Artificial intelligence (AI) and the know-do gap: Reflections from Kenya
Author: Leah Mwai
Background
The rapid advancement of artificial intelligence (AI) promises to revolutionize evidence-based healthcare practice. However, significant disparities in technological infrastructure, research and development (R&D) capacity and concentration of AI talent have resulted in its uneven adoption, favoring practitioners and communities in high income countries and high resource contexts.
As a knowledge and evidence professional working in Kenya and across Africa, I have witnessed diverse opinions ranging from excitement, skepticism, distrust, pessimism, denial, hopelessness, to utter panic when talking about AI. In my opinion, these conflicting views highlight critical and inconvenient realities that the evidence community must confront with regards to meaningfully leveraging AI in its work. Notably, evidence suggests that developing countries, encompassing most of Africa, currently contribute less than 5% of total global AI grants and research outputs, and are therefore seriously lagging in the critical resources required for effective AI adoption . Further, whilst the ethics of AI is an increasingly popular topic, Africa has hardly been at the table in academic and policy discussions, even when it comes to ‘global’ and ‘intercultural’ approaches. Biases and stereotypes about Africa persist.
Perhaps, this excerpt from the foreword of the book titled ‘Responsible AI in Africa: challenges and opportunities’ aptly encapsulates the status and prevailing attitudes on AI in Africa:
‘Responsible AI in Africa sounds almost like an oxymoron, and at best comes across as a marginal topic…. But AI is becoming more popular and like everywhere, it is already changing the world—also in Africa. There are start-ups, there is research, and there is innovation. Africa is also a place where the future is made’.
Remarkably, signs of progress are evident. Optimists highlight that ‘the arrow in Africa’s quiver’ is its young population which can help catalyze and leapfrog Africa’s AI transformation , and various countries are intensifying AI adoption activities. Kenya for instance is making efforts to implement its recently developed national AI strategy for 2025-2030 , and the need for data that can inform evidence-based actions has never been greater or more pressing. Below, I highlight four key gaps, opportunities, and recommendations where collaboration amongst evidence-based health care practitioners can support effective and equitable implementation of AI:
Key gaps, opportunities and recommendations
1. Promote wider collaboration to advance and implement minimum standards for reporting, evaluating and using AI generated knowledge
Whilst AI is increasingly being leveraged in various evidence-based healthcare processes, unrepresentative data, inappropriate AI algorithms and models can result in systematic favoritism or prejudice in evidence generation, translation and use, leading to discriminatory and detrimental outcomes. The negative consequences of AI biases extend beyond individual patients to the potential reinforcement and perpetuation of inequalities, and are likely to be most pronounced in marginalized communities and low income and low-resource contexts where there are lower levels of AI development and capacity. If such biases are not addressed, those from underrepresented groups, countries or contexts may receive suboptimal benefits or experience adverse outcomes, and this can erode trust in AI and evidence-based health care approaches. Whilst various efforts have been made to develop protocols, guidelines and standards, particularly based on FAIR principles, more efforts are required to ensure that contextual knowledge and lived experiences of practitioners and communities in low-income and low resource country contexts are adequately incorporated and reflected. Such standards should incorporate local values, ethics and cultural contexts by genuinely collaborating with local practitioners and communities. Collaboration can help to address this gap and pave the way for more inclusive and equitable AI.
2. Support capacity development, collaborative learning and knowledge exchange through dynamic knowledge platforms and communities of practice
Developments in AI are nascent and fast moving, and published sources are proving insufficient for providing information on the most recent contextual questions or gaps. Further, whilst contextualized AI skills and educational materials are required to enable practitioners to navigate the complexities of ethical AI adoption and use, existing evidence-based health care training courses and curricula do not adequately integrate these aspects. My professional observations on this are in agreement with documented evidence that highlights significant skills gaps, sparse and fragmented guidance that focuses primarily on completion of disparate tasks rather than comprehensive and coherent skill development, and lack of a standardized, adaptable competency framework for practice. Consequently, most practitioners lack confidence and feel unprepared to integrate AI into their work. Addressing these gaps is crucial not just for improving technical proficiency, but also fostering collaborative learning and interdisciplinary approaches, and a deeper understanding of the ethical and professional implications of AI. This presents an opportunity to collaborate in developing and revising training materials, and promoting continuous learning, dialogue and exchange through structured but dynamic knowledge repositories and communities of practice. Doing so will strengthen the overarching local and global ecosystem and contribute to a more equitable and inclusive global AI landscape.
3. Enable knowledge generation, translation and use in local languages and locally relatable formats
Despite language and context being some of the most significant exclusionary filters in shaping the framing, translation and use of knowledge, they are rarely (if ever) given the same attention as other methodological concerns in the practice of evidence-based health care. At the same time, whilst AI is making is easier than ever before to execute many professional evidence-based health care tasks (including creating graphics, summarizing research and making presentations), current AI systems predominantly operate within the framework of Eurocentric or English-speaking datasets and models, and are likely to exacerbate existing biases and inequalities. This preponderance of English-centric AI systems makes it more essential than ever before for evidence-based healthcare practitioners to intentionally promote and accommodate authentic local voices, and culturally grounded ways of expression, for instance by enabling inquiry, documentation, and dissemination of knowledge in local languages, local voices, and locally relatable formats. In Kenya for instance, initiatives such as Chini ya Mbuyu , Corona University and Talking Walls have been progressively promoting creative, co-generative, participatory and community-led approaches (that include digital storytelling, social media content creation, visual art , poetry and storytelling) to help break the ice, build confidence, and facilitate honest dialogue on complex and sensitive topics such as reproductive health, vaccine uptake, gender based violence and urban safety.
