AI Data Sovereignty and its importance in Africa

Welcome to Building Better Developers, the Develop-a-Nor podcast, where we work on getting better step by step, professionally and personally. Let's get started. Well, hello and welcome back. We are continuing our season of Building Better Developers, Develop-a-Nor with a focus on getting unstuck and moving forward in forward momentum and just starting the year correctly. I am one of the founders of Develop-a-Nor, also the founder of RB Consulting, where we help you do a technology reality check. We help you before you get into that big project, you start investing all that time, all those resources. We help you figure out what is where you at? What is the foundation that you have? How do we make sure that we make that solid enough that you are able to build on it and plan for your future so that you're successful not only today and next week, but six months and even six years down the road? Good thing and bad thing. Good thing is that we are in a flurry of interviews, I'm in a flurry of networking calls and being on podcasts. There's a lot of great things that are out there. There's a lot of cool ideas. There's a lot of great people. We've had a lot of fun with these. I have enjoyed many of these conversations. It is one of those things that's like, sometimes when you get out of your coding bubble a little bit, you step out of your chat with AI and actually talk to real people. You can have some really cool things that come to mind and it can help you get out of your rut and figure out how do you want to proceed? How do you get past that problem that's been blocking you a little bit? The bad thing is there's just not enough hours in the day. The bad thing is that I'm starting my days very early and I'm ending my days very late because it is a global, the world has shrunk. When you're dealing with people everywhere from New York to Hong Kong and all points in between, whichever direction you go to get to those two, then your day becomes almost a 24-hour day. It can be a little exhausting, but you know what charges me up every time is hearing Michael give his introduction. Introduce yourself. Hey everyone. My name is Michael Milash. I'm one of the co-founders of Developineur, Building Better Developers. I'm also the founder of Envision QA, where we build and test custom software that eliminates the bottlenecks. That way your business can run smoother and grow faster. Good thing and bad thing. Good thing, last time I mentioned I reorganized my office, I am actually finally getting to the point. Rob's a little bit ahead of me on this, but I'm looking at building my own private AI LLM, trying to put some systems together and build that. So I'm kind of excited about that, geeking out a little bit. Downside, because of the memory shortage, some of the things I'm looking to buy to build this thing are super expensive for no stupid reason other than they are selling it to AI. They've sent something to consumers and it's all enterprise. My daughter can relate because we're trying to build her next gaming rig and it's outrageous. It's like this is stupid. But hey, that's why it's the bet. I think they call it first world problems. You know, that's like, oh well. I totally relate. I was, I'm, you know, granted I got frustrated, accidentally like wiped out a whole bunch of documents in my LLM recently, but I was like, oh, I didn't mean to do that. But it was a good opportunity for me to like go and as I regularly do, reset some stuff, reorganize and just make it a little fight tighter, faster, better. And yeah, it's very easy to geek out on that. It is like I have spent probably more time than I need to in my off time in the last few weeks pulling some of that stuff together and really just like, you know, refining the processes that we have. But that is what we talk about when we talk about working on your business is making sure that you can turn it into squeaky clean processes that you could hand off to somebody else. And if you get hit by a bus, your company can just keep chugging right along. All of that. Have fun with it though. Oh, definitely. There's like, trust me, there's a lot of cool fun stuff I've been doing with this. It's nice to be able to just like crank this stuff out. All these little ideas that Norma is like, I don't have a week or two to do it when it's like, oh, I can have AI go do it and get it done in a couple hours. Makes a huge difference. It is amazing how many things I'm knocking off of my to do list for even like the simple automations, not including the bigger automations. We digress. We got to get back to our conversation with Dr. James Mysiri. Just for those that are out there, it's M-A-I-S-I-R-I is how you spell his last name. There will be a quiz somewhere along the way. So that is important to keep up with. Plus, also, there'll be links in the show notes for all of those things. It's a great conversation. Let's not push it off anymore. Here we go back to our conversation with Dr. Mysiri. Now, one of the things you mentioned, which is fascinating and one of those great uses of AI in that is, for example, like the farmers, they can take a picture of a leaf and they can figure out like what are all of the specifics about it and chasing it down. And I'm assuming that's from growing it to dealing