Prof. Jerry John Kponyo shares insights on Ethical AI in Ghana
Prof. Jerry John Kponyo, Principal Investigator and Scientific Director of the Responsible AI Lab (RAIL) contributed to a panel discussion led by FAIR Forward at the Masterclass on Ethical AI during the Ghana Digital and Innovation Week in Accra. His presentation emphasised the ethical considerations crucial to AI development in Ghana, particularly as the nation embarks on its journey of AI adoption.
Prof. Kponyo stressed that, as AI solutions are being built, it is critical to ensure that ethical principles guide every stage—from the inception of projects to their deployment. He underscored that this approach will shape a future where AI is trusted, inclusive, and impactful, especially in healthcare, agriculture, and industry sectors.
He highlighted that AI solutions in Ghana must prioritise integrity, equity, and respect for individuals while always being mindful of their social impact.
During his presentation, Prof. Kponyo introduced the FACETS Framework, a comprehensive measure proposed by AI research labs within the AI4D initiative to assess the ethical dimensions of AI systems. The framework covers the entire AI innovation pipeline, from envisioning to deployment, ensuring responsible AI development at every stage.
Prof. Kponyo elaborated that the framework accounts for various stages of AI innovation, beginning with problem definition and stakeholder involvement, ensuring that AI solutions are relevant and ethically sound.
He shared some examples of ethical AI practices in RAIL’s projects:
- Mini-Grid System Project: In this project, RAIL’s team engaged stakeholders in an island community, seeking their input before implementing an AI-driven mini-grid system. They also obtained permission from the Ministry of Energy and local leaders.
- Crop Disease Detection Toolbox: RAIL collaborated with local farmers, ensuring that the toolbox addressed their specific needs. Multiple languages were implemented in the system, enabling farmers across different regions to use the tool without language barriers effectively.
- Intimate Partner Violence (IPV) Project: Ethical deployment was paramount in this project, as the app allows victims to conceal their presence on their phones and log in anonymously. This ensures their safety while using the app.
One of RAIL’s key goals is to develop affordable assistive technologies that empower differently-abled individuals. Prof. Kponyo reiterated that these technologies should be accessible to people from all economic backgrounds, ensuring inclusivity and equal opportunities for participation in society.
Despite these advancements, Prof. Kponyo acknowledged the challenges of developing ethical AI in Ghana, particularly:
- Low AI Adoption: AI is still in its early stages of adoption in Ghana, making it challenging to convince stakeholders to invest in data collection and large-scale projects.
- Labour-Intensive Data Gathering: Projects like the Crop Disease Detection Toolbox require extensive fieldwork to collect images from farms nationwide. Ensuring that data represents diverse regions and dialects is time-consuming and demands multiple verification rounds, cleaning, and processing.
Prof. Kponyo highlighted the importance of building AI systems grounded in ethical practices. As Ghana continues to explore AI solutions, ensuring fairness, transparency, and inclusivity will be crucial for building trust and driving long-term impact.
RAIL collects datasets focused on English spoken with a Ghanaian accent, moving beyond foreign accents. This approach acknowledges the linguistic diversity within Ghana, where various tribes influence how English is spoken.
Here is a link to the bot: https://t.me/NLPHelper_Bot
By focusing on diverse datasets, RAIL minimises the risk of bias and ensures that accent classification systems can effectively serve users nationwide, promoting wider adoption of AI technologies.
Thomas Vogele, Senior Researcher and Head of International Business Development at DFKI Robotics Innovation Center, Germany, highlighted that AI development must aim to benefit society. He discussed notable projects such as AI-based environmental monitoring and circular economy initiatives. The circular economy project, also called innovative recycling, uses AI-analyzed multispectral imaging to localise and classify objects, improving recycling rates for bulky wastes. Vogele also emphasised the significance of the comprehensive AI and robotics regulation under the EU AI Act. The legislation ensures that AI solutions and services are safe and transparent and uphold fundamental rights, aligning with broader European efforts toward responsible AI development.