lyrical about them rather than ourselves.” This distinction shaped the entire conversation, with all subsequent speakers adopting the “open-source AI” terminology.

## Defining Open-Source AI and Current Applications

### Consensus on Transformative Potential

All panellists expressed strong consensus regarding open-source AI’s potential for Africa. Kojo Boakye positioned it as a technological leapfrogging opportunity: “This feels as though we are presenting an opportunity through open source AI to put those technical tools in the hands of those who probably most need them… I think open source presents an incredible opportunity to do that.”

Tobias Thiel emphasised comprehensive openness, arguing that “open source AI requires open access, open infrastructure, and open innovation for truly inclusive digital transformation.” Barbara Glover highlighted democratisation, noting that “open source AI provides opportunities for young innovators who cannot afford proprietary software to build AI models and tools.”

Adil Suleiman connected open-source AI to continental goals: “an enabler for achieving Africa’s Agenda 2063, offering transparency and community collaboration while being cost-effective for resource-limited Africa.”

### Practical Success Stories

The discussion featured concrete examples of successful implementations. Kojo Boakye highlighted that “Meta’s LLAMA models have been downloaded more than a billion times, with African winners of Impact Grants like Digital Green and Jacaranda Health creating solutions for farmers and maternal health.” He mentioned Impact Grants ranging “between 50 and 200,000” for such initiatives.

Barbara Glover referenced the Tanzania malaria elimination project using LLAMA and noted “over 2,000 submissions” for the AU’s AI and Robotics competition, demonstrating significant continental interest. Tobias Thiel emphasised localisation potential, noting that “open source AI allows creation of localised datasets and language support for underrepresented African languages like Swahili and Luganda.”

## Infrastructure Challenges and Reality Check

### Confronting Current Limitations

The discussion took a critical turn when moderator Alison Gilwald introduced sobering realities: “We have over half the population of the continent who are not connected to the internet… certainly don’t have the capabilities in the public sector, large parts of the private sector, and certainly not in most of the SMMEs… So how do we get to this opportunity that you’ve been speaking about from where we are now?”

This intervention shifted the conversation from aspirational to practical, highlighting challenges including limited internet connectivity, shortage of data scientists, insufficient computing power, and inadequate capabilities across sectors.

### Strategic Parallel Development

Rather than viewing infrastructure as insurmountable barriers, panellists advocated for simultaneous development. Adil Suleiman articulated: “I think the questions in front of us, whether to first work on the fundamentals, and then do AI, or do them in parallel. And I think we don’t have options. We have to do everything at the same time, which is always our challenge in every area.”

Barbara Glover reinforced immediate possibilities: “Young people don’t have to wait for government to catch up on infrastructure, to catch up on supercomputers and all these issues before they can, you know, harness the world of AI.”

## Governance and Sovereignty: Balancing Control and Innovation

### The Control Debate

A significant tension emerged around appropriate government control levels. Kojo Boakye expressed concerns about excessive control: “I do have concerns when people start speaking about control, because control, for me, negates in some ways… the ingenuity, innate ingenuity and brilliance of young Africans, who, if we put these tools in the hands of people right now, will create the solutions.”

Conversely, Adil Suleiman advocated for African sovereignty: “Africa needs sovereignty and control over AI aspects it wants to adopt, ensuring cultural and traditional requirements are met while building locally.” He provided a nuanced framework: “The metric would be if the African government and the African communities, the African families, they can control what is it, what part of AI they want to take, and what part of AI they don’t want to consider.”

### Collaborative Governance Approaches

Despite differing control perspectives, speakers agreed on collaborative governance needs. Tobias Thiel advocated for “fair, enabling, risk-conscious, and equitable regulation [that] requires collaborative partnerships bringing together public sector, private sector, academia, and civil society.”

The moderator highlighted the challenge: “The challenge lies in balancing protective governance with enabling frameworks that don’t stifle innovation.”

## Continental Strategy and Global Positioning

### African Union Initiatives

Adil Suleiman outlined AU’s comprehensive approach: “The African Union is developing frameworks for open source data and implementation strategies, focusing on creating enabling environments for safe AI deployment.” He mentioned a framework review scheduled for “end of November” during ministerial meetings.

However, coordination challenges were acknowledged, with the moderator raising concerns about “coordination challenges between different African Union initiatives (digital policies vs trade policies) that may conflict with each other.”

### Global Representation

The discussion emphasised Africa’s need for stronger global AI governance representation. Adil Suleiman stressed “Africa’s positioning in global AI decision-making processes, including advocacy for African representation in international AI governance,” specifically mentioning the upcoming WSIS+20 process.

Tobias Thiel supported this through GIZ’s work “on building inclusive AI ecosystems through initiatives like Fair Forward, partnering with local organisations to address data gaps in the Global South.”

## Economic Impact and Competitive Positioning

### Realistic Assessment of Advantages

Barbara Glover provided candid assessment: “Africa probably won’t be a leader in high-performance computing. It won’t be a leader, perhaps, in data centre location… So we need to look then at what is our niche? What is it that we can do better?” She referenced a New York Times article about global data center locations to support this realistic appraisal.

She advocated leveraging “Africa’s young population and existing IT skills for addressing continental challenges,” suggesting demographic advantages could provide competitive edges.

### Practical Business Applications

Kojo Boakye outlined multiplier effects: “Investment in connectivity, device affordability, and training programs creates multiplier effects when combined with government adoption.” He highlighted practical applications: “Open source AI can help small and medium-sized businesses understand cross-border operations and business development through LLAMA-based applications.”

He also referenced Dr. Ndemo’s work in Kenya, particularly the mobile device tax removal example, as demonstrating policy impacts on technology adoption.

## Implementation Strategies and Community Building

### Leveraging Current Opportunities

Rather than waiting for ideal conditions, speakers emphasised engaging with available opportunities. Kojo Boakye noted: “Communities need to come together and access available programs from various tech companies, with many opportunities already existing on platforms like LinkedIn.”

Barbara Glover advocated for “problem-solving approaches through domain-specific education and applications addressing real-time challenges in health, agriculture, and education.” She mentioned the upcoming AI and Health Conference in Cairo as an example of sector-specific focus.

### Sustainable Community Development

An online participant asked about community sustainability, prompting discussion about network effects and collaborative learning. Barbara Glover emphasised that “open source AI communities need support and sustainability through practical applications that solve tangible problems aligned with continental development goals.”

## Unresolved Challenges and Future Considerations

### Technical Openness Questions

Despite enthusiasm for open-source AI, questions remained about true system openness. The moderator raised concerns: “Questions arise about true openness of systems like LLAMA, with concerns about backend accessibility for independent development.”

Barbara Glover identified implementation gaps: “Trust issues in open AI systems and ineffective data harmonisation policies prevent innovators from accessing needed data despite supportive policies on paper.”

### Resource Constraints

Audience member Osei Keja posed a fundamental question: “How can Africa advance AI and other emerging technologies at a fraction of the billions/trillions spent in the global North while ensuring equity and long-term health of these tools?” This encapsulated the core challenge of achieving meaningful AI development within Africa’s resource constraints.

## Key Outcomes and Next Steps

### Concrete Action Items

The discussion identified several immediate actions: the African Union’s open-source data framework review scheduled for end of November, continued connectivity infrastructure investment, and development of African common positions for the WSIS+20 process.

GIZ committed to continuing support for 35 African states requesting help to domesticate AU data policy frameworks, while speakers encouraged communities to actively access existing tech company programs.

### Ongoing Tensions

Several tensions remained unresolved, including balancing African sovereignty with innovation enablement, achieving true backend openness in AI systems, and coordinating different AU initiatives to avoid policy conflicts. The fundamental challenge of scaling AI solutions from connected urban areas to rural areas lacking basic infrastructure requires continued attention.

## Conclusion

This comprehensive discussion revealed sophisticated understanding of complex trade-offs in AI adoption for development. While significant challenges remain, there is substantial consensus among stakeholders about open-source AI’s potential to catalyse Africa’s digital economy. The conversation successfully balanced aspiration with pragmatism, emphasising practical problem-solving approaches, collaborative governance frameworks, and the need for simultaneous development of infrastructure and AI capabilities.

The key insight emerging from the discussion is that Africa cannot afford to wait for perfect conditions but must pursue parallel development strategies that leverage current opportunities while building toward longer-term infrastructure and capacity goals. Success will depend on maintaining the delicate balance between enabling innovation and ensuring African sovereignty in AI adoption.

Moderator: Right, good afternoon, everyone, and welcome to the session on Open AI to Catalyze Digital Economy in Africa. I urge the few scattered people in the room to come and join us in the front. Maybe we can make more of a conversation of this. And thank you for staying for this last session on the second last day. I know it’s been very intensive. But we’d like to have this important conversation. We’ve got these great panelists with us. So let’s begin. We’ve also got Stéphane Langlois online. He’s going to do the content moderation. We’ve got Kojo Bokre, who’s the VP of Public Policy with Africa Middle East and Turkey, FORMETA. We have Dr. Tobias Thiel, who’s the Director for the GIZ for the African Union. I’m not going to do my bad German on that. We’ll have to do. And we’ve got Adil Suleiman from the African Union, Senior Policy Advisor, I believe, as I remember, in the Information Society Division 2, I believe. So thank you, Adil, also for stepping in for the African Union here. And then, of course, we also have Barbara Glover online, who’s the Programme Officer of the African Union’s High-Level Panel on Emerging Technologies, and has been centrally involved with the development of AI policy on the continent. So thank you, Barbara, for joining us, I believe from Johannesburg. I’m gonna start by just asking you all, before we get into the issues of what AI can do for SMMEs and for the digital economy in Africa, to tell us what you understand by this rather controversial, I think it used to be a kind of accepted good, but what do you understand by open AI before we try and look at what it can do for Africa?