with pests or pests and things like that and even diseases that the various farmer may run into. But one of the things that that sort of brings to light is where you have situations and problems that are unique to Africa, like we'll just take it as a continent, is there a data infrastructure because AI has to have the data before it can figure stuff out. So the concern would be if it's solving a problem that's overly, it's too much China data or too much Russia data or US data or something like that, that it doesn't actually apply. So that's probably the wrong answer, essentially, for when you switch it into an African environment. Is there a push or has there been a push from just a, we'll just call it a digitization of some of these nations even before we get to AI to try to get some of that data pushed into systems so that AI will then be able to use it when it gets to that point? So that problem happens more often than you believe, right? Because we don't create our own AI systems, we import them. And it doesn't affect, it doesn't land well in our context and it can cause more problems than it actually can fix. So I'll give you an example. Zimbabwe partnered with China for facial recognition systems. And when this AI came, because the AI systems was trained on Asian facial recognition systems, it basically was not working. It didn't work. It didn't work well within the Zimbabwean context. So it created a situation where Zimbabwe allowed China to collect data of Zimbabweans without the explicit consent, knowledge, or even awareness, allowed China to train the AI system based off that and control the data around that. So a lot of academics have a problem where we're seeing an extraction of African data companies within the global north that they have no say about, they have no control about because they don't have the infrastructure to hold it, to train it and so forth and so on. So it's a serious, serious crisis. And sometimes it's a life threatening crisis. Like, for example, I think there's a stat that 1% of the global health data originates from the African countries, right? So there's a case in South Africa where in 2019, black practitioners were starting to notice that they were being accused of fraud, like medical aid scheme frauds. And they were just like, you know, curious, like, why is this happening to us disproportionately? And when they did launch an investigation, they found that AI automated systems made those particular decision. And the suspicion is that's because the context that you know of black practitioners within low community environments are much different from the AI that's trained on those particular decisions. So they disproportionately punished those people. So we need to create our own AI's with our own data to make our own decision, because we're just giving data to those companies. We don't really, we're not really actively involved in how it's trained and all of that. We just pay the cost. We just buy the final product. Can I give you one more case? I think it would be interesting. So in South Africa, 68% of small medium enterprises in Africa is done by an AI system or is governed by an AI choice. Like if you want to loan something, if you're a small medium enterprise, you want to loan on from banks or financial institutions, 68% of them is governed by AI in one shape or form or the other. And the study in South Africa found that in construction businesses, a women led construction business is 53% less likely to actually get a loan. And when they do get a loan, it's much smaller at a high interest. And the reasons for that as many, but one of the reasons is because the AI systems are not trained with the African unique histories where certain women didn't have property rights or property rights in the back of the day and historical data points, which all feed into the decisions that is made. So we need to make our own systems with our own histories, our own values to help us move forward. So it's a serious, serious problem on our soil. Yeah, that is a fundamental issue that you guys are going to have to figure out because it is. Yeah, AI is great. If you have data that is essentially bad data, it's trash data in your case, then it doesn't help you. So one more question on that is, so you've obviously, being a doctor that you've spent a lot of time in this, you understand it, obviously you research it. How does this land when you're talking to other people? Is it something that is common sense enough that people are like, yes, this makes sense. We need to have our own data. We need to have our own sources. Or is it something that you get? There's a lot of pushback on that. No, so I have a because I'm a professional speaker as well. Right. So I have a very common keynote that's called If Africa doesn't, if Africa does not shape AI, AI will shape Africa. Where I'm basically showing how these important technologies are even affecting our culture, like the case studies within our culture, our culture has been shifted out daily lives. And it's always mind boggling to the audience. Like it's always an eye opener. It's like they had no idea whatsoever. I mean, that and also to the degree that decisions are made around their lives by algorithms, they had no clue, like at all, at all. So there's generally in the so far, there's always