The Access Gap is an Innovation Gap: Democratizing Healthcare through AI and Inclusive Education

By Anie Udofia | Technology For Good

When landmark studies reveal that certain commercial AI systems used in healthcare assessments report error rates of up to 34% for darker-skinned women, it is not merely a glitch, but a prescription for inequality (Buolamwini et al., 2018). It reveals the underlying homogeneity of the creators designing the networks, serving as a drastic example of how the ‘closed-loop of creation’ in health technology actively perpetuates inequity. This is the central flaw of our paradigm: we deploy advanced tools to communities without any possibility of tailoring, giving them a finished, rigid product. The result is a disturbing contradiction, where algorithms capable of screening millions systematically exclude a vast segment of the population from the skills needed to adapt, or invent them.

Therefore, to rectify the embedded biases in health AI and achieve true equity, the most critical intervention is to democratize the very power of creation. It is essential to shift public and private investment from solely deploying finalized tools to systematically building the innovation capacity within these sectors of society. Hence, I propose a three-pillar framework for policy-makers and institutions, such as the World Health Organization, national public health agencies, and mission-driven public partnerships: Demystify, Incubate, and Mechanize.

The first pillar, Demystification and Foundational Literacy, directly attacks the disempowering opacity of AI. The black boxes, or the initial unexplainability of AI, prohibit people from interacting with its components. As Seymour Papert argues in Mindstorms, comprehension of complex systems is established through constructionist learning as opposed to instructionism, pointing towards the superiority of hands-on and project-based health AI education (Papert, 1980). Accordingly, public investment must extend beyond generic coding lessons to incorporate culturally contextualized education that provides transparency and explainability for high-stakes AI decisions in healthcare settings. An example of such education includes the reasoning behind reduced performance on darker skin tones and equipping members with the foundational skills to examine such datasets and co-design inclusive data collection protocols for their own communities. This approach embodies what scholars Vossoughi and Vakil brilliantly term “pedagogies of practice”, rooted in community-identified concerns, as opposed to abstract “pedagogies of deference”, where technical facts are accepted passively without understanding (Vossoughi et al., 2018). Therefore, this pillar actively contributes to the goal of shifting community members from being passive patients into informed evaluators and potential co-designers.

However, literacy solely provides language for critique, not the actual tools. Thus, the second pillar, Targeted Innovator Incubation, is employed to provide the resources to translate deep understanding into a culture that fosters community-owned innovation, unique to the particular demographic. This encourages creative experimentation, modelled after the most effective learning environment most of us have experienced: kindergarten. In Lifelong Kindergarten, Mitchel Resnick, the MIT Professor behind Scratch, describes this environment as thriving on the 4 P’s: Projects, Passion, Peers, and Play (Resnick, 2017). Applied here, a participant of Pillar 1, now educated on the biases of AI, could advance solutions through tangible projects, such as a low-cost capture kit, fueled by passion to solve locally-identified health crises, developed in collaborative peer teams refined through a “path to creativity”, or play. Therefore, this pillar tackles the “Lost Einsteins” effect by narrowing opportunity gaps, reducing the disproportionate likelihood that children from high-income families, who are up to ten times more likely to become inventors than their low-income peers, will dominate innovation outcomes (Bell et al., 2018). Consequently, inventive success is driven more by demonstrated ability rather than by socioeconomic status.

To translate this cultivated capacity into a tangible impact, the final pillar, Mechanism for Targeted Impact, provides the launchpad and market. This entails establishing “Health Equity Grand Challenges”, focused on providing enticing incentives for solving community co-defined problems through competitive grants (Peña et al., 2019). An example of a prompt for a Health Equity Grand Challenge could be: “Develop an accessible AI tool for early detection of diabetic retinopathy in X rural region”. Such a prompt would centralize resources on urgent matters of a particular location without instigating ethical dilemmas that jeopardize the order of the targeted zone (Vryn et al., 2025). 