Kojo Boakye: Kojo, do you want to give us a run at that? I can try, I can try, and then I may attempt to reframe it slightly. But first of all, I’d like to say a huge thanks to the government of Norway for putting on another wonderful IGF, as well as many of the support staff from what I assume is the Nova Spectrum team. And then Alison did a wonderful job of introducing the three of us, myself and my two learned colleagues. But also, hopefully many of you know, Alison Gilwald, the work of Research ICT Africa, the impact it has had on policymaking, not only on the continent that is most close to my heart, but globally as well. So give her her applaud, I hope you’d already read the bio. Thank you so much, Kojo. I think this is the second panel I failed to introduce myself. It’s not because I expect everybody to know me, I was just getting onto the interesting topics. So I think the question was, as you posed it, was what was the impact that open AI may have on Africa? And my colleague, open source. No, I’m actually asking you a question before that. Okay. I’m asking you, what do you understand as meta, in your role as meta, how do you define, how do you understand open AI? So I would reframe it more as open source AI rather than open AI. That’s fine. My competitors are open AI. would be delighted that I’m going to wax lyrical about them rather than ourselves. So, I understand it as a huge opportunity for the continent, first and foremost. I’m not going to get into the kind of technical parameters. Anyone who’s been in the conference over the last four days will understand there are closed models and open models, and they’re on a spectrum, some more closed than others, and meta has gone a long way since 2023 to put open source models, alarmer models in the hands of entrepreneurs, NGOs, governments, researchers, academics, anybody who wants to use open AI, open source AI, you’ve got me now, open source AI in a way to resolve their issues, meet their commercial goals, meet their public sector goals, all kinds of stuff. What I understand most is the opportunity in front of us. So, there was a chap here from one of our fellow tech companies, Microsoft, who put on this history of technological revolution. The cotton genie from the industrial revolution wasn’t there, but a number of things were, the combustion engine, a few other things, this timeline, and I’ve always been struck by this as I think about open source and the fact that we are putting it in the hands of people, the fact that it has, alarmer models have been downloaded more than a billion times now, that at each of those stages, the continent that is closest to my heart and beyond, the region I look after, Africa, Middle East and Turkey, probably weren’t at the forefront of those technical revolutions, whether it was land, labour, capital, technical devices in the hands of people, you and I have been working on mobile, telephony and internet for a very long time, and we’ve always felt to be behind the curve a little bit, with all due respect. This feels as though, despite the fact that major, big tech companies appear to be driving it, at least on behalf of Meta, this feels as though we are presenting an opportunity through open source AI to put those technical tools in the hands of those who probably most need them. This idea of of catching up through mobile telephone or catching up by putting the internet into people’s hands has worked to an extent. But because we’ve been behind the curve, we probably haven’t seen the impact we’d like to see. I think open source presents an incredible opportunity to do that. If we get, and as we’ll go on to speak about, policy and regulation and all that kind of stuff right, equity and safety and responsible AI and all of that kind of stuff right, I think it presents, first and foremost, a wonderful opportunity.

Moderator: Okay, so lots to come back to there. Tobias, why don’t you say, either you or what GIZ understands open AI to me.

Tobias Thiel: Yeah, first of all, thank you so much, Alison, for having me. It’s really a great pleasure to be part of this critical conversation on the benefits of open source AI. I mean, on your question, I would define it mostly as an opportunity for the continent. I think that’s really what captures for me and also the position of GIZ in terms of what’s the essence of open source AI, because I think we believe that for digital transformation to be truly inclusive and developmental, it really needs to be opened, needs to be openly sourced, and that means there needs to be open access, there needs to be open infrastructure, and there needs to be open innovation, and I think that really captures the spirit. It’s of course part of the principles for digital development, and I think an essential part really if we want to leverage the opportunities for the African continent and beyond.

Moderator: Thank you so much. Barbara, I’m going to ask you to help us. We’ve been told open source AI is an opportunity. I think both have qualified by saying they’re talking about open source AI, but perhaps you can help us by telling us you know, what is closed AI and what is open AI? Why should we be excited about open AI?

Barbara Glover: Yeah, thank you, Alison, and thank you to all the organizers for this event. I think that open source AI presents immense opportunities, at least for the African continent. We are looking at opportunities where our young innovators that we work with, who ordinarily would not be able to have the costs or the funds needed to, you know, get proprietary software when it comes to AI innovations, have that opportunity to still build AI models and tools. I’ll give you a good example of the work, for instance, we’re doing with Meta on the MSMEs and how we can utilize some of these e-learning platforms to offer our young innovators with the opportunity to also be in the AI ecosystem and develop the tools and software building on this. The closed AI option, closed source AI option would be one that would require a certain level of investment that for low or middle income countries will be something that, I mean, the average young innovator will not be able to afford. For instance, when you’re looking to some of the repositories in GitHub related to machine learning and AI, you would see a breadth of resources that are being useful to our young people that help them to develop the tools, whether it has to do with AI applications in healthcare, AI applications in African local languages, they have that opportunity or that leapfrog to be able to develop tools that are needed to address pressing challenges. So I do see open source AI as a great, great opportunity that we should be leveraging.

Moderator: Thanks so much, Barbara. Adil, maybe you can just build on that by telling us what role open AI or how AI fits into the African Union’s continental. AI strategy?

Speaker: Thank you very much, Alison. I think this is a very pertinent question, because you need to know what is it so that you can tackle it correctly. But if you don’t know the definition, so I’m wearing my policy hat, so I’m not going to give you a technical definition, but as my colleagues said, it’s an opportunity for Africa. Particularly, Africa is trying to achieve its Agenda 2063 Agenda. And we see it as an enabler to achieve this agenda in light of the recent AI strategy that we’ve developed. I think, as mentioned by our sister here, that the close AI is very expensive. For Africa, of course, we are resource limited, so if open AI is going to be good for Africa, we have a used dividend that we can use entrepreneurs who can be engaged in the open AI. There is transparency where community can work together to build systems, open AI systems, and I think that’s also an opportunity. We are aware of the challenges of open AI in terms of security, in terms of intellectual property issues, but I think the benefits outweigh these issues. And as mentioned by the first speaker, that I think the policy, our role is to make it, to provide the conducive environment for the African youth, the entrepreneurs, so that they can use it safely, and then we can scale up AI across the continent. I think also, it’s also an opportunity. So, I think we have a lot of opportunities to have, and Africa, we want, in terms of AI. So, it’s an African-led AI. I think we have islands of excellence across Africa, where countries that are excelling in certain aspects of open-source AI, we can leverage all this area of excellence in terms of infrastructure, in terms of capacity, in terms of systems and coding and all this technical stuff. So, together, I think we can build something really huge and fantastic towards the realization of the AI strategy. By the way, one of its goals is for the socio-economic development, and then enabling the attainment of the Agenda 2063, but also for inclusivity, so that everybody is going to be part of this journey, has got this undertaken, and I think since it’s cheap, it’s available, and I think that then Africa can make the best out of this technology and systems, so that we can leapfrog into the 21st century.

Moderator: Thank you. Adil, and in fact Tobias and Kojo too, and I think probably the way that Barbara expressed that, it’s kind of implied in what you said, but I mean, you’ve all spoken about these opportunities and you’ve all indicated that there are challenges in seeing these benefits. I mean, currently, as applied with some of these things, initiatives that you’re trying to do on the continent, who can be beneficiaries? We’ve got over half the population of the continent who are not connected to the internet. We’ve got a dearth, we’ve got fantastic, brilliant data scientists, they lack access to data, they lack access to computing power, and we still… certainly don’t have the capabilities in the public sector, large parts of the private sector, and certainly not in most of the SMMEs that you’re about to talk about, which are much higher percentages, 75% are actually above 75% probably are over informal sector. And even if they have access to the internet, they’re not using it for internet purposes and those things.

Speaker: They have a smart device and many of them don’t. So how do we get to this opportunity that you’ve been speaking about from where we are now? So just let Adil just finish that because he raised the inclusion thing and then I’m going to come to you. Yeah, I think it’s a very pertinent question, by the way. And it’s valid not only for AI, for everything we do in digital. We have this, we lack fundamentals. We have to work on the fundamentals on the digital space. But to me, what I see the opportunity, the issue we’re going to face when we do scaling, when you scale the systems, I think you will face this issue because you don’t have connectivity. But you can build the focal, the focus point of AI in the cities, like when you start building this center of excellences in the cities where you have connectivity. And I think, yes, we’re going to face the issue of scaling when we come to launching AI in the rural, remote areas and things like that. But of course, there is also, in terms of infrastructure, where we can deploy DPIs and maybe community-led solutions instead of having AI deployed, all the benefits of AI deployed. Like, I gave you example for instance, in the Agenda 2063, we are focusing on education, health, right? FinTech and agriculture, right? So we see the communities where like for instance, we need to focus more on agriculture. So then we tailor our solution, conductivity is going to be part of the equation. But then we are going to be solution-based. We talk to our international partner, to the government, to the private sectors, researchers, and then interpreters, and then even the company, big tech company, so that all together we join hands. And I think part of the deployment of these solutions is going to be the fundamentals, like the issue you mentioned, which is, of course, we are facing across the board. But these fundamentals are going to be embedded in the solution that will provide it, like if it’s education, if it’s health services, that’s where it can provide, and so forth. I know it’s not easy, but of course, this fundamental, of course, is going to be a big issue. I think the questions in front of us, whether to first work on the fundamentals, and then do AI, or do them in parallel. And I think we don’t have options. We have to do everything at the same time, which is always our challenge in every area.