4. Re-imagine research rigour, embracing complexity and uncertainty
While aspects such as reproducibility and peer review remain core and valuable aspects of the scientific method, local communities do not consider them the only path to truth, and have their own contextually developed ways of knowing such as distilling and passing knowledge over generations, and testing, validating and refining knowledge communally through lived experiences. Further, consistent with ‘One Health’ principles that emphasize interconnections between human, animal and ecosystem health, issues in healthcare have become more complex and interdisciplinary, and there is an increasing demand for knowledge products that address broader questions and provide actionable insights drawn from a broader, more integrative perspective on existing evidence. At the same time, AI is transforming the way in which evidence is commissioned, generated and consumed (including how literature reviews are conducted). Evidence practitioners should embrace this as an opportunity to collaborate in establishing updated dimensions of rigour that incorporate human values that local communities truly care about such as sustainability, wellbeing, justice and resilience. Further, amid AI’s disruptive impact on the evidence and knowledge synthesis and translation process, there is an urgent opportunity for practitioners to collaborate on establishing updated, more relevant formats and standards for systematic reviews and other evidence and knowledge products and outputs (particularly determining and delineating the role of human judgement vs AI, and enhancing our ability to explore and make sense of complexity across disciplines, systems and time).
Conclusion
I have highlighted important and strategic areas and opportunities for global, cross-country and local collaboration that can strengthen the ecosystem, and contribute to a more dynamic, equitable and inclusive global evidence-based AI practice landscape. The call for the evidence community to coordinate and collaborate more intentionally has never been louder, more earnest, and more urgent. Let’s do it!
Key Takeaways
Currently, developing countries account for less than 5% of total AI grants and research outputs, and are disadvantaged in terms of critical resources for progressing AI adoption. Yet, signs of progress are evident and many countries including Kenya are intensifying AI adoption activities.
Stronger, more structured and more dynamic cross-country partnerships, knowledge platforms and communities of practice are critical for strengthening capacity and establishing a more equitable, inclusive and sustainable global AI policy and practice landscape.
The adoption of AI provides a critical opportunity and juncture for reimagining research rigour and standards in a way that accounts for the increasing uncertainty and complexity of the knowledge production and translation process, and centres local community values, outcomes and impact.
AI is making it more essential than ever before for practitioners to embrace locally-resonant co-creation and community-led models of knowledge generation, and there are already some inspiring and progressive case studies to learn from.
References and resources
Cisco and Carnegie Mellon University Africa (2025). AI and the Workforce in Africa: Realizing the Region’s Potential Through Public and Private Sector Collaboration : https://newsroom.cisco.com/c/dam/r/newsroom/pdfs/Cisco-CMU-Whitepaper_AI-and-the-Workforce-in-Africa.pdf
Damian Okaibedi Eke; Kutoma Wakunuma; Simisola Akintoye (2023) Responsible AI in Africa: Challenges and Opportunities: https://link.springer.com/book/10.1007/978-3-031-08215-3
Phillip Mwaniki. Artificial Intelligence readiness: Kenya ranked 8th in Africa as AI set to create jobs, boost continent's economy. Citizen Digital. August 14, 2025 02:30 (EAT): https://www.citizen.digital/article/artificial-intelligence-readiness-kenya-ranked-8th-in-africa-as-ai-set-to-create-jobs-boost-continents-economy-n367970
Kenya artificial Intelligence Strategy 2025-2030. Kenya Ministry of Information, Communication and The Digital Economy: https://ict.go.ke/sites/default/files/2025-03/Kenya%20AI%20Strategy%202025%20-%202030.pdf
Huerta, E.A., Blaiszik, B., Brinson, L.C. et al. FAIR for AI: An interdisciplinary and international community building perspective. Sci Data 10, 487 (2023). https://doi.org/10.1038/s41597-023-02298-6
Arianna Valentini and Alep Blancas. The challenges of AI in higher education and institutional responses: is there room for competency frameworks? UNESCO International Institute in Latin America and The Caribbean. 2025: https://unesdoc.unesco.org/ark:/48223/pf0000394935.locale=en
Chini ya Mbuyu initiative by Akili Dada: https://akilidada.org/chini-ya-mbuyu-podcast/
Talking Walls and Corona University initiatives by Hope Raisers: https://www.hoperaisersinitiative.com/covid19
To link to this article - DOI: https://doi.org/10.70253/PTAX1954
Disclaimer
The views expressed in this World EBHC Day Blog, as well as any errors or omissions, are the sole responsibility of the author and do not represent the views of the World EBHC Day Steering Committee, Official Partners or Sponsors; nor does it imply endorsement by the aforementioned parties.
