Building AI with - not for - communities

Digest #84

From enabling early disease detection in healthcare to optimizing crop yields through precision farming techniques, AI is one of the most promising technologies in decades for those working in international development. However, technology cannot succeed in isolation; it needs people at its core. To unlock the transformative power of AI, two key groups must be engaged meaningfully: the communities these technologies are designed to help, and the governments which are so critical to AI’s regulation and uptake.

There are a multitude of ways to achieve this, from local advisory boards that include teachers, healthcare workers, or small-scale farmers in a community-led approach to AI design, to strong capacity development for decision-makers and civil servants. With properly funded and supported strategies, the possibilities for AI in driving development are limitless. Without these investments and collaboration, we risk technology causing more harm than good, reinforcing biases, or leaving marginalized groups even further behind.

Using human technology to improve artificial technology

We often hear of situations where AI is utilized without proper care and human oversight; one example that sticks with me is the Aadhaar biometric identification system. Aadhar’s AI-powered technologies were used to streamline public service delivery in India, including food distribution and social welfare programs. However, early implementations highlighted the risks of excluding local voices. Errors in biometric authentication and deactivation processes led to the denial of essential services for vulnerable populations and suspected identity theft, disproportionately affecting marginalized groups. This was primarily caused by the lack of inclusion of local stakeholders in the implementation of the system, leading to a massive underestimation of accessibility requirements for most local users, and ultimately stifling the ability of the AI technology to help those it was intended to serve.

But this does not have to be the way. One example of successful community engagement has been askNivi, a digital health platform that is making a meaningful difference in 15 countries in Sub-Saharan Africa and South Asia, including India, Kenya, and Nigeria. It provides personalized, real-time advice to users on topics such as family planning, vaccinations, and general wellness, through platforms like WhatsApp and SMS, making it both accessible and confidential. It has reached millions of users, many in areas where reliable health information is scarce, helping improve health literacy while encouraging the uptake of vital services. Crucially, askNivi’s success lies in its integration of human oversight, with local health professionals monitoring and refining AI responses to ensure cultural sensitivity and accuracy.

Health topics often drive stigma, but the platform provides a safe and anonymous way for individuals to ask sensitive questions and get the support they need. During askNivi’s deployment in Kenya, human feedback loops helped refine the chatbot’s responses to better align with local healthcare policies and practices. By incorporating regular input from local health professionals, the platform adjusted to cultural sensitivities—for example, avoiding stigmatizing language around sexual health—and improved the accuracy of referrals to nearby clinics. Without this essential oversight, the risk of disseminating misleading or inappropriate advice would have undermined trust and effectiveness. This highlights a broader point: involving local stakeholders who will bear the long-term impacts of digital and AI interventions—not as an afterthought, but as an integral part of the design and implementation process—is not optional, it is foundational to ensuring ethical, inclusive, and contextually appropriate outcomes.

Human-AI collaboration ensures that digital solutions align with the cultural and practical needs of the communities they serve. Including stakeholders early-on in the process builds local capacity, reduces dependency on external expertise, and ensures solutions are tailored to local needs, leading to more adaptable and sustainable systems. With consistency, this approach fosters trust and community acceptance and increases adoption rates and the impact of AI interventions. Without their input, even the most advanced technologies risk becoming ineffective or harmful, perpetuating a new form of technological colonialism.

Empowering governments to lead the AI revolution

Governments are another key stakeholder in the deployment of large-scale AI programs, and crucial in ensuring their equitable application. However, despite being ever present in government strategies and the media, AI remains an esoteric concept to many civil servants, particularly outside of technical departments. Through my work leading a digital skilling program for nearly 500 civil servants in Ghana, Malawi, and Senegal, I have seen how governments can serve as the foundation for ethical AI deployments when given the right tool for success, such as meaningful capacity strengthening opportunities. When these are not prioritized, and government stakeholders are unfamiliar with the intricacies of responsible data and AI themselves, they are unlikely to insist that human technology integration occurs and the inherent biases of technologies are allowed free reign.

One common challenge we encountered in our project across all three countries was an initial resistance to adopting new technologies, often due to limited digital and AI literacy and concerns about data misuse. Many civil servants found it difficult to see themselves as active participants in technology deployment, often assuming such tasks were best left to technical specialists. For many, this mindset was rooted in a lack of confidence in their digital skills or a perception that technology was too complex to align with their day-to-day responsibilities. However, through targeted training that focused not only on technical upskilling but also on their unique role in shaping outcomes, participants began to realize that their expertise in governance, policy, and local context was vital to the success of digital initiatives.

By engaging in real-world applications—where their decisions directly influenced outcomes—they started to view themselves as central to the process rather than peripheral to it, ultimately building confidence and a sense of ownership in the digital transformation journey. In Ghana, we supported representatives from the Ministry of Education to experiment with the integration of data and AI tools to enhance the accuracy of school monitoring systems. These trainings are achieving lasting change, and the Government of Ghana has continued investment in building capacity within key ministries and professional development for civil servants on data and AI. This shows how important it is to invest in both the human side of AI, as well as the technological, in order to reap the benefits.

Putting people at the heart of AI deployment

As technology races forwards and more processes become automated, we need to recognize where human involvement is non-negotiable. Ultimately, ensuring this happens is about empowering people to participate in the data and AI revolution and enabling them to act with confidence to transform how essential services are delivered. This approach demands collaboration across sectors, consistent investments in capacity-building, and a commitment to ethical practices.

At askNivi, we strongly encourage the entire data ecosystem to center people and their upskilling at the heart of AI deployments. It may not be the most “exciting” work but it is absolutely the most vital. In the absence of funding to support the acquisition of important AI and data skills, organizations in low and middle income countries may be forced to forego this activity altogether. So, it’s important that funders also do their part by availing financial resources that make effective use of human technology a reality. If we can ensure that AI is deployed with people at the heart of its strategy—both members of the communities it affects and the government decision-makers to support its application—I believe it can be transformative in addressing the global challenges we face.