Moderator: But just before Kojo comes in, so we know that the major problem for digital inclusion is not just connectivity. We’ve got most of the continent covered, and we’ve still got these very, very low penetration rates, especially in these developed countries. So this really presents a skill in terms of having the skills in order to become representative in these databases and things like that. So I think it’s a bigger challenge than connectivity. It’s this big human development challenge that we have to address at the same time. And maybe AI can help us with that. But you’re speaking specifically about the digital economy and using AI to catalyze the digital economy in this session. So, Kojo, why don’t you tell us what Mette’s vision is in this regard? So I’ll do that as I attempt to answer that other question as well, which I think you rightfully put, because all four panelists, and I wish there was a fifth, who was less enthusiastic about open-source AI, because I think that might make an even more interesting conversation.

Kojo Boakye: But I think the question you then put was, that I felt you were putting, was, oh, there’s great opportunities in front of us. How do we make sure the millions of people who aren’t yet, that it happens? And I think Adele’s right. There needs to be this continued investment in connectivity. I still believe that’s an issue. You might talk about mobile coverage being extensive and people- I think it’s the human skills to use the internet. I agree, so I’m gonna touch on that as well. I still think there’s work we need to do, and as Mette, we have done, on subsea cables and the capacity to take things internationally and across the continent that’s so important if we need to get the price point to where it is completely affordable. And my colleagues, I sit on the board of the GDIP with Sonia George and many others that you know, which are still looking at affordability and inclusion, and it’s still a challenge. And I think investments like the 2Africa cable that Mette has invested in, the first cable to link West to East Africa, or if you’re East African, East to West Africa, whatever way you like, around circumventing or going around Southern Africa the first time helps with that, as do investments by other companies who, for this purpose, will remain nameless, but you can, you know, Google have done some fantastic work there as well, so not so nameless. You then need to start thinking about the kind of investment. It’s not a build and they will come. It’s been downloaded a billion times, but we’ve also done work since 2023 to catalyze usage. So our Lama. Impact Grants were a fantastic initiative launched in 2023. Thousands of applications for these grants, between 50 and 200,000, and two of the winners of that Impact Grant were African players, Digital Green and Jacaranda Health, who have gone on to do fantastic work to resolve some of the issues that are most important. So Digital Green does kind of information to farmers through SMS in rural areas built on LAMA, helping revolutionize yields and farming outcomes. Jacaranda Health does work with maternal health, with women having children post and pre, and improving outcomes. Those sorts of things help. At the same time, our team has just launched, closing tomorrow, if you have time do apply, the Accelerator Grant for LAMA as well, which is giving NGOs, commercial players, mentors, or potential commercial players, mentors and support to use these technologies to meet these goals. And I think that’s the part of the way to go in there. Continued investment in connectivity, efforts to bring down the cost of devices. I saw Dr. Ndemo in the halls earlier and I still call him a mentor of mine, in part because of the effort he made in Kenya to take the price of the tax on mobile devices off. And then you see penetration, not rocket science, penetration jumped by 200,000. We need vision initiatives like that. You need training programs like the ones I’ve mentioned, Impact Grants, Accelerator Grants. But at the same time, something that we’re beginning to focus on is how do you help governments? I travel across Africa, Middle East, and Turkey. I see the government of Turkey are doing things, UAE and KSA doing amazing things, Ethiopia doing amazing things. Bosun, the minister from Nigeria was here. They’ve bought chips themselves and are starting to use LAMA as well. How can companies like ours work with governments to get them to use it as well? And I think that creates a multiplier effect. If you have countries in which, some may argue, the social contract between government and people has broken down completely. When you give governments the opportunity through LLAMA to try and revolutionize judiciary, education systems, health systems, et cetera, those people who may not be able to use LLAMA, who don’t want to code on LLAMA but just want the benefit of it, have access to some of the benefits of it. Finally, Alison, because I know I’ve taken a bit of time. The other thing that we feel is super important, in part to give people access, in part in my view, to dispel this concern about the risks of AI, which are valid and we need to work on them, is to get it on people’s devices. Amongst the first countries in the world to see meta AI, and I know there are other apps that many people use, but meta AI on people’s devices were places like Ghana and Nigeria, where we put it, embedded it in WhatsApp and Facebook to give people access and to give people an opportunity to use what is, at the moment, a very basic form of AI. If you ask, you know, what’s Kojo Bwachi talking about today? It might.

Moderator: I’m not that famous, but it might. In fact, it even comes, even if you don’t ask it, you’ll get it anyway.

Kojo Boakye: We’re trying to ward against that, hopefully you can lock it down, but giving people access to it is going to be super important in helping them resolve issues. If you’re in Ghana or you’re in Nigeria, where uptake has been fantastic and retention has been fantastic, we’re seeing young Africans, young, old, middle-aged Africans, using AI and clearly, clearly gaining benefit from having access as well. So, it’s those investments, it’s that vision, it’s executing along the whole, if you want to call it a value chain, but along every area that’s important to making sure the opportunities that all four of us have spoken about actually come to fruition.

Moderator: So, Kojo, I’m going to go very quickly after this to hear some more of the possibilities, use cases, interesting things that we can catalyse AI, we can get AI to catalyse the economy, but the things you’ve been speaking about would enable the economy, get people actually using this technology in their daily lives and those who are connected and online and meaningful. online. But, you know, a lot of the critique has been about the embedding of these systems in our local systems. And so, you know, we’ve become locked into these. I mean, Adele was speaking about, you know, deployment of DPIs and the obviously enormous potential for AI in producing, you know, making those more effective, efficient. But the danger, of course, there is that although people, you know, data scientists who are using LLAMA say that they’ve got, there’s a kind of front-end openness to it, but there’s actually not a back-end openness to this that would allow them to be really developing, you know, stuff that they can build on independently and pass on. So, just to respond, because I think that is what some people would regard as a risk, is that you actually, you know, big tech systems

Kojo Boakye: are now, you know, embedded in your local public sector, your social systems, everything else. So, if I understand, a couple of things. If I understand your question correctly, I think we want to give people control over their use of meta-AI within their apps. So, users. Yeah, users. Data scientists. When it comes to data scientists and researchers who download and use LLAMA, I think they will argue that the weights we share, and I think with LLAMA4, I wish I was an engineer, but with LLAMA4 we’ve shared more information than ever before in terms of transparency, one of the major benefits of open source, than ever before would give them control. The ability to fine-tune, the ability to assess the data that we share, that we use to train the model is pretty open. I think I started my intervention with the first question by suggesting there is no binary open and closed. I think there are models that will be deemed closed, far more closed. We believe our model is open, fine-tunable, trainable, and adaptable, and I think in part the feedback we get from the community, some better than others most of the time, but the feedback we got from LLAMA4 wasn’t great initially, but has improved. The feedback we get from the community not only helps us create a better model next time, but is also evidence of the fact that the data science community, the people who don’t just want to use AI but want to train it and fine tune it and all those kind of things, are actually see a lot of

Moderator: benefit from doing it. So I push back slightly on that. Okay, we’ll come back. Let’s give Tobias a chance. Tobias, you, I mean, maybe just also because I saw you were nodding to some of these. What is GIZ doing in terms of developing more sort of organic, maybe indigenous, or, you know, enabling people to be able to respond to these open source AI options that are becoming available?

Tobias Thiel: Yeah, so I would say in the first place, GIZ, of course, approaches the issue of AI primarily from a developmental angle and not from a commercial one. So as a development agency, we don’t pursue any economic profit sort of objectives. And we have quite a vast digital portfolio on the continent. I mean, whether at the continental level with our friends from the African Union, whether with the regional economic communities, bilaterally down to the community level. And perhaps one interesting case that we have, we have an initiative called Fair Forward, artificial intelligence for all. And there we’re basically working to really build inclusive AI ecosystems. So we partner up there with a lot of local organizations for initiatives like the Lacuna Fund or the Open for Good Alliance to address one of the, I would say, major AI gaps in the global south, which is data. And so what we do, we help local researchers and innovators to use, create high level or high quality labeled and localized data sets. Because, I mean, at the end of the day, the model, the data is the backbone that really matters and the idea is then to use these data sets to tackle real-world challenges in sectors such as agriculture, energy, education, health, I mean basically across the board of the sectors that we work and maybe to make it a bit more tangible. I think a few examples of how we do that where we really try to create a value added for local communities is there for example AI is used through picture recognition for local farmers where you can take a picture of a plant and diagnose sort of pests and develop adequate responses or satellite data that’s being analyzed by AI to identify good places to build local renewable energy and in communities that are not electrified which then will also have a positive impact on the community overall. Perhaps another important part of this program, of this initiative is language. Language is really a core pillar and there we work with Mozilla’s common voice and we basically support the co-creation of open speech data sets in also languages that usually are not covered. I mean you have, I mean we know there are thousands of different languages on the continent so one particular case we work on Swahili and Luganda, both Bantu languages that are under covered basically in commercial AI systems to make them more inclusive. We also work with a project called AI Made in Africa where we try to make existing data available to researchers and innovators so that they can really create solutions that African people can use. So, I think it’s important to recognize that there’s a bigger picture here, and I think that what we do in terms of access is important, but it’s also not enough. I mean, I think when we see what has been discussed here in the various panels that we had this week, there are really a lot of challenges, and you mentioned also initially that we can’t do only one thing, and the way we do it, we can’t do it all at once. So, I think it’s important to recognize that there’s a bigger picture here, and I think it’s important to recognize that we can’t do only one thing and wait for the other, but we need to tackle these challenges simultaneously, and I think beyond data access, I think there’s also an issue in terms of digital literacy. I think we need to build stronger digital literacy, tailored capacity building, and that both for regulators, but also for innovators. I also believe that it’s important to have a proper governance structure, and I think that’s also where we need to have a proper maintenance strategy, that they are improved, maintained, and secured in the long term, and I think that’s also where open source governance can play an important role. And lastly, and I think that’s really perhaps one of the probably the most challenging point, is that I think we need to have a good governance structure in terms of the data-driven infrastructure, and that’s because that’s really the most resource-intensive part, especially in underserved regions. So, yeah, I think we have to be also a bit realistic in terms of what open AI can, open source AI can do. I think it’s not a silver bullet either. I think we have to be a bit realistic in terms of what open AI can do. I think we have to be a bit more leapfrogging, but at the same time, if we’re really realistic, I think we have to recognize that it cannot fully fix the inequalities, the geoeconomic inequalities between the global north and south. There’s simply some hard facts in terms of GDP, in terms of market sizes and so on that I’m sure you know much better than me, but what kind of considerations go into that. But despite this, I really believe that open source AI can make a difference if we connect the dots. I think it’s the best shot we have to really support leapfrogging for the continent.