Figure 1 - Grand Challenge Framework by Liz O’Brien

The basis of this Grand Challenge model has been proven by institutions such as the USAID and the Gates Foundation, which applied this framework to generate distinct innovations that positively contributed to solving global obstacles. In particular, the successful development of the PharmaJet Tropis® needle-free injection system for polio-endemic regions in Nigeria was the direct result of the original Grand Challenges in Global Health initiative by the Bill & Melinda Gates Foundation. This system was so influential that Nigeria is still currently using fIPV for the implementation of Polio treatment to this very day, highlighting the considerable scope of impact these challenges have (Mohan et al., 2025). Moreover, when these challenges equally adhere to the Community-Based Participatory Research (CBPR) principles, community members are not only recognized for their ideas but also deemed as equal partners throughout the entire process, giving a voice back to direct beneficiaries (Peña et al., 2019).  In addition to seeding ideas through Grand Challenges, public procurement of these community-developed solutions is indispensable. As highlighted by Monteiro et al. (2024), procurement serves a powerful purpose, developing sourcing streams that prioritize equity first and providing a stable market for these technologies, as well as a consistent flow of value. Thus, an active government role is crucial, as the intentional purchases of community-developed solutions by these bodies design a sustainable market that ensures these innovations can continue to make a difference.

Overall, this analysis reveals that the path to true equity in healthcare AI is not exclusively found in distributing better technology but in democratizing its creation, where the medical fate of a young girl in an underserved community is no longer in the hands of a biased algorithm. The three-pillar framework of Demystify, Incubate, and Mechanize provides the actionable blueprint to breaking the ‘closed loop of creation’, setting the stage for innovation to serve all, built by all. 

References: 

Bell, Alex, Raj Chetty, Xavier Jaravel, Neviana Petkova, and John Van Reenen. 2018. “Who Becomes an Inventor in America? The Importance of Exposure to Innovation.” Opportunity Insights. Accessed December 26, 2025. https://opportunityinsights.org/wp-content/uploads/2019/01/patents_paper.pdf.

Buolamwini, Joy, and Timnit Gebru. 2018. “Gender Shades: Intersectional Accuracy Disparities in Commercial Gender Classification.” Proceedings of Machine Learning Research 81: 1–15. https://proceedings.mlr.press/v81/buolamwini18a/buolamwini18a.pdf.

Mohan, Deepa, Mercy Mvundura, Sarah Sampson, Victor A. Adepoju, Gambo B. Bakunawa, Chukwudi Umebido, Anthony Ekeh, et al. 2025. “Evaluating the Impact of Needle-Free Delivery of Inactivated Polio Vaccine on Nigeria’s Routine Immunization Program: An Implementation Hybrid Trial.” Vaccines 13 (5): 533. https://doi.org/10.3390/vaccines13050533.

Monteiro, Bruno, Ana Hlacs, and Pedro Boéchat. 2024. “Public Procurement for Public Sector Innovation: Facilitating Innovators' Access to Innovation Procurement.” OECD Working Papers on Public Governance, No. 80. Paris: OECD Publishing. https://doi.org/10.1787/9aad76b7-en.

Papert, Seymour. 1980. Mindstorms: Children, Computers, and Powerful Ideas. New York: Basic Books.

Peña, Vanessa, and Chelsea A. Stokes. 2019. Use of Grand Challenges in the Federal Government. IDA Document NS D-10699, Science & Technology Policy Institute, Institute for Defense Analyses. Accessed December 29, 2025. https://www.ida.org/-/media/feature/publications/u/us/use-of-grand-challenges-in-the-federal-government/d10699final.ashx.

Resnick, Mitchel. 2017. Lifelong Kindergarten: Cultivating Creativity through Projects, Passion, Peers, and Play. Cambridge, MA: MIT Press. https://www.academia.edu/34937979/Lifelong_Kindergarten_Cultivating_Creativity_through_Projects_Passion_Peers_and_Play_From_A_to_X.

Vossoughi, Shirin, and Sepehr Vakil. 2018. “Toward What Ends? A Critical Analysis of Militarism, Equity, and STEM Education.” In Education at War: The Fight for Students of Color in America’s Public Schools, edited by Arshad I. Ali and Tracy Lachica Buenavista, 117–140. New York: Fordham University Press. https://www.researchgate.net/publication/329887415_Toward_what_ends_A_critical_analy_sis_of_militarism_equity_and_STEM_education.

Vryn, Michelle Flores, and Meena Das. 2025. “Building Community-Centered AI Collaborations.” Stanford Social Innovation Review, February 5, 2025. Accessed December 28, 2025. https://ssir.org/articles/entry/community-centered-ai-collaborations.