BUILDING AN EQUITABLE DATA ECOSYSTEM FOR AI DEVELOPMENT IN GHANA
In October, GFA Consulting Group's Digital Innovation Unit and the Africa Center for Economic Transformation (ACET) convened a workshop at the 2024 Ghana Digital Innovation Week on Data Ecosystem for AI Development in Ghana. The panel featured some of the leading voices in that field, including Naa Lamle, a machine learning expert; Isaac Newton Acquah, Tech Sector Growth Lead for the Netherlands Trust Fund V; Godfred from Trotrolive; and Léandre Godefroy from d-Node.
Ghana has made remarkable progress in digital transformation, with internet penetration of around 70%, driving the generation of digital data crucial for a data-driven economy. The Ghana Data Protection Commission, under the Ministry of Communications and Digitalisation, recently held a validation workshop for the Ghana Data Strategy, which focuses on advancing the SDGs through better data governance, partnerships and innovation. This strategy is also in line with Ghana's forthcoming AI policy.
“However, building a robust ecosystem for data and AI development requires greater collaboration between the government, community, and the private sector, with the latter playing a pivotal role in driving innovation and investment in critical data systems,” says Blaise Bayuo, Senior Fellow and AI Technical Lead at ACET.
Takeaway from the groups' discussions and panel season
"We asked our panellists and audience to share their opinions on each pillar, and provide insights on the best approach to building an equitable data ecosystem for AI development in Ghana. The responses were mixed, with a stronger emphasis on the role of the community in pushing and engaging government to build foundational systems for data and AI," said Daniel Otto, Data Scientist at GFA's Digital Innovation Unit. As a result, seven key areas emerged where the experts see a need for change.
- Open datasets and platforms - Panel experts highlighted the challenges of accessing basic datasets from public institutions, even for non-sensitive sectors such as education. Despite the existence of the Right to Information Act in Ghana, public officials are often reluctant or unable to provide data. There was a general call for the government to adopt a data-first approach, making government data available through platforms and repositories for the development of AI models. Building the skills of civil servants to regularly update and manage datasets is also crucial. Naa Lamle shared the challenges she faced in accessing data from public institutions for AI model development, and believed this could help address the capacity of public institutions to curate and publish data. Participants suggested that governments could create revenue models to sustain such platforms, but the main focus should be on reducing access barriers. Improving existing projects, such as the Ghana Open Data Initiative, to provide up-to-date data could be a significant step forward.
- Standards and best practices for private datasets - In addition to public data, participants highlighted the role of private companies in contributing to the data ecosystem. However, guidelines and best practices for private companies to contribute to open data pools still need to be developed. There is also no organisation or coalition that certifies private data and guarantees its fair use by third parties. Panellists called for the creation of a community-led organisation to collaborate with data protection commissions to maintain and guide data contributions for AI development. Godfred Addai Amoako of Trotrolive and Léandre Godefroy of d-Node expressed their belief that the community should lead the collection of local datasets to complement government efforts.
- Tracing and incentivising voluntary data contributors - The global data and AI ecosystem has yet to develop mechanisms to reward private companies and individuals who voluntarily contribute data, or benefit when their data is used by third parties to create value. This could work in a similar way to software licensing, with different agreements for commercial and non-commercial use. Isaac suggested using blockchain as an infrastructure layer to track data use and reward contributors where appropriate, bringing transparency to the use of citizen and private data for AI development.
- Incorporate data and AI-related training into school and out-of-school programs - A data-first culture should be part of national skills development programs, preparing young people for the future and fostering local data-driven solutions. The role of data in digital transformation should inform curriculum design. While not everyone needs to be a data or AI expert, a basic understanding of their importance to the country's digital transformation is essential for widespread adoption. Companies such as Microsoft and Google offer virtual training in data and AI, and panellists called for public education on these opportunities, in addition to local university courses and out-of-school training programs.
- Develop citizen-centred AI governance regimes - AI governance should empower users to understand, control and seek redress when their data is used unfairly. Ghana's digital ecosystem needs practical governance structures that communicate with users, in local languages, about the implications of the growing data and AI revolution. Provision should be made for less digitally literate populations to access data rights and protection services in simple, understandable languages at their local assemblies. Future data and AI policies, strategies and programs must build on local values, needs and realities to create a fair and responsible data and AI ecosystem.
In addition, data and AI should represent and resonate with the diverse communities within Ghana's digital ecosystem. Government policy formulation and implementation could also benefit significantly from public engagement, which lays the foundation for the three pillars. The community needs to form data and AI coalitions for policy advocacy, implementation support and public education.