Moderator: Thanks very much, Tobias. Adil, maybe just sort of building on that, we’ve got the AI continental strategy in place and you’ve said that the open AI or open source AI is a principle behind it, but a lot of the challenges at the moment lie around the implementation of that. I think we have a scattering of different AI blueprints, things that have come at different times doing different things, representing different interests. A lot of them other than the continental strategy, foreign funded, foreign driven, different agendas on those. How do we ensure that we begin to get the framework, and I think with the continental data policy framework, we do have some of the governance underpinnings for that, but how do we extend that implementation so that this enabling environment that can manage the risk, that can enable local entrepreneurs, local scientists to use open source in the public interest or for? Yeah, sure. I think we actually, we are working on another framework for open source data. And it’s going to be reviewed and considered

Speaker: by the minister by the end of the year, early November. So, and also, we are also drafting an implementation strategy at M&E, also part of this implementation strategy. Same is going to be also looked at during the ministerial meeting end of this year. So I think we. So, we are not, at the Commission, we are not addressing open AI as such, but they are doing all the homework, all the enabling environment for open AI, so that when it’s being recognized and it’s being launched, then everything is there, fundamental is there, to be able to make sure that we launch it safely and also with all the barriers removed. I think, originally, I think you raised a very good point about the fact that we need to make sure that when we do open AI, AI in general, and specifically open source AI, we need to make sure that, you know, the AU Agenda 2063 and the AI strategy is talking about Africa, you know, Africa first, and all that, and the sovereignty of Africa. I think we need to make sure that these solutions are not going to be, to cause us more problem than benefits. I think open AI give us this opportunity, so that we are going to be doing things within Africa, and making sure that, you know, we don’t repeat the experience of, you know, the social media and all that, because, you know, we don’t have control, we didn’t have control over, we don’t have control over this, all this new technology coming to Africa. And I think, for me, the metric would be if the African government and the African communities, the African families, they can control what is it, what part of AI they want to take, and what part of AI they don’t want to consider, because otherwise, then you lose your sovereignty. And I think one way is to do that. do it at home, even though it’s going to be difficult, but you do have to do it locally so that he can control what aspect of AI, because there is also a cultural and tradition aspect of it, because the culture in the West is not the same as the culture in Africa. So we just need to make sure that when we build it, so we build it the way that is going to satisfy all this requirement in terms of the cultural and in terms of tradition and so forth. So I think for me, as I said, the metric, like in my family, do I have control what kind of AI systems or applications that is going to be in my family or we don’t have? As a country, do I have control what aspect of AI I want to use and what aspect of AI I don’t want to use? If I feel that, I think we failed in terms of the Africa we want and all the technology we want. Okay, maybe it’s a question we can get an answer to from the floor. Do we have control from countries, from a sovereignty point of view, do we have control over what AI is coming into our countries?

Moderator: And what’s the risk of control? Sometimes there’s a couple of things. Yeah, absolutely. There’s a couple of things. Can we just get Barbara in on that, because we’ll come back to that, because I think we actually are getting onto some of the risks and some of the risks of control, which you’ve identified as well. Barbara, you’re working with scientists and maybe scientists who are also working as entrepreneurs actually with startups and various things. So just speak about, I mean, is open AI seen, or open source AI seen as this sort of fabulous opportunity? Because I think a lot of these, you’ve spoken about them yourself with your partnerships with Meta and how people have responded to these. What feedback do you get, because I do know you’ve dealt with data scientists who are often feeling sort of squeezed out of, for example, local language, large local language modelling, because, you know, the data and that translation, all the technology already exists, you know, with Google or, you know, AWS, or somebody can come in and do something, you know, very quickly and that actually, local scientists could do if they were given the opportunities and things to do that. So just in the context of open source AI, just speak about some of the positive and possibly negative things if you do come across them. Yeah, so thank you for that. I think that I’ll start with the negative and end with the good.

Barbara Glover: In terms of open AI, I think the, one of the raising concerns has been trust, just trust in the open AI system and the need for more understanding of what that means and getting more young innovators to be interested, or I mean the scientists and the data scientists that we work with. And I think also open AI cannot thrive if there isn’t effective data harmonization or policies even that encourage sharing of data. So, well, I wouldn’t want to name, but there are some innovators, for instance, that have approached certain bodies to get some data that can feed into the work that they are doing. But even though there are policies that support this, they are not able to access that data because the data harmonization frameworks are not being implemented as effectively as it should be. And so then they get frustrated in that there is open source data on paper, but it’s not being seen on ground. But one thing that highlights the positives of us looking more and more at open source AI approach. is the fact that when you, I don’t know how many people in the room read the article that New York Times released, I think a week ago, looking at the location of data centers across the globe. So you see so many of the places had blue dots all over, and then you come to Africa, and it’s just a street down there in South Africa. These are realities. So Africa probably won’t be a leader in high-performance computing. It won’t be a leader, perhaps, in data center location from what is currently being seen, or even in cloud infrastructure. So we need to look then at what is our niche? What is it that we can do better? And I think that these are the opportunities that open-source AI provides. It provides our young people with that opportunity to create a niche for Africa in this space, whether it’s in, we have over 60% young population, and these are talented people that are using these IT skills to dance on TikTok. How can they leverage some of that in using it to address real challenges that are happening on the continent? And those are the opportunities that I believe that open-source AI provides that we should not miss. We were just wrapping up, for instance, our AI and Health Conference here in Cairo, and one of the discussions was on us trying to avoid boiling the ocean, trying to avoid, you know, I have seen more and more in these AI conversations on the continent, you know, African solutions, African people, and the messaging around that. But we also want to look at how real, realistic some of these ideas are. What are the, what is it that we can leverage on Africa, as Africa, what kind of unique niche we have that we can utilize more effectively? And I believe that’s where this open-source structure provides that opportunity. Young people don’t have to wait for government to catch up on infrastructure, to catch up on supercomputers and all these issues. before they can, you know, harness the world of air. I think it’s a great opportunity that we should not be missing. And that’s one thing that as African Development Agency, we have tried to bring to conversations when we meet with local innovators. And what is it that you can do with the now? You know, realistically as Africa, we guess we have dreams of how it should look like, but realistically in the settings, what is it that you can leverage? I was listening to a presentation again by Meta where they were showing what has been done, I think in Tanzania, where they’re using the Llama platform for malaria elimination efforts. And then the presenter was saying, you know, they didn’t even approach Meta to do this. They went ahead, got into the platform, accessed this and they use it for great stuff. And I thought, yeah, this is what we want to see our young people do. And this is what open source AI provides for us. And we should be leveraging these things, address your own problems without necessarily waiting for the great infrastructure that, you know, and the great data centers that other countries may have. And that we unfortunately may not have as a continent. We are shaping that conversation more and more. I’m happy to hear from our partner, our colleagues from the African Union Commission that you developed in the open source framework. I think that this is timely. And I think as a development agency, we’d be happy to shape that inputs and be able to ensure that our young people are leveraging this within the African region strength to be able to harness AI effectively.

Moderator: Barbara, thanks so much for that. We’re going to come to some of the challenges, but I’m also aware of the time. And I know you’ve also got exciting initiatives that you want to speak about that are going to catalyze the digital economy in Africa. But before we, so you can, we can go into the risks and straight onto the opportunities that you want to speak about. But I thought if we could quickly just see if there’s some questions from the room before we go into the next section. Or Christelle, if we. we’ve got any questions online. I think there is a question there. Thank you. Thank you very much. My name is Osei Keja from Ghana. I would like to ask two questions.

Audience: Okay, I’ll keep, I’ll just answer one. The path of AI is sort of path of riches when it comes to Africa. And distinguished panelists have articulated the challenges and all that. But I’m curious, how can Africa, we advance AI and other emerging technologies at a fraction of the billions, the trillions that have been spent in the global North while still ensuring equity and the long-term health of all these new tools or say new technologies. What current models can be replicated? Okay, let me keep this. Thank you so much.

