In the ESPA Talks section, we discuss with our external consultant, Dr. Mahamed Lamine Guindo, the issue of AI and data, particularly the challenges for African states.
1. Everyone is talking about AI now and it is seen as a new technological revolution. But concretely, what does artificial intelligence consist of and what makes it so special?
Everyone is talking about it, that's true. Whether at the market or in public squares.
AI is a machine that has learned from a considerable amount of information: texts, images, sounds, videos. By analyzing this data, it learns to recognize situations, make predictions, make certain decisions, and even produce new content. We can say, to simplify, that it seeks to replicate certain abilities that we associate with human intelligence.
And this same mechanism allows it today to create things. By typing a simple phrase, you get a video of a wrestling match in an arena that never took place, or a complete mbalax piece, with the voice, the sabar, and the chorus, in less than a minute. It guesses image by image, note by note, from thousands of videos and songs it has already seen and heard. Young people are already producing clips, ads, and songs from their phones, without a studio or camera.
It does not think like us, it must be said. But it does in three seconds what an entire office used to do in a week. And that changes everything: for the first time, the machine is not just replacing our arms, it is starting to replace part of our minds.
The real question is therefore not "what can the machine do?", but "who owns it, and who decides what it does?".
2. We already see that AI has become a battleground for geopolitical rivalries. For you, what are the major stakes surrounding AI for states?
The issue is simple: whoever masters AI will have the economy, the military, and information tomorrow. All three at once.
Today, Americans have the best chips and the best models. The Chinese are close behind with fewer resources. But between us, are they not already at the same level? China likes to hide its game. We saw it with DeepSeek: the whole world was surprised overnight. The gap is now measured in months, no longer in years.
Chips have become what oil was last century: we fight over them, we prohibit selling them to rivals.
And us, in Africa? For now, we are mostly customers. We use tools made elsewhere, and our data, our photos, our voice messages on WhatsApp feed machines that belong to others. The issue for us is this: remain a customer or become a player.
Well, then there’s always someone next to you who claims to be an expert. (Laughter)
3. In the military field, the American company Anthropic disclosed in a report in September 2026 attempts to use Claude AI for military purposes by certain groups or states. Israel has also used AI in the genocide it is conducting in Gaza, according to several organizations. Could AI not risk being diverted from its original role and becoming a major military tool whose precision could leave much to be desired?
I will be frank: we are no longer in science fiction. It is already a reality.
The report published by Anthropic in September is chilling. Thirty-nine cases of diversion in eight months. An armed group in Yemen that wanted to use it to write weapon guidance code. And I emphasize this point: Africa is directly concerned. According to this report, in Mali, a surveillance project capable of tracking about 25 million SIM cards. In the Central African Republic, propaganda manufactured by AI and broadcast on the radio. It’s not in Washington or Beijing, it’s at our doorstep.
The company claims to have blocked these accounts. Good for them. But tomorrow, other tools, less scrupulous, will block nothing.
And still, this report only shows what a company has been willing to make public. No one really knows what is happening in the laboratories of the major armies, American, Chinese, Russian. What we discover today, they may have been testing for years.
And in the meantime, in Africa, apart from a few countries like South Africa, Egypt, or Nigeria that are starting to manufacture their own drones, our armies are not really doing cutting-edge research. They buy equipment designed elsewhere, with software they do not control. It’s a shame, and it’s dangerous: an army that does not understand its own tools is entirely dependent on the one who sells them.
On Gaza, according to an investigation by the magazine +972, it is Israeli officers themselves who have testified. They describe a system, Lavender, that has designated up to 37,000 people as targets. Some claim that they accepted killing fifteen to twenty civilians for a single low-ranking fighter. Fifteen to twenty. The Israeli army contests this, of course. But the mere fact that this debate exists shows where we stand.
As for precision, the real danger is not just that the machine makes mistakes. It’s that we hide behind it: "It’s not me, it’s the algorithm." No. Behind every keystroke, there is a human who decides. And a machine should never have the final word on a human being's life.
You are reading this content for free. Subscribe to access all our geopolitical intelligence productions.
4. The Chinese and Americans are engaged in a frantic race in AI. The Europeans still seem to favor legislation and regulation to avoid excesses. In your opinion, how should Africa position itself in these strategic choices?
The Americans are running to win. The Chinese are running to catch up or to surpass; you never know with them. And the Europeans are writing the rules of the game… because they are no longer really on the field. (Laughter)
Africa should not copy anyone. It would be like buying a suit tailored for someone else: it never fits well.
What we need is simple. First, to speak with one voice. Alone against Google or a Chinese giant, Senegal counts for nothing, let’s be honest. But fifty-four countries together, nearly one and a half billion people, there we count.
Next, tools that speak our languages. An AI that understands neither Wolof, nor Pulaar, nor Serer, nor Diola will never serve the grandmother in the village.
Finally, to clearly tell the big companies: "You want our market? Very well. But on our terms." And these terms need to be calculated.
First condition: train our youth, not just to use ChatGPT or Claude, but to build their own models. Using AI, everyone will do it. Those who create it will decide.
