Искусственный интеллект в Африке: настоящая революция происходит не в чат-ботах, а в шахтах и банках

At a recent Artificial Intelligence (AI) conference in Johannesburg, I watched people marvel at a robot walking across a stage. It was an impressive demonstration of how quickly AI and robotics are advancing. But during a conversation afterwards, Tertius Zitzke, the chief executive officer (CEO) of 4Sight Holdings, a South African technology company, offered a useful reminder. Someone had to put that robot in a vehicle, drive it to the conference, switch it on and eventually put it back in its crate. The joke points to a bigger problem with the way we talk about AI. We are fascinated by the visible parts of artificial intelligence: chatbots, humanoid robots, copilots and sophisticated AI assistants. But some of the most consequential AI adoption may be happening in much less visible places. It is taking place within the systems that process invoices, manage workflows, optimise energy consumption, operate mines, and support business decision-making. Africa’s AI revolution may not look much like the one dominating Silicon Valley. It may have fewer viral chatbots and consumer-facing AI startups, but more machines quietly making decisions inside mines, factories, banks, telecom networks and corporate back offices. From 4Sight’s push to automate routine business processes to ABB Group’s vision of connected mines, where AI can optimise operations from pit to port, a deeper transformation is taking shape beneath the layer consumers see.  For African businesses, the real test of AI may not be whether they can build the next ChatGPT, but whether they can use it to make some of the continent’s biggest industries more productive, safer and competitive. The technology is becoming less of a product people interact with and more of an invisible operating layer running the economy.  The AI that doesn’t look like AI The enterprise AI conversation is moving away from what an AI model can say towards what it can actually do. That is the strategy behind 4Sight’s approach. Zitzke’s view is that “artificial intelligence” is not necessarily the best description for what businesses need. The company calls its approach 4Sight Automated Intelligence, focused on automating routine business processes at a deeper level. The distinction matters. A chatbot can answer a question about an invoice. An automated system can read the invoice, extract the relevant information, compare it with previous transactions and trigger the next step in a company’s workflow. Zitzke offered a simple example involving utility bills. Historically, a business might capture the amount, date, and invoice number before paying the bill. But the document contains much more information. An AI system could read the metre readings, compare water consumption with previous months, and flag an unusual increase. If consumption suddenly doubles, the system could alert the business to investigate. Perhaps a pipe has burst. The value is not the AI’s ability to have a conversation. It is the decision the business can now make because the machine has extracted and interpreted information that was previously sitting unused inside a document. That is a different proposition from adding another chatbot to an employee’s desktop. 4Sight’s strategy is built around the idea that this kind of automation can translate into measurable business efficiency. Zitzke pointed to the company’s reported 54% revenue growth in its most recent results, alongside a 5% increase in its staff complement, as evidence of the efficiency it is seeing within the business. That should not be read as proof that AI alone generated the growth. But it illustrates the economic logic behind enterprise automation: businesses do not necessarily have to employ fewer people to get more output from the same workforce. As Zitzke put it, a company can fire people to cut costs, or use technology to make the people it has more efficient. The difference is important for Africa, where the AI debate is often reduced to whether machines will destroy jobs. The more interesting question may be what happens to those jobs when the repetitive parts are automated. What about the physical economy? If 4Sight shows how AI is moving through the office, ABB illustrates what happens when the same logic reaches the physical economy. Mining is one of Africa’s most important and technologically complex industries. In a conversation with TechCabal on Monday, Charl Marais, ABB South Africa’s Local Division Manager for Process Automation Process Industries, described one of the industry’s biggest challenges as a problem of “islands of automation”. A modern mine may have sophisticated systems controlling different parts of its operations, but those systems do not necessarily operate as a single, connected intelligence. A grinding circuit might run on one programmable logic controller, a concentrator on another automation system, and power control elsewhere. Each system may work well independently. The problem is that the mine does not necessarily have visibility across the entire operation. Once these systems become interconnected, mines can begin to understand how decisions in one part of the operation affect another. A bottleneck in a crushing circuit, for example, could affect energy consumption elsewhere. Greater visibility across the value chain allows operators to optimise production and energy use in ways that were previously difficult. This is where the AI story becomes bigger than software. The opportunity is not simply to put an AI assistant in front of a mining engineer. It is about connecting physical infrastructure, operational data and automation systems so AI can help optimise the entire operation. ABB has been working on automation in mining for decades. What is changing is the convergence of automation, digitalisation and AI. Marais sees a future in which mines can be managed from operation centres, including remote facilities, with technology helping operators make