Moderator: That’s a great question. And hopefully our panel can answer that as we move on to that next section. Because I think the question of ensuring, you know, equities, safety, a lot of the things that you were speaking about, really speak to this enabling environment we want. So we do, you know, we do want to enable people to be able to use the technologies. We do want to enable investments in these technologies. And, you know, it doesn’t all have to be large language modeling and generative AI. There are lots of other local AI models and systems that we can be developing. But what are the governance, maybe not regulation, but I think probably regulation and governance frameworks that we need that are both. protective, but also enabling, that will ensure people have access to open source systems or to be able to deploy those for other purposes. Kojo, did you want to just, because you started out talking about the control, but…

Kojo Boakye: I think it’s a mix of things, and I was grateful for Osei’s question about equity. I wasn’t sure whether the question was about when the world is, you know, tech players are investing 65 billion as we are this year, how do you have equity and all kinds of stuff. I think we’ve kind of spoken to the fact that open source, at this point in time, with regard to AI and the opportunity in front of us, is probably the only way to level the playing field. Certainly at Meta, that’s how we see it, by putting models into people’s hands, and they don’t need to invest 65 billion to create a model that’s specific to Komasi, where both Osei and I are from, that you can fine tune and build something local for that region to meet that issue. I think that’s going to go some way to doing it. I don’t think you’re ever going to be at a stage, or in the near future, not ever, I don’t think you’re going to be at a stage in the near future where an African company is going to say here’s 65 billion for 2026 or 2027 to invest in compute. So I don’t know if you’re going to get equality in that kind of way, I’ll stress that. Although I pray and earnestly believe God will answer, at some point in the near future we will have that. On the issue of levelling the playing field, I’ll go back to my bread and butter, policy and regulation, I think there are some obvious and key things that we can do. First of all, we have to be included in the conversation, and include ourselves in the conversation. They are ongoing. I was speaking at UNGA in 2023, at Secretary Blinken’s event at the time, and we were speaking about everything from responsible AI, the need for red teaming and benchmarking and all that. of that kind of stuff, which is a conversation that has continued, and that speaks to some of the ways we might drive equity, inclusion in terms of language and all kinds of stuff, which META is working on with No Language Left Behind and Africa Next Voice with the Gates Commission, all kinds of stuff, to drive increased equity. But we need to have policymakers included. The reason why I was asking about control, which I hope is linked to this a bit, Alison, is that control can be misinterpreted. I’m a historian. My first degree is African history. My second degree was development. My master’s was development studies, political economy, looking at the history of Africa, how it developed and all kinds of stuff. I appreciate we have to learn from the past and past mistakes, and Adil spoke to the concerns around social media, and that should inform how we look at AI, completely get it. I respect the office and every government. My view is that they should retain sovereignty and maintain it. I’m on record as saying that now, but I do have concerns when people start speaking about control, because control, for me, negates in some ways or may, in some jurisdictions and with some governments, negate what I have a bias and believe in, which is the ingenuity, innate ingenuity and brilliance of young Africans, who, if we put these tools in the hands of people right now, will create the solutions. And I’ve spoken to some of those, whether that be Vax Lama in Nigeria or Jacaranda Health or Fundamate in South Africa, now serving three million people. These are ideas that were born in the minds of Africans and developed and executed using Lama and AI, and those things, despite having some semblance of control in their policy and regulatory environments, have gone on to do amazing things. So I get it, and I get governments want to understand what’s happening and have some control and put policy and regulation in, but I do have, I do say we should have caution

Moderator: if we want to grab the full opportunity that this presents. I think we need to have some caution around it. So perhaps just to quickly pick that up, and then we’re coming back to the clever things that have begun to be offered to drive, to catalyse the economy, but I think when we were talking about control… We’re talking about controlling big tech, not the small tech. But just to take that idea, because I think it is an important one. I think where you don’t have democratic human rights, people-centered frameworks, and you’ve got access to very powerful technology, and you control that access, very bad things can happen. But assuming you’re within normative frameworks, international frameworks that we have, and I think particularly in Africa and the work that we are doing within the G20, our first African G20, I think there’s a lot of recognition that, in fact, you do not get equity. You do not get innovation unless you create the conditions for innovation. And unless you actually regulate access to data, you’re not going to just get it. Or you’re not going to get necessarily what you need. Unless you access, unless you regulate mechanisms to open up compute power, these kinds of things, you’re not going to get it. So I think there’s a strong argument. I’m completely with you on political control of this. It’s problematic, and it needs constraints and safety nets and all sorts of things. But in order to make this work in Africa, we do have to regulate the conditions that will protect people’s homes. But as I said, importantly for this session, but also for Africa, the enabling conditions for this to drive and catalyze the digital economy that we’ve spoken about. So Adil, just in the continental framework, there’s reference to the kinds of governance that one would require for AI on the continent. Yeah, I think maybe also I’m going to link it to the answer to the question from the floor. Yes, I think I am optimistic, unlike the questioner,

Speaker: because I think this is an opportunity. Africa can make the best of it. have people wishing, or not wanting to be inspired. We haven’t found that way. Or some of them are voting. I think there is also that way that is the way people are thinking, in terms of that there are the ways sort of towards the commitment, the way people are thinking, and the way that we are doing it. I think we need to move, we need to have a holistic approach to capacity building. You start from even early education, PhD, and then just make sure that we have it from A to Z. Everybody is well-capacitated. At least the community is aware of how many people can accomplish something. There, all that’s what we want to do. I’m a strong believer in the minds and skills of African, and they’re predominantly young people. Young people can do wonder if you give them the tools, so I believe that there are more than that. We need young people. You don’t have to work with ten pounders to reach America. The other thing also, it’s true, if you work individually, if you don’t go in countries, you don’t go very far. We have to work collectively on AI, and that’s why I talk about the sovereignty aspect. I’m not talking about the political control, whatever. You just do things that you do. that you, because we want the Africa we want, right? So we need to work towards that. This is the Africa we want. Like for instance, we are doing AI for development. We are not doing a lethal, you know, we’re not going to pursue lethal use of AI. But if you leave it open, it will come to you. And then the world is going to be a disaster. It’s going to be uncontrollable, the situation. So that’s why, that’s what I meant when I said, so you have to have some control. You need to do development, you need to have inclusion, you need to have all this, you know, objective that we have under our EU Agenda 23. So to do that, you need to make sure that is in-house, you do it in-house. Even though you will struggle in the beginning, but I think over time, you will get there. Yeah. So Tobias, I mean, some of the issues we’re talking now, this kind of collective and harmonized frameworks and things are the things that would provide Africa with the kind of economies of scale and scope that these data and AI kind of markets need.

Moderator: And GIZ has been working with African Union and institutions, regional economic communities on the continent to achieve this. But just, you know, how far are we? And what are the kind of urgent issues now to move that forward? I’ll just give you a very quick example. You know, with the African Union Data Policy Framework, we had a whole lot of fantastic principles, including the idea of, you know, creating public value, because there was all this public data that we needed to get out there and, you know, make available to people. But which, in order to deal with some of these enormous, you know, competition challenges and asymmetries of, you know, from big tech providers or tech providers, that we would want, you know, preferential access to open data, to public data, and for local entrepreneurs and that kind of thing before it gets gobbled up. But that was completely, you know, prevented by the African Continental Free Trade Area digital protocols, which are cut and paste from WTO, and so, you know, trade always trumps everything else. So, you know, just in terms of you’re working on the continent, you’ve been supporting these initiatives. There obviously is still a lack of coordination at some level, you know, around all African Union initiatives, but we’re still in our kind of silos, our digital sector silo and our trade silo. You know, what are the challenges there from supporting this, you know, Agenda 2063, etc., that we all committed to? Yeah, I mean, maybe before I answer your question, just allow me a few sentences to react on what my peers on the panel said with regards to regulation.

Tobias Thiel: I mean, I’m a strong believer that it’s really not about control. I think we have too much of a – there’s too much danger in calibrating this wrongly, especially when you see the different types of leadership that you have on the continent or beyond. And I think in that sense, it’s – I think it would be easy for us to agree that we need a fair and enabling and risk-conscious and equitable regulation. I think for me, the more important question is actually what is the process to get there? And I think what’s really essential there is the need to have really collaborative partnerships where you bring the variety of different perspectives together, and that’s the public sector or regional institutions like the African Union, that’s the private sector, that’s academia, that’s civil society, that should also be actually users and beneficiaries, perspectives, youth, women. So I think there’s – I think if we ensure that we have the right process, I think that would really be the way to help us calibrate regulation in a – people like to adopt these policies but implementing them in a proper way and I think, in that sense, the African union plays a really important role as a continental norm setter. And I think some of the frameworks you mentioned, data policy framework, I think was an extremely important prerequisite. And I think it’s a very important prerequisite, and I think it’s a very important prerequisite. And I think this is, I mean, the program that we have to help domesticize, domesticate, the AU data policy framework has currently requests from 35 different African states to help put this framework, which is unless it’s implemented, it’s a document that has no value. And so, I think, it’s a very important prerequisite for what we’re doing in Africa, and then it becomes again a question of resources. I think another important aspect also to calibrate regulation well is I think also one of bargaining power. I mean, I had in the context of this week, I had a number of bilaterals with a number of African states which cost 30. Sheretty Oassima belongs to four countries, that cost 83 million inhabitants, where you have not more than 300,000 connected users and even that is probably an overcount, what kind of bargaining power do you have? And that’s where I think it’s also important that AU member states really get together, whether under the framework of the AU, whether under the framework of that set up and even if it’s under some certain Patriotism which the European Union does in a global context when it comes to negotiating. So, I think those are some realities that we really need to

Moderator: take into consideration. I’m sorry, I’m not going to return to the question that I asked after that, because we’re running out of time, and I want to see if there’s an online question from Christelle. Christelle, can you hear me, and can I hear you? Yes.

Christelle Onana: There is a question online. How can open source AI communities interact with each other, and how can open source AI communities in the African continent be supported and sustained? This is from Ghana Hub, IGF. Great. Thank you, Christelle. Is there another question? I think somebody was attempting to ask a question. There’s a gentleman here. Okay.

Moderator: Christelle, is there another question online? It was online. Okay, thank you. We’ve got another question in the room, and then, Barbara, we’re going to start with you so that you can lead off the answering questions. Yeah. Generally, to the panellists, I think the question of Africa’s position on AI is something that Adil has pointed out.