a great response, a great need for change. Like the audience is very big on change and just big on building awareness one stage at time. So you've talked about, you know, after building their own AI, that there's the need for you guys to build your own AI models, your own AI infrastructure. Understanding that basically how these models are trained determine how well they work in the particular regions they're in. In the US, you know, we've got GROG, we've got Chattupti, OpenAI, all these systems. China's got their own. What currently is working best for you? Because here there's a lot of like concerns about racism and inconsistencies in certain AIs as to how they're trained. What are you looking at right now as the AI models that are useful enough, powerful enough that they're not going to unintentionally influence the young minds or those using AI in your country in the wrong way? That's a good question that are not going to unintentionally influence the youth in a certain direction. I don't know if I can give a credit to any AI model system right now that's not going to do that. I really, really don't. I can't give flowers to any of the systems right now. Truth is, it's a miss. It's truly a miss and we're just scrambling to try and figure it out. I don't think we're there yet or I can't even see the light of the tunnel with regards to that at this current point, you know. No, we're still figuring out. We're still trying to figure it out. I don't know if I can give flowers to any system like that right now. I was just curious because I keep trying everything and keep using things for certain things. It's like one model is good for this and another model is good for this and you just avoid certain questions with these models because you know you're not going to get the right... You may not get the perspective that you want. That's one of the biggest, I guess, concerns with these models, right? Because corporations are building these. We don't know how they're trained. Speaking of training, so as Africa grows there, as you start building this initiative, as you start looking at building these data centers and building your own models, what is your vision on how to train these models in a safe, secure way that will benefit Africans? So a lot of this discussion on literature right now is based on Ubuntu ethics, like training AI models based on Ubuntu ethics. Ubuntu is basically an African philosophy that says, I am because you are. So it's very communal. Like you don't think... You know, it's not an individualized culture like you'll see in many parts of the West, but it's very communal. You don't do anything that would hurt that as a net negative within society. Everybody should be accounted for, even what you deem as lower class whenever you make decisions within these technologies. So when we train... The push in academia is that when you train data, that's the ethical framework of how you should make sure your AI output should be in terms of fairness. And the problem is when we import these technologies, there's direct contradictions with that. So for example, in the educational sector, like in a lot of traditional African education is very, very communal. Like it's very, very communal. And there's been a critic that these AI systems like AI tutors practice individual learning, like screen time learning, which is changing the actual way that that specific African culture learns. If that makes sense. So yeah. It does. And that's actually an interesting idea because if there would be a way... And this could be something that could be specific to Africa, to your culture, is if you can figure out how to take these AI models and instead of making them individual thinkers, making them group thinkers, like somehow train these models to... Just the way you explain that just makes me think that that could be an opportunity there to take AI to the next level. Instead of it being like one, be a group thinker instead of like a single thinker. That's kind of cool. And that would mirror the culture, the ground exactly, which is exactly what I'm saying. So to do that, are you looking at maybe introducing instead of like single AI, but like maybe introduce them into these communities and have the communities maybe try to build these systems together, which might holistically build the AI system that is more conducive to the African way of thinking? Yeah. So I'm a big believer that no matter what AI is introduced in these communities, you need to go speak to the community. And whether it's the community leaders, the traditional leaders within that particular community and have these conversations within on what that specific AI will do within the community. Otherwise, it will end up having adverse consequences. At this stage right now, to my knowledge, there is zero of that. I've seen that in academic literature where a lot of people are advocating for it theoretically. But practically on the ground, I mean, I can't imagine open, maybe that's unfair, but I can imagine open AI coming down to Zambia, sitting down with community leaders saying, we want to do this, we want to do that. Maybe that's unfair. Maybe they would. But I'm not seeing it at this current stage in point in time. In fact, what I do see, especially from these NGOs, is not communicating with communities. In