Africa and Europe seek their own path in AI in science

Africa and Europe are seeking to chart their own course in the rapidly accelerating global artificial intelligence (AI) race, with a partnership built around scientific collaboration, shared values and solutions to societal challenges rather than technological advancement alone.

“The United States and China are moving very far ahead,” Maria Cristina Russo, the deputy director-general for research and innovation at the European Commission (EC), told a high-level meeting on AI in science between the African Union (AU) and European Union (EU) on 26 October.

“The EU and partners such as Africa need to develop their own way of doing AI,” she added.

The meeting, themed, ‘Building an Equitable and Inclusive Partnership for Scientific Discovery’, brought together policymakers, scientists, universities, research organisations and innovation specialists to explore how the two continents could harness AI for scientific discovery without reinforcing inequalities in access to technology, knowledge and its benefits.

Originally planned as a two-day gathering in Addis Ababa, Ethiopia, including a ministerial session, the event was converted at short notice into a one-day virtual discussion following the postponement of the AU-EU Research and Innovation Ministerial Meeting to 2027 in light of the security situation in Ethiopia.

DLR Projektträger, which coordinates the EC’s Global Service Facility, provided organisational support for the rapid transition. The expert-level discussions proceeded, but political consideration and adoption of their recommendations were deferred to the ministerial meeting.
 

This is the second in a series of articles by University World News – Africa providing independent coverage of initiatives emanating from the African Union-European Union High-Level Policy Dialogue on Science, Technology and Innovation, and the ongoing implementation of the 10-year AU-EU Innovation Agenda (2023–2033).

A partnership built on shared ambitions

The meeting’s programme described AI as a means of accelerating scientific discovery and supporting evidence-based decisions in fields ranging from public health and food security to climate change, biodiversity, water management and sustainable energy.

For Africa, the opportunities extend beyond scientific discovery to strengthening its research systems and advancing development.

“AI represents not only a technological opportunity, but also a way to achieve the aspirations of Agenda 2063, The Africa We Want,” Professor Saidou Madougou, the director for education, science, technology and innovation at the African Union Commission, (AUC), told the meeting.

He situated the emerging partnership within the AU’s Continental Artificial Intelligence Strategy and the overarching Science, Technology and Innovation Strategy for Africa, (STISA) 2034, which prioritise responsible technology development, scientific infrastructure and research cooperation.

Cooperation on AI between Africa and Europe builds on a longstanding partnership, with both continents seeking to benefit from their complementary scientific expertise and shared commitment to equitable research collaboration.

The AU-EU High-Level Policy Dialogue (HLPD) on Science, Technology and Innovation (STI) dates from 2010.

Its cooperation framework was strengthened by the AU-EU Innovation Agenda, adopted in 2023, which seeks to turn research into products, services, businesses and jobs across four priorities: public health, green transition, innovation and technology, and capacities for science.

Russo said the partnership had generated around €6 billion (about $6.7 billion) in joint research and innovation activities, including contributions to advances against malaria, HIV and tuberculosis through the European and Developing Countries Clinical Trials Partnership.

AI in science and science diplomacy, initially focusing on research infrastructures, is now being added to the cooperation agenda.

Europe has launched its Strategy for AI in Science, as well as the Resource for AI Science in Europe, or RAISE, an initiative intended to connect computing resources, scientific data, research talent and funding. Russo invited African researchers and innovators to participate in relevant Horizon Europe calls.

The ambitions extend beyond technological access.

Professor Denise Pires de Carvalho, the assistant director-general for sciences at the United Nations Educational, Scientific and Cultural Organization (UNESCO), posed a fundamental question: “How do we make sure that artificial intelligence expands the frontiers of science for everyone, not only those who already stand closest to them?”

She argued that inclusive AI required stronger research institutions, scientific freedom, better working conditions and opportunities for researchers, alongside ethical governance.

Gender disparities also needed attention. Women account for only about one-fifth of the data and AI workforce globally, she said, calling for measurable targets for women’s participation and leadership in joint research projects.

Who controls the knowledge?

The question of who creates, controls and benefits from scientific knowledge emerged as a central concern in the meeting’s cross-cutting plenary on data sovereignty and governance.

AI has advanced dramatically since the first AU-EU meeting devoted to AI in science in October 2025, at a pace that has outstripped human oversight.

Yet its benefits remain unevenly distributed, with persistent divides in access to representative datasets, high-performance computing and scientific expertise, alongside a concentration of influence over decisions about research priorities.

AI models trained predominantly on data from the Global North could be less effective, biased or even harmful when applied elsewhere.