Audience: They’re quite proactive in establishing, and as we know, the WSIS plus review process is coming up. Africa has taken a proactive step, even before this particular process comes up, to issue various frameworks, as you’ve actually pointed out. The AUDPF, the position paper on the GDC, the African contribution, the AI strategy. Are we seeing, in the future, more and more of this kind of approach? Are we seeing more and more of this kind of strategy? Are we seeing, in the future, a position where continentally or at member state level, a departure for the adoption of open source AI to boost the accessibility of the community? I think that’s a good question. I think that’s a good question. to the digital economy, is it something that you and the panelists would support, even if it’s in contravention or it’s in opposition to Africa’s position on the use or regulation of AI? So, that’s my question to you, because as we are quick to see the adoption of an open science of AI as an opportunity, we also need to look at the sustainability from a policy standpoint, the adaptability, what that really means when it comes to member states enforcing their interests. And I think you’ve touched on this question a bit more, that’s why I sat down, but the question then arises, you know, with the WSIS plus 20 process coming up and all these other parallel processes globally, continentally, do we see a situation where Africa has to then concede and take up open source AI for economic benefit in contravention to its policy standpoint? Thanks. Before you leave, can you just clarify the position that you’re saying is different from Africa’s position? In WSIS, which specific position are you speaking of? No, no, no, I’m not saying there’s a different position, I’m just saying the process is coming up and one of the elements that are going to be discussed is the AI and data governance as a bit of, yeah. Okay, so where should we be, where should we stand on that? Exactly. Yeah, okay.

Moderator: Thank you. Barbara, if you would answer, take any question you’d like, but I think the first question particularly is geared to the work that NEPAD’s doing in this area and trying to enable access to open AI.

Barbara Glover: Yes, thank you for that. I think our approach has been to look at open source AI, not in just the idea of training code. just for the purpose of training coders or getting people into AI just for the purpose of it. But we are trying to look at a problem-solving approach to some of these communities. For instance, with our Youth in AI Robotics competition, where we have these innovators grouped into categories. Each of these categories are addressing a problem, whether it’s healthcare, education, mining industry challenges, whatever pressing issues we have, as aligned to Agenda 2063 and then the Science, Technology, and Innovation Strategy for Africa. So we are framing those community outputs into tangible impacts. And I think that’s how we can ensure that these communities become sustainable because you are solving real-time problems. And it’s also when it comes to education systems that build up these communities, where we want to look at domain-specific education. So you’re not just giving broad AI and then the person is a master of everything, but how do we contextualize some of these communities to provide education in specific sectors, whether, again, in health or agriculture education, where AI is seen as, I think that’s what one of the panelists said, as a roadmap to development and not just the new technology. Cutting away from the buzz and just looking at practically how we can use it to address many of these problems. We are doing that already with our innovators in AI and Robotics competition. This year, we just rounded up that competition. We had over 2,000 submissions. And the idea is that once we profile them and look at what levels they are, whether they are prototypes or those that are in the field, we’ll then also move them into an accelerator hub to look at how we can get them to look at some of these innovations and solutions in a more domain-specific approach, addressing problems that have been. are identified as challenges on the continent

Moderator: from our policy makers and our decision makers. Thank you so much, Barbara. And I think this nuancing that Barbara is trying to get to this discussion, what AI are we talking about? Who can actually use this? What is the purpose of it? The sectoral applications of it is really going to help us move this ahead. What is pragmatic to do here? What is the potential there? And perhaps something we haven’t actually discussed very much is we’ve spoken a lot about regulation. But within regulation, of course, great utility and incentives. So maybe also just as you wrap up, to just speak about from your perspectives where you are what those might be, but to answer the more political question on where Africa’s position on this should be within the context of this action within the World Summit of Information Society.

Kojo Boakye: So I’m going to try and attempt to answer both questions. One, my view has been for at least 35 years that closed mouths don’t get fed. So on the question about the community and how they can leverage open source AI more, I think there is first a need for the community to come together. And there are so many across this continent that are already doing that. But grow those communities is important. And then for those communities to access the kind of programs that Meta puts out, whether it be the impact grants I’ve spoken about, the accelerator, as well as other companies put out. Every time I go on LinkedIn, I see a program being provided by Google, by Microsoft, by other smaller players involved in this space. So I think there’s a lot out there for the community and we should continue trying to invest in it and grow that piece. The gentleman’s question, I didn’t quite catch, Alison. Is there this decision we’ll have to make on whether we use open source AI and sacrifice something else is kind of what I got. I speak to ministers, heads of state, regulators across the continent. and we started off with this panel by all in unison saying this is a huge opportunity that we need to seize. As I think about kind of how governments are approaching this for the most part, they’re all trying to seize this opportunity and even the AU as a body, an august body for kind of coordinating and trying to drive the harmonization is doing it. We’ve done some work with Aldenepa, the AU development agency, to develop a Lama-based app to help small and medium-sized businesses. The core of this session was about how do we grow the economy to help them understand how they move their businesses from, how they open businesses in countries across borders. All of that speaks to this, and I think that’s a ringing endorsement from the African Union as well as governments to kind of the opportunity in front of us. I’m not sure at this stage about getting more granular. I see a hard decision to make for the continent at the moment.

Moderator: Maybe I’m not understanding the question properly, but that’s how I see it. So I think part of the question was just, I think there’s this big commitment, a lot of enthusiasm, but I think it’s not unconditional, that it needs certain conditions to be enabling, it needs certain conditions to not perpetuate, amplify inequalities, et cetera, et cetera. So I think that was sort of the question. Tobias, we are time up. I’m gonna give you a minute and I’m gonna give Adeel a minute and we’ll be talking to ourselves and probably lose our sound as well. Okay, so I tried to cope with the challenge of answering seven questions with zero seconds of time. Yes, something like that. No, I mean, maybe just a few selected points.

Tobias Thiel: I mean, first of all, I would find it presumptuous to lay out what I believe should be the African approach to AI regulation, implementation, and so on. I think I would hope for it to be an African-led process that’s truly African, that’s truly inclusive, that’s… I think, I think that’s the key element of the AI, the AI that’s enabling business-friendly, but also risk-conscious. I think those are, from my perspective, the key elements, but I think the local ownership is really perhaps the most important part. So, yeah, on that, the other thing is that I’m really convinced that the future of AI in Africa will not be built in silos. I think that’s the key element. So, I think that’s the key element. I think that’s the key element, and I think that’s the key element from collaboration and from speaking now from the perspective of GIZ. I mean, we see ourselves very much, we’re not a tech, primarily a tech organisation. We see ourselves as a convener, as an enabler, and in that sense, we’re happy to play our role also to provide a bit of specific information for h application services, the

Speaker: difference between a single sacrifice application and we might see others do the same. Consτο we want AI for peaceful development, and we are also advocating for ethical use of AI. So, that does have, this has been a very important part of what we are trying to do, and we are also advocating for the inclusion of AI on the WSIS plus 20, and we are going to articulate the same position. So, also, we are also advocating for a seat for Africa when it comes to AI policy-making and AI governance. I think it’s very important. I think the membership of the G20 that you mentioned, I think this is a step forward, but I think it’s important to make sure that the members of the G20 are also part of the G20, and we are left behind in many of the decision-making in the previous century, but I think we need to position ourselves more when it comes to the one, in the G20, to make sure that we are not left behind in the decision-making in the previous century, but I think we need to position ourselves more when it comes to the G20. global decision-making when it comes to AI, I think it’s very crucial that we are part of this decision-making process, and this is what we are going to articulate when we develop our WSIS plus 10 position, common, African common position on WSIS plus 10. 20, sorry.

Moderator: Press 20. Thank you so much, Adil, and thanks very much to our panelists and our audience who’ve sat with us for so long, it’s such an end of a long day, and to Kristel and people online, thank you so much for staying with us, and finally to our audio and crew people here for indulging us with a few extra minutes. Thank you so much. And to you for the excellent moderation. Thank you.

Agreement points

Open source AI represents a significant opportunity for Africa

Speakers

– Kojo Boakye
– Tobias Thiel
– Barbara Glover
– Speaker

Arguments

Open source AI represents a huge opportunity for Africa to catch up technologically, putting technical tools in the hands of those who need them most

Open source AI requires open access, open infrastructure, and open innovation for truly inclusive digital transformation

Open source AI provides opportunities for young innovators who cannot afford proprietary software to build AI models and tools

Open source AI is an enabler for achieving Africa’s Agenda 2063, offering transparency and community collaboration while being cost-effective for resource-limited Africa

Summary

All speakers unanimously view open source AI as a transformative opportunity for Africa, emphasizing its potential to democratize access to AI technology, enable leapfrogging in technological development, and support continental development goals while being cost-effective for resource-constrained environments.

Topics

Development | Economic

Need for collaborative and inclusive approaches to AI governance

Speakers

– Tobias Thiel
– Speaker
– Moderator

Arguments

Fair, enabling, risk-conscious, and equitable regulation requires collaborative partnerships bringing together public sector, private sector, academia, and civil society

Africa needs sovereignty and control over AI aspects it wants to adopt, ensuring cultural and traditional requirements are met while building locally

The challenge lies in balancing protective governance with enabling frameworks that don’t stifle innovation

Summary

There is strong agreement that AI governance should be collaborative, inclusive, and balanced – involving multiple stakeholders while ensuring African sovereignty and cultural considerations, without stifling innovation through excessive control.

Topics

Legal and regulatory | Development

Importance of practical, problem-solving applications of AI

Speakers

– Kojo Boakye
– Barbara Glover
– Tobias Thiel

Arguments

Meta’s LLAMA models have been downloaded over a billion times, with African winners of Impact Grants like Digital Green and Jacaranda Health creating solutions for farmers and maternal health

Focus should be on problem-solving approaches through domain-specific education and applications addressing real-time challenges in health, agriculture, and education

GIZ works on building inclusive AI ecosystems through initiatives like Fair Forward, partnering with local organizations to address data gaps in the Global South

Summary

All speakers emphasize the importance of using AI to solve real-world problems in sectors like healthcare, agriculture, and education, rather than pursuing AI for its own sake. They advocate for domain-specific applications that address concrete challenges facing African communities.