Zambia, for example, childbirth is not just a medical thing. It's extremely cultural. These women have been passing on what to do from generation to generation. They know what roots to eat or what juices to take in. They go to the traditional healers for specific herbs to protect the baby physically and spiritually. That's what they believe as well. But they introduced an AI app into that community to help. But that output of the AI apps only suggest Western practices and Western cultures. So as more young people lean towards those practices, their own indigenous knowledge and way of being tends to become invisible. That's not to say AI apps are bad or anything like that. They do save lives, but it's just to show the change that such a technology can have. Because that AI app is not trained on the Zambian culture. Maybe the AI app could tell them, take this herb at this particular time. I think that's again, it goes back to a fundamental difference of how AI works. Even if OpenAI decided to come down and spend some time, they have to understand that whole Ubuntu approach. They have to understand what the culture and what the science and the spiritual healers and all of those facets that you guys live with. This is the data we're working with. Because, for example, if it gives you a Western solution and says, go get this drug that is not available, then it doesn't help. It's like, okay, well, it needs to have, for an AI engine to be useful, it has to have the right context to talk the same language that you do so it can actually give you solutions that are feasible. Like I said, there's a whole season worth of discussion on this alone. I think in some of the things that you guys are running into and facing. This has been fascinating and this is blown right by as far as our time here. One of the things before we go right ahead. Sorry, I just thought if I can say that there's no difference, especially for your audience, which is an entrepreneur, there's no difference between a problem and an opportunity. They're fundamentally made out of the same thing. If there's a problem, you can find an opportunity behind it and a solution behind it. I think whatever investors are listening, I really think the future is in Africa in terms of AI. If you can develop an AI problem, an AI for African problem, it will be extremely, extremely profitable for within your end and for you. While you're highlighting the many problems we have, these are many solutions for those who are listening. I think you did point that out where you said that maybe in the United States and China or EU are way ahead of where Africa is there, but that is also an opportunity for them to learn from those stakes that have been made in these other places that they advanced and come into Africa and say, okay, now we've tested this stuff out. We know how the solutions should go. We know how this should grow so that we can help them help Africa catch up. Hopefully, it's way almost always works is if you get somebody way ahead and they treat somebody behind them that helps that person behind them to catch up faster as opposed to taking the long slow route that the researchers took the first time around. I love that you mentioned that and that is definitely an opportunity for anybody looking forward. Every one of those problems that we've listed are also problems that if you can solve those problems, those are opportunities for impacting a lot of people, a lot of lives and somewhere along the way as a business, you're going to be successful when you do something like that. Like I said, this has been fascinating. What are the best ways for people to get a hold of you and either to just talk to you more or also, if there's any links that you would want to throw out or ways to support some of the efforts that you're backing and that you're pushing for? There's three ways you can hit me on my email at jamesmysiri. I don't know if my name is going to be there on the screen when you edit out. It's going to be jamesmysiri, the number one at gmail.com or you can message me on linkedin at jamesmysiri or you can book me for a talk because I do a lot of keynote talks to raise awareness. I think that's the answers I have now. Excellent. Well, we'll make sure that there are links in the show notes for those that are listening and those that are watching. They'll be there as well because this is one of those areas that we hadn't spent as much time exploring, but I think it is an area where there was a lot of exploration left to be done, a lot of opportunities here. Especially anybody out there that's trying to think of a new side hustle or a new product to launch or anything like that, then I think you have a whole continent full of potential items there to look into. So thank you so much. Thanks for your time. Yes, and positive for it. This is doing it for the community, not just for yourself because that is the way. Thank you so much for your time and hanging out with us. For those of you who are listening, thank you for your time, for spending some time with us and investing some time in your career and also in hanging out with us and our guests. As always, go out there and have yourself a great day, a great week, and we will talk to you next time. Great success. Keep learning, keep growing, and we'll see you in the next episode.