### AI Development and African Urbanism This is a space for me to think about the intersection between the progress we are seeing in AI, and African development. I've been thinking about this since I attended a deep learning summer school in Africa organized by a group at Deepmind, known as the Deep Learning Indaba. I've also had the opportunity to participate in career advising from 80,000 hours, an offshoot organization of the broader EA community, where I spoke with people from the Global Priorities Institute at Oxford. Below was my general application for that which, though written back in 2021, is still quite in-line with how I think about this. I believe it is an underexplored perspective in the AI Governance space. ### AI Governance in the African Context My main ‘thesis position’ is that the story of the 21st century is largely going to be the story of the African continent, in that, what happens in Africa is going to affect the future of the rest of the world for a long time to come. The single biggest data point that influences my thinking here is that by current population projections, Africa will grow from 1.3bn people (current) to ~4.3bn by century’s end (2100). Developing adequate, formal thinking frameworks for how to solve the challenges that this growth will entail (that are sufficiently specific and appropriate to Africa as a region), and that also make use of the advancements we are seeing in new AI systems, is very likely one of the most impactful things anyone can do, in my opinion. In short: valuing longtermism relies on getting Africa right, and I feel that we aren’t on track to do so. ![[Pasted image 20250105213858.png]] *Ranking of the 100 most populous cities in the world and how they change across the 21st century, from the Washington Post article: Africa's Rising Cities*: [Washington Post](https://www.washingtonpost.com/world/interactive/2021/africa-cities/) I see this as the confluence of two megatrends in the 21st century: The speed of AI development and the need for a better approach to development in Africa given the pace of urbanization. Consider what it means for a continent to build cities that will house hundreds of millions of people in an era of artificial intelligence. The question isn't whether these cities will be different from those of the 19th or 20th centuries – they undoubtedly will be – but rather how they might leverage the rapid advances we're seeing in AI. The opportunity to integrate AI systems thoughtfully into infrastructure, governance, and daily operations presents both a serious challenge and an extraordinary opportunity. The massive urbanization ahead isn't just a challenge of infrastructure – it's a challenge of governance and social systems. When we talk about AI safety and governance globally, we often focus on scenarios involving highly advanced AI systems. But in Africa, we need to think about AI safety in more immediate, practical terms: How do you ensure an AI system managing water distribution in a rapidly growing city is reliable and fair? How do we develop AI-driven public transportation systems that serve all communities equitably? How do we implement AI governance structures that respect traditional decision-making processes while also embracing and making use of all of this technological progress? ![[Pasted image 20250105212745.png]] *Lagos will become the worlds most populous city by 2100* The climate crisis adds another layer of complexity. In the Sahel region, where temperatures are rising 1.5 times faster than the global average, we need AI systems that can help communities adapt to rapidly changing conditions. How can we fund systems in Africa that can help design climate-resilient agriculture, optimize water usage, and support adaptive urban planning? How can we efficiently scale new start-up ventures within Africa in a way that is correctly incentivized for in places that are external to Africa, and in ways that avoid the returns being bidded down by simply further investing into the developed nations? In some ways I think these challenges actually play to some of the continents strengths. One of Africa's big characteristics has been its ability to leapfrog certain developmental stages. Many African countries skipped landlines entirely and went straight to mobile phones. The same thing happened with banking – instead of building traditional bank branches, many countries moved straight to mobile banking and mobile payment. In the same way, there's an opportunity to think carefully about how AI systems are adopted and developed in Africa, ensuring this technological leapfrogging leads to genuinely beneficial outcomes. The effective altruism community often talks about longtermism – the idea that we should be particularly concerned about humanity's long-term future. I would argue that getting Africa right is crucial to any longtermist project. We cannot meaningfully talk about humanity's long-term future when we're not adequately considering how nearly half of humanity will develop and grapple with the challenges of this technology over the coming century. I find myself wondering about questions like: - How do we ensure AI systems actually serve local needs rather than just importing solutions developed for different contexts? - What does AI safety look like when you're building systems that have to work in environments with very little reliable infrastructure? - How do we develop governance frameworks that can scale with the fastest-growing cities in human history? These questions matter because getting AI governance right in Africa isn't just about Africa – it's about developing models that might work better for everyone. The solutions that emerge from addressing these challenges could show us new ways to think about AI alignment, safety, and governance globally. This is my attempt to create a space for these discussions. I don't have all the answers, but I think the very fact of raising these questions and working toward answers, is one of the most impactful things we can do. I believe the story of AI development in Africa will test a lot of our assumptions about technology and progress. When we think about existential risk and long-term futures, we need to consider not just how AI might develop in research labs, but how it will be implemented in the places where humanity is growing the fastest. When the history of AI development is written, it should tell a story not just of technological advancement, but of wisdom, foresight, and genuine human flourishing -- and that story should include Africa.