Topics

Development | Sociocultural

Similar viewpoints

Both speakers emphasize the importance of empowering young Africans to innovate with available tools rather than waiting for perfect conditions or excessive government intervention. They believe in the inherent capability of African youth to create solutions when given access to technology.

Speakers

– Barbara Glover
– Kojo Boakye

Arguments

Young people can leverage open source platforms to address real challenges without waiting for government infrastructure improvements

There’s a need for caution around government control that might negate the ingenuity of young Africans who can create solutions when given tools

Topics

Development | Economic

Both speakers advocate for leveraging Africa’s diverse strengths and capabilities across the continent, including language diversity and existing areas of excellence, to build a uniquely African approach to AI development that serves local needs.

Speakers

– Speaker
– Tobias Thiel

Arguments

Africa can build islands of excellence and leverage collective capabilities across the continent to achieve African-led AI development

Open source AI allows creation of localized datasets and language support for underrepresented African languages like Swahili and Luganda

Topics

Development | Sociocultural | Multilingualism

Both speakers advocate for pragmatic approaches that work with current realities rather than waiting for ideal conditions. They emphasize the need to pursue AI development in parallel with infrastructure development, starting where conditions are most favorable.

Speakers

– Barbara Glover
– Speaker

Arguments

The approach should be realistic about Africa’s current capabilities while maximizing available opportunities rather than waiting for ideal infrastructure

Infrastructure limitations require simultaneous work on fundamentals while deploying AI solutions, focusing on cities with connectivity first

Topics

Development | Infrastructure

Unexpected consensus

Cautious approach to government control over AI

Speakers

– Kojo Boakye
– Tobias Thiel
– Moderator

Arguments

There’s a need for caution around government control that might negate the ingenuity of young Africans who can create solutions when given tools

Fair, enabling, risk-conscious, and equitable regulation requires collaborative partnerships bringing together public sector, private sector, academia, and civil society

The challenge lies in balancing protective governance with enabling frameworks that don’t stifle innovation

Explanation

It’s unexpected that a Meta representative, a development agency representative, and a moderator would all express similar concerns about excessive government control. This consensus suggests a shared understanding that while governance is necessary, it must be carefully calibrated to avoid stifling innovation and entrepreneurship.

Topics

Legal and regulatory | Development

Need for simultaneous rather than sequential development approaches

Speakers

– Speaker
– Barbara Glover
– Kojo Boakye

Arguments

Infrastructure limitations require simultaneous work on fundamentals while deploying AI solutions, focusing on cities with connectivity first

The approach should be realistic about Africa’s current capabilities while maximizing available opportunities rather than waiting for ideal infrastructure

Investment in connectivity, device affordability, and training programs creates multiplier effects when combined with government adoption

Explanation

There’s unexpected consensus across different organizational perspectives (AU, development agency, private sector) that Africa cannot afford to wait for perfect infrastructure before pursuing AI development. This pragmatic approach represents a shift from traditional sequential development thinking.

Topics

Development | Infrastructure | Economic

Overall assessment

Summary

The discussion reveals remarkably high consensus among speakers on key issues: open source AI as an opportunity for Africa, the need for balanced governance approaches, emphasis on practical problem-solving applications, and pragmatic development strategies that work with current realities rather than waiting for ideal conditions.

Consensus level

Very high consensus with minimal disagreement. The main tensions were around the degree of government control needed, but even this was more about calibration than fundamental disagreement. This strong consensus suggests a mature understanding of both opportunities and challenges, and indicates potential for coordinated action across different sectors and organizations in advancing open source AI for African development.

Different viewpoints

Level of government control and sovereignty over AI systems

Speakers

– Kojo Boakye
– Speaker

Arguments

There’s a need for caution around government control that might negate the ingenuity of young Africans who can create solutions when given tools

Africa needs sovereignty and control over AI aspects it wants to adopt, ensuring cultural and traditional requirements are met while building locally

Summary

Kojo Boakye warns against excessive government control that could stifle innovation by young Africans, emphasizing the importance of putting tools directly in people’s hands. In contrast, the Speaker (Adil) argues for African sovereignty and control over AI systems to ensure they align with local cultural values and traditions, advocating for local development capacity.

Topics

Legal and regulatory | Development | Cultural diversity

Approach to addressing infrastructure limitations

Speakers

– Moderator
– Speaker

Arguments

Over half of Africa’s population lacks internet connectivity, there’s a shortage of data scientists, limited access to computing power, and insufficient capabilities in public and private sectors

Infrastructure limitations require simultaneous work on fundamentals while deploying AI solutions, focusing on cities with connectivity first

Summary

The Moderator emphasizes the fundamental barriers including connectivity, human capital, and institutional capacity gaps that must be addressed comprehensively. The Speaker acknowledges these challenges but advocates for a parallel approach, starting with urban areas that have better connectivity while working on broader infrastructure simultaneously.

Topics

Infrastructure | Development | Digital access

True openness and independence of open source AI systems

Speakers

– Moderator
– Kojo Boakye

Arguments

Questions arise about true openness of systems like LLAMA, with concerns about backend accessibility for independent development

Meta’s LLAMA models have been downloaded over a billion times, with African winners of Impact Grants like Digital Green and Jacaranda Health creating solutions for farmers and maternal health

Summary

The Moderator raises critical questions about whether supposedly open source systems like LLAMA provide true backend accessibility for independent development. Kojo Boakye counters by highlighting the practical success stories and widespread adoption of LLAMA, suggesting that the level of openness is sufficient for meaningful innovation.

Topics

Legal and regulatory | Development

Unexpected differences

Risk tolerance regarding government control versus innovation freedom

Speakers

– Kojo Boakye
– Speaker

Arguments

There’s a need for caution around government control that might negate the ingenuity of young Africans who can create solutions when given tools

Africa needs sovereignty and control over AI aspects it wants to adopt, ensuring cultural and traditional requirements are met while building locally

Explanation

This disagreement is unexpected because both speakers are advocates for African development and open source AI adoption, yet they have fundamentally different views on the role of government control. Boakye, representing a tech company, surprisingly advocates for less control to enable innovation, while the AU representative advocates for more control to ensure sovereignty – a reversal of typical expectations about corporate versus governmental perspectives.

Topics

Legal and regulatory | Development | Cultural diversity

Assessment of current open source AI accessibility and effectiveness

Speakers

– Moderator
– Barbara Glover

Arguments

Questions arise about true openness of systems like LLAMA, with concerns about backend accessibility for independent development

Trust issues in open AI systems and ineffective data harmonization policies prevent innovators from accessing needed data despite supportive policies on paper

Explanation

While both raise concerns about accessibility, their focus differs unexpectedly. The Moderator questions the technical openness of the systems themselves, while Barbara Glover identifies policy implementation gaps as the main barrier. This suggests different perspectives on whether the problem lies with the technology providers or with local governance structures.

Topics

Legal and regulatory | Data governance | Development

Overall assessment

Summary

The main areas of disagreement center around governance approaches (control versus freedom), infrastructure development strategies (comprehensive versus parallel approaches), and assessments of current system accessibility. Despite these disagreements, all speakers share enthusiasm for open source AI’s potential for Africa.

Disagreement level

Moderate disagreement with significant implications. While speakers agree on the fundamental opportunity that open source AI presents for Africa, their different approaches to governance, control, and implementation could lead to conflicting policy recommendations and strategies. The disagreement between corporate and governmental perspectives on control is particularly significant as it could affect how African countries engage with international AI governance processes and technology adoption policies.

Partial agreements

Similar viewpoints

Both speakers emphasize the importance of empowering young Africans to innovate with available tools rather than waiting for perfect conditions or excessive government intervention. They believe in the inherent capability of African youth to create solutions when given access to technology.

Speakers

– Barbara Glover
– Kojo Boakye

Arguments

Young people can leverage open source platforms to address real challenges without waiting for government infrastructure improvements

There’s a need for caution around government control that might negate the ingenuity of young Africans who can create solutions when given tools

Topics

Development | Economic

Both speakers advocate for leveraging Africa’s diverse strengths and capabilities across the continent, including language diversity and existing areas of excellence, to build a uniquely African approach to AI development that serves local needs.

Speakers

– Speaker
– Tobias Thiel

Arguments

Africa can build islands of excellence and leverage collective capabilities across the continent to achieve African-led AI development

Open source AI allows creation of localized datasets and language support for underrepresented African languages like Swahili and Luganda

Topics

Development | Sociocultural | Multilingualism

Both speakers advocate for pragmatic approaches that work with current realities rather than waiting for ideal conditions. They emphasize the need to pursue AI development in parallel with infrastructure development, starting where conditions are most favorable.

