“AI is not supposed to and will not replace people. But people who use AI effectively will replace people who do not. Organizational leaders need to set in place change management processes, training, and upskilling that focus on people to be able to see lasting positive change that technology, such as AI can bring.”

In the rapidly evolving landscape of artificial intelligence (AI), the intersection of innovation, social impact, and ethical governance has become increasingly pivotal. Rositsa Zaimova, CEO and Co-founder of Dalberg Data Insights (DDI), embodies this intersection through her leadership, vision, and commitment to driving positive change with her team that puts ethical practices at the core of designing and building AI systems.  

She has a background in Economics and International Management and has been recognized as a Forbes 30 Under 30 in Europe. Her leadership at DDI has been instrumental in harnessing AI for tangible social impact. She is an East Africa Acumen Fellow and an Ambassador for One Young World.

DDI projects are about more than technological advancement, they seek to effect technological transformation. Tell us more about your approach.

I am interested in a future where AI empowers communities, fosters inclusive growth, and prioritizes ethical governance. By deeply understanding user contexts, ensuring accurate responses, and setting high standards for AI readiness, we address biases, paving the way for fair and diverse AI technologies. 

A great example of this is our work towards the development of a groundbreaking decision-support tool supported by the Bill & Melinda Gates Foundation. This tool, equipped with a user-friendly Gen AI-enabled interface (like a chatbot), simplifies complex data insights for public health officials, bridging the gap for users with varying levels of data expertise. There are two concrete things we have put in place to ensure the tool is contextual and relevant. First, it is accessible on WhatsApp, which is the preferred means of access to information for our users. Second, we have set up a benchmarking mechanism, which allows users to rank the relevance of the answers they get, which feeds back the model with information used to continuously improve the accuracy and relevance of the responses.

While AI ethical concerns are a hot topic, what elements are particularly unique to the social sector?

This approach is about more than convenience; it is about how we enable technology to translate to tangible improvements in outcomes and quality of life for communities at large. For example, we designed and developed decision support tools for Ministries of Health that monitor immunization indicators. These in turn allow public health officials to understand data such as the trends in vaccine coverage rates over time within specific populations or geographic areas. That helps decision makers make informed decisions.

When considering the impact of your projects, what metrics or indicators do you prioritize to measure success, and why do you find them valuable?

Measuring the success and impact of an AI or data project requires a comprehensive approach that takes various dimensions into consideration. These include performance, ethical considerations (in relation to access, performance, bias, and transparency), user satisfaction, and societal impact. Our team has been very intentional in ensuring accuracy, fairness, and interoperability for the AI and data systems and tools that we develop. For example, when building an AI analyst that allows public health officials to interpret data from the Health Management Information System (HMIS), we were responsive to using WhatsApp as an interface for the chatbot since it was the preferred communication platform for our users. Working together with all relevant parties is essential to ensure that AI technologies are not only cutting edge but also culturally responsive and truly advantageous for those users.

What are you most excited about?

Imagine a community health worker in Kenya, equipped with the right information to provide vital care and counselling when needed, or a head of a vaccination program in Madagascar effortlessly interpreting real-time changes in vaccine distribution for a specific district, or a farmer in Senegal gaining a deeper understanding of and access to government services by simply posing a question. These are just a few examples of how technology can drive impact. The good news is it is already happening. The bad news is that it is just not happening at the pace that is required to ensure that the digital and AI divide is not continuing to grow.

Helping our partners embrace the transformative power of artificial intelligence is central to our vision for the future.

We understand it can be overwhelming for many to understand all the opportunities and risks that come with AI. Being able to help impact-driven teams identify the right problems to solve with AI and walking the journey with them from design to responsible development and rollout of AI applications is what I am super excited to be working on.

Can you describe an instance where partnerships and alliances were pivotal in advancing a particular project or initiative?

We are building the Tanzania Information Statistical System Portal (TISSP) as an online data analytics and visualization application that pulls together data from different government agencies into a single portal.  We will provide a dynamic platform for generating customized analyses on cross-sectoral data. As a result, users can understand, for example, how funding decisions impact development outcomes. We are only able to do this impactful work because of the collaboration with UNICEF Tanzania, the Line Ministries, and the Statistics Bureaus (NBS and OSGS), who all play a critical role in the design, development, and rollout of such a system.   

Most recently, in 2023, DDI supported MESH—a social enterprise in Kenya that offers an online community platform connecting young entrepreneurs in the informal sector—in defining fairness criteria and de-biasing techniques to mainstream a fairness approach in their service platform. By applying Responsible AI frameworks and monitoring fairness metrics such as Equal Treatment, Demographic Parity, and Equalized Odds, the MESH team is able to identify and mitigate bias for prioritized AI use cases. 

Ultimately, finding sustainable business models will be key in piloting and scaling AI innovations for impact-driven organizations.

What’s next?

As we move forward into an AI-driven future, it’s crucial that the progress we see is guided by ethical innovation. This means that our advancements must be in harmony with universal human values, societal standards, and the needs of everyone, including diverse, marginalized communities. Our new AI platform is designed along these terms. Augment Impact is designed to serve as a trusted AI and data platform that enables the design and development of custom AI agents tailored to specific use cases and grounded in the unique individual, organizational, or geographic contexts of our partners.

Contact: rositsa.zaimova@dalberg.com