Monday, 6 July 2026

The power and challenges of AI in Africa

IEC e-tech

article here

How is artificial intelligence (AI) used on the African continent and what are the specific demands of African countries? We survey different countries in Africa and look at their AI needs and challenges.

Governments across the world are prioritizing AI as the basis for national competitiveness and African countries are no different. A rapidly growing number of programmes, strategies, and pilot projects are being launched across the continent. As a result, public policy is being shaped to address gaps in infrastructure and specialist skills that might limit these countries’ ability to compete globally.

There are notable differences in the levels of economic maturity, compute and internet capacity as well as institutional frameworks in each country of this vast continent, but at the same time most of them share the same concerns.

One of the main ones is a common belief that in order to be independent of technologies trained elsewhere and or owned by third parties, each state should ideally develop its own sovereign AI systems. These should be engineered to reflect linguistic diversity,  address the  requirements of informal economies as well as resource‑strained public services.

A May 2026 study by PwC found that an overwhelming majority of African organizations are already engaging with AI but also that few had succeeded in building out the technology “in ways that generate sustained growth, operational reinvention or competitive advantage.”

As AI rises up on the agenda, countries that invest in inclusive strategies, automation, innovation, and broad AI education - backed by serious funding and robust national policies - will be positioned to lead.

This article snapshots a selection of African countries and their pathways to engaging with AI.

Kenya: from maternal chatbots to helping farmers

The Kenya AI Strategy 2025–2030 aims to establish a strong digital infrastructure for AI. It outlines priority AI applications envisioning, for instance, a maternal health chatbot conversant in local dialects, AI-driven teaching systems and translating agriculture data into “farmer-friendly audio formats in local languages.” However, as the strategy document makes clear, current infrastructure limitations “could slow down AI adoption and limit the scalability of AI driven solutions across the country.”

It also acknowledges that Kenya’s current regulatory environment for AI is “fragmented”, causing “inconsistencies and inefficiencies” in AI  governance, in turn, “making it challenging to create a cohesive strategy for AI development and deployment.”

“Like many countries, the government was initially slow to act on AI governance,” says Monica Okoth, Assistant Manager in the Electrotechnical & ICT Division at the Kenya Bureau of Standards (KEBS). “Policy makers didn’t fully understand the technology, so there were no policies for a long time.” She says it was the private sector which led the government to draw up an AI strategy and that KEBS was brought into the process largely through private sector engagement.

Data ownership is a challenge

In particular, there are concerns about data ownership. Big tech companies have previously collected indigenous knowledge or language data without full disclosure, in addition to plundering the continent’s resources for their requirements. (so-called techno-colonialism). “For example, there was a case where biometric data was collected from communities without proper government knowledge or transparency. The issue isn’t the data collection itself since we need more data to reduce bias,” Okoth says, “but ensuring full disclosure, consent, and clarity on how the data will be used.

Another concern is whether AI can take into account the diversity of African populations. “Kenya has 42 languages, so developing models for all of them would require huge investment,” she adds. “Because English and Swahili are widely spoken, some therefore argue that we don’t urgently need large language models (LLMs) in every indigenous language.”

Digital safety, privacy and security were part of a more than USD 1 billion investment led by Microsoft and an Emirati AI company to build and run a data centre for East Africa in Kenya’s Rift Valley. The project promised to operate a “trusted data zone” under which local data would be stored and governed by local laws.

However, since being announced in 2024, the project has yet to kick off, with reports of development being tied up in red tape and concerns about how much electricity from the grid is needed to run it.

Role of KEBS and standards 

Recognizing the importance of international standards, KEBS joined as a participating member of ISO/IEC JTC 1/SC 42: Artificial Intelligence, in 2020. “We pushed for standards to be embedded in Kenya’s national AI strategy, because without them, implementation would be weak,” Okoth explains. “One of the government pillars now includes developing standards for conformity assessment and responsible AI.”

KEBS has already adopted several international standards and is developing local ones. For example, Kenyan Standard KS 2007:2025, a Code of Practice for AI, outlines non‑functional requirements and includes a full clause on trustworthy AI. It assigns responsibilities across the data providers, developers, software managers, and end users.

“Implementation is challenging if standards aren’t enforced or understood, so we also focus on conformity assessment. With donor support (such as funding from the UK government) we’re developing AI and AI‑data maturity models. These help organizations, especially start-ups and government agencies, self‑assess their readiness for responsible AI adoption or development. We’re also creating playbooks to guide them,” Okoth says.

Nigeria: building AI based on wide variety of languages

With over 500 languages, Nigeria is one of the most linguistically diverse countries in the world. A LLM currently in development aims to capture and build on this resource. N-ATLAS, short for Nigeria’s African Tongue Language AI System, is designed to handle multiple languages native to Nigeria, including Yoruba, Hausa, Igbo, and Nigerian-accented English.

Launched last September, the centrepiece of Nigeria’s National Artificial Intelligence Strategy (NAIS), it is a public-private initiative led by the National Information Technology Development Agency (NITDA), the National Centre for AI and Robotics (NCAIR) and local AI firm Awarri.