Speakers

– Barbara Glover
– Speaker

Arguments

The approach should be realistic about Africa’s current capabilities while maximizing available opportunities rather than waiting for ideal infrastructure

Infrastructure limitations require simultaneous work on fundamentals while deploying AI solutions, focusing on cities with connectivity first

Topics

Development | Infrastructure

Key takeaways

Open source AI represents a significant opportunity for Africa to leapfrog technologically and achieve digital inclusion, with all panelists agreeing it’s more accessible than closed/proprietary AI systems

Africa can leverage open source AI to build local solutions addressing continental challenges in health, agriculture, education, and finance without waiting for ideal infrastructure

Success requires simultaneous investment in fundamentals (connectivity, digital literacy, computing infrastructure) while deploying AI solutions, rather than sequential development

African sovereignty and local control over AI development is crucial, but must be balanced with enabling innovation rather than stifling it through excessive government control

Community building and collaboration are essential – both within Africa (leveraging islands of excellence) and globally (accessing existing programs and partnerships)

The focus should be on problem-solving applications rather than AI for its own sake, with domain-specific education and solutions addressing real-time challenges

Africa needs stronger representation in global AI governance and decision-making processes to avoid being left behind in policy development

Resolutions and action items

African Union to review and consider open source data framework by end of year/early November during ministerial meeting

African Union developing implementation strategy with monitoring and evaluation components for AI strategy

Continued investment needed in connectivity infrastructure, including subsea cables and affordability initiatives

Need to develop African common position on WSIS+20 that includes AI governance and advocates for African representation

GIZ to continue supporting 35 African states requesting help to domesticate the AU data policy framework

Communities should actively access existing programs from tech companies (Meta’s Impact Grants, Accelerator Grants, and similar programs from Google, Microsoft)

Focus on building domain-specific AI education and applications rather than broad AI training

Unresolved issues

How to achieve true backend openness in open source AI systems – concerns remain about whether systems like LLAMA provide genuine independent development capabilities

Balancing African sovereignty and local control with enabling innovation – the tension between protective governance and innovation-friendly regulation

Coordination challenges between different African Union initiatives (digital policies vs trade policies) that may conflict with each other

How to scale AI solutions from urban centers with connectivity to rural areas lacking basic infrastructure

Trust issues in open AI systems and ineffective implementation of data harmonization policies despite supportive frameworks on paper

The fundamental challenge of achieving equity when Africa cannot match the $65 billion investments made by global tech companies

How to ensure cultural and traditional requirements are met in AI development while maintaining technical effectiveness

Suggested compromises

Start AI deployment in cities with existing connectivity while simultaneously working on rural infrastructure rather than waiting for universal coverage

Pursue African-led AI development through collective continental effort, leveraging different countries’ areas of excellence rather than individual national approaches

Focus on practical applications that solve immediate problems while building toward longer-term infrastructure and capacity goals

Develop risk-conscious but enabling regulation through collaborative partnerships involving public sector, private sector, academia, and civil society rather than top-down government control

Combine open source AI adoption with continued advocacy for African representation in global AI governance rather than choosing between local development and international engagement

Pursue domain-specific AI education and applications rather than trying to build comprehensive AI capabilities across all sectors simultaneously

I would reframe it more as open source AI rather than open AI… My competitors are open AI. would be delighted that I’m going to wax lyrical about them rather than ourselves.

Speaker

Kojo Boakye

Reason

This comment immediately clarified a crucial distinction that shaped the entire discussion. By differentiating between ‘open AI’ (the company) and ‘open source AI’ (the concept), Boakye prevented confusion and established the proper framework for discussion. His candid acknowledgment of competitive dynamics also brought commercial realities into focus.

Impact

This reframing fundamentally redirected the conversation’s trajectory. All subsequent speakers adopted the ‘open source AI’ terminology, and the discussion became more precise and technically grounded. It also introduced the competitive landscape as a relevant factor in the conversation.

This feels as though we are presenting an opportunity through open source AI to put those technical tools in the hands of those who probably most need them… we probably haven’t seen the impact we’d like to see. I think open source presents an incredible opportunity to do that.

Speaker

Kojo Boakye

Reason

This comment introduced the powerful concept of technological leapfrogging specifically through open source AI, positioning it as potentially different from previous technological waves where Africa was ‘behind the curve.’ It reframed the discussion from technical capabilities to equity and access.

Impact

This perspective became a central theme that other panelists built upon. It shifted the conversation from what open source AI is to what it could mean for African development, establishing the aspirational tone that dominated much of the discussion.

We have over half the population of the continent who are not connected to the internet… certainly don’t have the capabilities in the public sector, large parts of the private sector, and certainly not in most of the SMMEs… So how do we get to this opportunity that you’ve been speaking about from where we are now?

Speaker

Moderator (Alison Gilwald)

Reason

This intervention was crucial because it challenged the optimistic framing with hard realities about digital divides, infrastructure gaps, and capacity limitations. It forced the discussion to grapple with the tension between aspiration and current conditions.

Impact

This question fundamentally shifted the discussion’s tone and direction. It moved the conversation from opportunities to implementation challenges, forcing panelists to address practical barriers and leading to more nuanced discussions about phased approaches, urban-rural divides, and the need for simultaneous infrastructure and capacity building.

I think the questions in front of us, whether to first work on the fundamentals, and then do AI, or do them in parallel. And I think we don’t have options. We have to do everything at the same time, which is always our challenge in every area.

Speaker

Adil Suleiman

Reason

This comment captured a fundamental strategic dilemma facing African development – the impossibility of sequential development when facing multiple urgent needs. It acknowledged the complexity of development challenges while advocating for a parallel approach.

Impact

This observation resonated throughout the remaining discussion, with other speakers referencing the need for simultaneous action on multiple fronts. It helped establish a more realistic framework for discussing implementation strategies.

Africa probably won’t be a leader in high-performance computing. It won’t be a leader, perhaps, in data center location… So we need to look then at what is our niche? What is it that we can do better?

Speaker

Barbara Glover

Reason

This was a remarkably candid assessment that challenged aspirational rhetoric with pragmatic realism. By acknowledging limitations in certain areas, Glover redirected focus toward identifying Africa’s competitive advantages and realistic opportunities.

Impact

This comment introduced a more strategic, niche-focused approach to the discussion. It led to conversations about leveraging Africa’s young population, addressing local problems, and finding realistic pathways rather than trying to compete directly with global tech powers.

I do have concerns when people start speaking about control, because control, for me, negates in some ways… the ingenuity, innate ingenuity and brilliance of young Africans, who, if we put these tools in the hands of people right now, will create the solutions.

Speaker

Kojo Boakye

Reason

This comment introduced a crucial tension between sovereignty/governance concerns and innovation freedom. It challenged the assumption that more control is always better, arguing that excessive control could stifle the very innovation that open source AI is meant to enable.

Impact

This sparked a nuanced discussion about different types of control – political control versus regulatory frameworks that enable innovation. It forced the panel to distinguish between protective governance and restrictive control, leading to more sophisticated policy discussions.

The metric would be if the African government and the African communities, the African families, they can control what is it, what part of AI they want to take, and what part of AI they don’t want to consider, because otherwise, then you lose your sovereignty.

Speaker

Adil Suleiman

Reason

This comment provided a concrete framework for thinking about sovereignty in the AI context – not as blanket control, but as the ability to make informed choices about adoption. It connected individual agency with national sovereignty in a novel way.

Impact

This reframing helped resolve some of the tension around ‘control’ by focusing on choice and agency rather than restriction. It provided a more nuanced understanding of what digital sovereignty might mean in practice.

Young people don’t have to wait for government to catch up on infrastructure, to catch up on supercomputers and all these issues before they can, you know, harness the world of AI.

Speaker

Barbara Glover

Reason

This comment powerfully articulated the democratizing potential of open source AI, suggesting it could bypass traditional development bottlenecks. It positioned open source AI as a tool for circumventing rather than requiring extensive infrastructure development.

Impact

This perspective energized the discussion by suggesting immediate possibilities rather than long-term prerequisites. It reinforced the leapfrogging narrative and provided concrete hope for near-term impact despite infrastructure limitations.

Overall assessment

These key comments fundamentally shaped the discussion by establishing crucial distinctions (open source vs. open AI), introducing productive tensions (opportunity vs. reality, control vs. innovation), and providing strategic frameworks (niche focus, parallel development, sovereignty as choice). The conversation evolved from initial optimism through realistic challenge to nuanced strategic thinking. The moderator’s reality-check questions were particularly effective in preventing the discussion from remaining at a superficial level, while the panelists’ responses demonstrated sophisticated understanding of the complex trade-offs involved in AI adoption for development. The discussion successfully balanced aspiration with pragmatism, ultimately arriving at a more mature understanding of both the opportunities and challenges facing Africa in the open source AI landscape.

How can Africa advance AI and other emerging technologies at a fraction of the billions/trillions spent in the global North while ensuring equity and long-term health of these tools?

Speaker

Osei Keja from Ghana (audience member)

Explanation

This addresses the fundamental challenge of resource constraints in Africa while trying to compete in AI development and ensuring equitable access to AI benefits.

Do we have control from countries, from a sovereignty point of view, over what AI is coming into our countries?

Speaker

Moderator (Alison Gilwald)

Explanation

This question explores the balance between leveraging open AI opportunities and maintaining national sovereignty over AI deployment and governance.

How can open source AI communities interact with each other, and how can open source AI communities in the African continent be supported and sustained?

Speaker

Ghana Hub, IGF (online participant)

Explanation

This addresses the need for building sustainable ecosystems and networks among AI communities across Africa to maximize collaboration and impact.

With WSIS plus 20 process coming up, do we see a situation where Africa has to concede and take up open source AI for economic benefit in contravention to its policy standpoint?

Speaker

Audience member

Explanation

This explores potential conflicts between Africa’s policy positions on AI governance and the practical need to adopt open source AI for economic development.

How do we ensure effective implementation of data harmonization frameworks that are currently not being implemented effectively despite existing policies?

Speaker

Barbara Glover

Explanation

This addresses the gap between policy frameworks on paper and their practical implementation, which is hindering access to data needed for AI development.

How do we address the lack of coordination between different African Union initiatives, particularly between digital sector and trade policies?

Speaker

Moderator (Alison Gilwald)

Explanation

This highlights the need to resolve conflicts between different policy frameworks that may undermine each other’s objectives, such as trade agreements preventing preferential access to public data for local entrepreneurs.

What governance and regulatory frameworks are needed that are both protective and enabling for open source AI deployment?

Speaker

Multiple speakers (Moderator, Tobias Thiel, Adil Suleiman)

Explanation

This addresses the challenge of creating regulatory environments that manage risks while enabling innovation and access to AI technologies.

How can we build stronger digital literacy and tailored capacity building for both regulators and innovators?

Speaker

Tobias Thiel

Explanation

This identifies the need for comprehensive skill development programs to support effective AI adoption and governance across different stakeholder groups.