informed decisions while reducing the need to expose workers to dangerous environments. The machine, in other words, is not necessarily replacing the worker. It is changing where the worker sits, what information they have and which decisions they need to make. Africa’s constraints could fuel its AI future There is an uncomfortable irony here. Africa is often described as being behind the rest of the world in industrial automation and digital infrastructure. Its businesses contend with unreliable electricity, ageing infrastructure, skills shortages and pressure to keep costs low. But those same constraints could make automation more valuable. A mine that can optimise electricity consumption has a direct economic incentive to do so. A company that can process thousands of invoices without manually capturing every field can reduce administrative friction. A mining operation that can move workers out of dangerous environments while retaining their expertise has a powerful safety incentive. This is why Africa does not need to replicate Silicon Valley’s AI journey to benefit from the technology. Its opportunity may lie in applying AI to sectors that already dominate its economies. The continent has enormous mineral resources, growing financial and telecommunications industries and large enterprises with huge amounts of operational data. The question is whether that data can be used to make better decisions. For mining, the stakes are particularly high. Africa is home to some of the world’s most important deposits of copper, cobalt and other critical minerals needed for the energy transition. If demand continues to rise, African producers will face pressure to increase output while controlling costs, energy consumption and environmental impact. That makes industrial intelligence less of a futuristic experiment and more of a competitiveness issue. The jobs question is more complicated None of this makes the employment debate disappear. In fact, it may make it a bit more complicated. Zitzke’s view is that AI will change every job. Marais, meanwhile, says automation is designed to work with people and move them away from dangerous environments rather than simply replace them. Both views point towards the same reality: the job itself is likely to change. Consider accounts payable. If AI can extract information from invoices, reconcile them, and push transactions through a workflow, a business may need fewer people to handle manual invoice processing. But that does not necessarily mean the employee has no role. The work can move towards exception handling, oversight, analysis, and process management. Zitzke gives the example of corporate lawyers. Instead of simply reading contracts and providing legal opinions, legal teams may be expected to understand the implications of AI across every department of a business, including governance, security and compliance. New responsibilities appear as old tasks disappear. That is why the global AI jobs debate is more nuanced than a simple count of jobs lost to machines. The International Labour Organisation estimates that one in four jobs globally is potentially exposed to generative AI, while finding that transformation is more likely than outright replacement. For Africa, however, the transition raises a difficult question: will workers receive the training needed to move into these new roles? The biggest obstacle may not be AI This is where the common thread between 4Sight and ABB emerges. The biggest obstacle to enterprise AI may not be the technology itself but the organisation. Zitzke repeatedly emphasises that successful AI adoption requires more than buying technology. Businesses need structured data, mapped processes, skills development and, most importantly, adoption and change management. His description is blunt: “Training without coaching is pure entertainment.” That is a useful warning for companies rushing to deploy AI tools. Employees need to understand not only how to use AI, but how to work alongside autonomous systems, establish guardrails and know when a human needs to intervene. The same principle applies to industrial environments. Marais’s description of “islands of automation” shows why simply adding AI to an existing system may not be enough. If the underlying systems remain disconnected, AI cannot see the whole picture. In the office, the problem can be fragmented workflows. In the mine, it can be fragmented automation. In both cases, the value comes from connecting the pieces. The invisible AI economy This is why the robot on the stage may ultimately be less important to Africa’s AI story than the systems nobody sees. The most important AI system in an African business may never have a face, chatbot window, or humanoid body. It could be an algorithm deciding how a mine uses electricity. It could be software identifying an abnormal water bill before it becomes a costly problem. It may be an AI agent processing invoices or moving information between departments. It could be a connected industrial system that helps an operator understand what is happening across an entire mine, rather than seeing only isolated parts. The AI that matters most to African businesses may be the AI customers never see. That changes the choice Africa needs to make. The continent does not necessarily need to win the global race to build the biggest AI model. It needs to become exceptionally good at applying intelligence to the industries that already power its economies. The real AI revolution may therefore happen somewhere most people will never see it, behind the screen, beneath the chatbot and inside the machines that keep Africa’s economy running. True scale demands moving beyond surface-level integrations to robust execution. We’ve filtered the noise out of Moonshot 2026, optimising the conference strictly for high-calibre connections between startup founders, global financial operators, enterprise leaders and individuals rewiring Africa’s technical frameworks.  Get 20% off Early Bird tickets for a limited time.