Second condition: be careful with servers. A foreign data center set up in Dakar does not make us sovereign. It remains subject to the laws of its country of origin: an American company may be required to deliver our data to Washington, even if it is stored here. It also consumes a huge amount of electricity and water, in a country that lacks both, for very few jobs once built. So yes to investments, but with Senegalese partners in the capital, our sensitive data placed under Senegalese law, and a real obligation to train and hire locally.
We have the youngest population in the world. It’s our trump card, but we must know how to play it.
5. You have notably worked on issues related to the use of AI in agriculture, a sector in which Africa is still lagging behind. Can we expect an accelerated revolution in agriculture with the use of AI among Africans?
Having worked for several years on these technologies, I can tell you that AI can now intervene at all stages, from the sky to the plate.
In the sky, satellites coupled with AI monitor crops, detect droughts, and predict harvests. For a state, knowing in advance how much it will produce means knowing how much to import and avoiding shortages. Above the field, drones identify sick areas and treat only where necessary, using fewer products and less water. In the field, a simple phone is enough: you photograph a sick leaf and AI tells you what it is.
And after the harvest, AI becomes a true scientific police of food. In a few seconds, it detects adulterated oil, fertilizer loaded with heavy metals, or peanuts contaminated with aflatoxin, a toxin linked to liver cancer. It’s a public health issue: we often hear "he died young." The real problem is that we don’t check what we eat.
Mais il faut regarder la réalité en face. Nos agriculteurs sont surtout de petits exploitants, sur de petites parcelles, avec peu de moyens. Beaucoup n’ont ni électricité, ni connexion internet fiable, ni accès au crédit. Les données agricoles africaines sont rares, et la plupart des outils sont conçus ailleurs, pour d’autres sols et d’autres climats. Et au Sénégal, seulement environ 11 % des ménages agricoles possèdent du matériel motorisé.
C’est pourquoi, pour moi, la priorité aujourd’hui, c’est la mécanisation. On ne peut pas piloter un champ par satellite quand on le travaille encore à la daba. Il faut d’abord des machines, partagées entre producteurs regroupés, de l’eau maîtrisée et des routes pour vendre les récoltes.
Without water, without fertilizer, without roads, no technology will save the farmer. AI is a good advisor, not a magic wand. It will then come to multiply our efforts. We don’t skip steps.
6. In Senegal, how do you perceive today the use of AI and especially the framework in which it is developing? Is research sufficiently supported to make progress on this issue?
Sur le papier, nous sommes bons. Une stratégie nationale depuis 2023, le New Deal Technologique depuis 2025, la Senegal AI Factory présentée à Tokyo cet été, et la première place de l’UEMOA en matière de préparation à l’IA.
Mais je vais vous dire une chose : être premier de l’UEMOA, c’est être le meilleur élève d’une classe qui a du retard.
Le problème du Sénégal n’est pas le manque de stratégies. Des stratégies, nous savons très bien en écrire, nous en sommes même champions ! (Rires) Le problème, c’est le financement et le passage à l’action.
Nos chercheurs n’ont pas les machines pour travailler. Nos meilleurs cerveaux partent à Paris, à Montréal, aux États-Unis. Et demain, nous achèterons très cher ce que nos propres enfants auraient pu construire ici. Cela, honnêtement, me fait mal.
We need a real budget, scholarships to retain our talents, and powerful computers for our universities. Otherwise, we will make beautiful speeches, and it will be others who do the work.
7. What does “digital sovereignty” concretely mean in the age of AI? Can a state truly be sovereign in AI without having its own infrastructures, data, models, and computing capabilities?
It’s simple: to be sovereign is to be able to say "our data, we decide."
Let’s ask ourselves the question: if tomorrow a foreign company cuts the service, do our civil registry, our hospitals, our taxes continue to function? If the answer is no, we are not sovereign. We are tenants. And a tenant can be evicted by the owner whenever they want.
Should we make everything ourselves? No, no one does that. Even Americans have their chips made in Taiwan. But we must at least keep sensitive data at home, train our own engineers, and never depend on a single supplier. You don't put all your eggs in one basket.
8. AI has also become a major source of misinformation, especially with deepfakes and particularly in Africa. How do you see this evolution? Does it worry you?
Yes, that worries me. A lot.
Here, information first circulates on WhatsApp and TikTok. An audio, a video travels around the country in one evening. And in family groups, there is always an uncle or a mom who shares everything without checking! (Laughter)
Today, with just a few minutes of your voice, they can make you say anything. A fake video of a minister, a religious leader, or a candidate costs almost nothing to produce.
But the worst part is not that we believe in the false. It’s that we no longer believe in the true. Tomorrow, a politician caught red-handed will simply say: “It’s a deepfake.” And some of the public will believe him.
What to do? A clear law. Fact-checkers who work in Wolof and Pulaar, not just in French. Require platforms to react quickly, especially during election periods. And above all, teach our children, from school, to ask themselves a question before sharing: "Who made this video, and why?" Just that would already be a great victory.