“N-ATLAS is designed to understand and generate Nigeria’s diverse voices, ensuring that the country’s linguistic richness is preserved and reflected in the global Artificial Intelligence (AI) ecosystem,” the Federal Ministry of Communications, Innovation & Digital Economy says.

For instance, farmers could use voice-based queries in Hausa to access weather forecasts or market advice. Igbo-speaking students might interact with AI tutors in their native tongue, improving learning outcomes in areas with low literacy rates. Or the LLM could be tuned to customer service where people’s queries “might differ significantly due to cultural and linguistic factors.”

The NAIS acknowledges existing structural challenges in the way of achieving its vision to make Nigeria a global leader in ethical and inclusive AI innovation. To upgrade its data centre and edge computing infrastructure, there are plans to offer tax breaks to investors. To build up its software capacity, organizations like Data Science Nigeria (DSN) offer upskilling and mentorship programmes. The Nigeria AI Scaling Hub is a research incubator focused on AI solutions in health, education, and agriculture, funded by the Gates Foundation. Additionally, the national curriculum for basic and secondary schools has been revamped to incorporate learning about AI, robotics, and cyber security. As DSN put it, “Africa cannot negotiate its place in an AI-driven world without the technical workforce required to hold that position.”

South Africa is leading the way

By most measures, South Africa is the continent’s leading country when it comes to digital and AI development. One report ranks South Africa as the only African country exceeding the global average of working age adults using AI tools. Another identifies more than 725 AI start-ups in the country at the end of last year, significantly more than the next best on the continent (Nigeria with 456).

Its National AI Policy Framework, released in 2024, placed “a strong emphasis on a human-centred approach in AI systems”, stipulating that applications should augment rather than replace human decision-making. And yet plans to cement this as national policy had to be put on hold earlier this year after it was revealed that some of the rules in the draft were AI-generated. At least six of the document’s 67 academic citations were found not to exist, suggesting that the draft did not in fact have sufficient human oversight.

“We need to make sure we have all the right voices at the table: people deeply immersed in AI, but also people in policy, law, and other areas,” explains Professor Benjamin Rosman, Director, Machine Intelligence & Neural Discovery, [JF1] at Wits University told CNBC Africa.

He is tasked with leading an advisory panel to the government to develop a new national AI policy. In his interview to CNBC Africa, he stressed that people working in the field were already seeking to bring in different perspectives to their work. Designing something “future-proof” should stand them in good stead, he explained.

Much of the investment in South Africa’s data centre and cloud infrastructure has come from overseas. Notably from Microsoft, which recently pledged a USD 329 million expansion in cloud infrastructure and AI training for Africa’s youngsters (a cumulative USD 1,4bn investment in recent years). Google and Amazon also all operate major platforms in Johannesburg and Cape Town.

These moves by overseas tech companies are viewed with a degree of suspicion by several south African pundits involved in the field. Rosman, for instance, queried whether South Africa should be happy relying on US technology and that subscription fees for it are leaving the country. “The country needs to build its own critical mass, widen participation, and ensure it is aligned with global best practices around AI, and plot a way forward,” he advised.

Egypt is an early adopter

The aim of Egypt’s National AI Strategy (2025-2030) is to have AI contribute as much as 7,7% of total GDP by the end of the decade. Achieving this would require training 30 000 specialists, supporting 250 startups, and expanding use of AI tools to over 35% of the total population (40 million people).

Egypt was an early mover on the continent, having issued an early AI framework for growth in 2021. The second edition is “much more ambitious” with a significant shift from solely using foreign technology to creating a sovereign AI ecosystem.

Core to that goal is development of Egypt’s first nationally built and owned LLM, called Karnak, which was announced at the AI Everything MEA summit in Cairo last February. Karnak is intended to “preserve linguistic and contextual relevance, especially for Arabic,” and to underline Egypt’s push to become “a regional AI hub linking Africa, the Levant, and the Gulf states.” 

Applications powered by Karnak include a personalized AI tutor supporting Egyptian history education in high schools, a specialized translation tool and a suite of natural language processing engines for colloquial Arabic.

Both editions of the AI strategy put ethical and responsible use of the technology at its core. As established in the Egyptian Charter for Responsible AI, the state is adapting international standards like UNESCO’s Recommendation on the Ethics of AI and ISO/IEC 42001, the International Standard for AI management systems, to local contexts, emphasizing human-centeredness, transparency, fairness, accountability, and privacy.

The work of African standards organizations 

Several initiatives have emerged recently to develop cross‑border AI standards for Africa.

EAC (East African Community) recently called for the responsible development of AI and stronger regional collaboration to drive inclusive growth and innovation. AFSEC (African Electrotechnical Standardization Commission) is creating a new technical committee on AI, including AI-enabled products and services, with specific consideration of African regional requirements, conditions, and priorities. It will be launched next month at its 10th General Assembly.

ARSO (African Regional Organization for Standardization): established a technical committee on information technology with a proposal for a subcommittee to deal with emerging technologies such as AI and IOT in May of this year.

“The goal is to create Africa‑centric AI standards,” explains Okoth. “It doesn’t make sense for each country to reinvent the wheel when resources are limited. Many African countries share similar cultural and economic contexts, so pooling resources to develop continental standards is more efficient and impactful.”

 


No comments:

Post a Comment