Dr Phil Mjwara, South Africa’s Permanent Delegate to UNESCO and a former director-general of the country’s Department of Science and Innovation, illustrated the scientific consequences through AI-assisted drug discovery.

Major biobanks are dominated by people of European ancestry, he explained. This restricts the ability of AI systems to identify biological targets and predict how treatments will work across genetically diverse populations.

“A model cannot learn biology that is absent from its training data,” he said.

Including “African genetic diversity therefore stops being a matter of equity alone and becomes a matter of scientific necessity”, he added.

Mjwara called for a continental framework governing genomic data and benefit-sharing, including clearer arrangements for ownership, provenance, consent and recognition of those contributing data. He also proposed translating international ethical principles into contractual requirements for research funders.

Sabine Henzler, the director for strategy, work programme and resources at the EC’s Joint Research Centre, warned that AI could reinforce established research paradigms, narrow the questions scientists pursue and weaken human oversight of knowledge production.

She also cautioned that growing reliance on AI in scientific research could erode researchers’ independent judgment.

“We will, I think, be speaking much more in future about cognitive decline. Critical thinking is crucial, and we have not yet understood how we will be able to maintain it,” she said.

The implications go to the heart of scientific independence. AI can make research faster and more accessible, but its benefits depend on whose data inform the models, who can test their conclusions and whether scientists retain the capacity to challenge the results.

From principles to practice

Four parallel breakout groups examined how the AU-EU partnership in AI could address these challenges through data and infrastructure, scientific networks, innovation ecosystems, and skills and talent.

Their recommendations reflected a common concern: equitable collaboration requires more than access to technologies developed elsewhere.

Data and infrastructure: This group identified fragmented datasets, uneven interoperability and unequal computing capacity as major obstacles.

Its rapporteurs outlined proposals for shared regional computing facilities, better-connected repositories and arrangements enabling researchers to access trusted data across institutions and countries.

They cautioned against duplicating expensive infrastructure that could become obsolete and urged greater attention to sustainability, including lower-energy AI and models adapted to local languages. They emphasised that infrastructure, alone, was insufficient without human capacity.

Scientific networks: The main point emerging from this group was that AI cooperation should build on established Africa-Europe partnerships, rather than create parallel structures.

Existing networks highlighted include the Africa-Europe Clusters of Research Excellence (CoREs), jointly led by the African Research Universities Alliance, or ARUA, and The Guild of European Research-Intensive Universities; Global Health EDCTP3; and the AU-EU research and innovation partnerships on Food and Nutrition Security and Sustainable Agriculture, or FNSSA and Climate Change and Sustainable Energy, or CCSE. The latter two are long-term partnerships under the AU-EU HLPD.

Reporting the group’s conclusions, the rapporteurs emphasised African leadership, locally representative data and evaluation of AI tools in the contexts of where they would be used.

Innovation ecosystems: What happens after scientific discovery, was the focus of this group.

Its recommendations addressed the challenge of translating promising AI research into market-ready products and services, particularly where computing resources, technology-transfer mechanisms and investment are limited.

Participants proposed stronger links between universities, research institutions and businesses, alongside private and blended financing, appropriate regulation and safeguards for data ownership and vulnerable users.

The group also stressed the importance of investing in scientific methods and people, rather than assuming that progress always depended on acquiring the most powerful infrastructure.

Skills and talent: For this group, the starting point was that research talent already exists across both continents, but opportunities to develop and deploy it remain unequal.

One finding was that AI could enable small, geographically dispersed research groups to undertake more sophisticated science.

However, persistent visa barriers, insecure funding and limited career pathways continue to obstruct collaboration, even in the AI era.

The group proposed building on existing programmes, including the Africa-Europe CoREs, the EC’s Marie Sklodowska-Curie Actions, which support researcher training, mobility and career development, and the African Research Initiative for Scientific Excellence, or ARISE, which supports early-career African researchers with EU funding through the African Academy of Sciences.

It also proposed continental research training grants, more flexible mobility arrangements and small grants for early-career researchers.

A particular concern was brain circulation – enabling African researchers to gain international experience without permanently losing their expertise to institutions elsewhere.

The way forward

Across the discussions, the proposed approach was less about replicating the AI strategies of the world’s technology superpowers than combining scientific capacity across Africa and Europe to address specific problems.

The meeting’s recommendations will be consolidated for further technical consideration ahead of the postponed ministerial meeting in 2027.

Whether the partnership can deliver on its ambitions will depend on decisions yet to be made about investment, institutional leadership, data governance and who ultimately benefits from AI-enabled discovery.