IndabaX Kenya AI Summit

Introduction

Responsible AI in politics and democracy

The theme of the summit was “Akili Unde Katika Siasa na Demokrasia”, which emphasized the need for responsible AI and ethical use of Artificial Intelligence(AI) in political and democratic processes. It emphasized on development and application of AI technologies in ways that uphold transparency, fairness and the protection of civic rights since artificial intelligence has highly impacted good governance. AI tools can be used to detect and limit the spread of misinformation during elections by flagging fake news on social media platforms.

Fueling the Future: The Workshops

The summit began with an insightful keynote speech by John Kitunyi, a student at Dedan Kimathi University of Technology who shared a compelling vision of a smarter judicial system in Kenya through the application of AI. In his talk titled Enhancing Access Efficiency and Fairness in Kenya Judicial System, he discussed how AI can be used in addressing persistent challenges in the legal sector, by automating tasks such as legal document analysis, case management and decision support to promote fair access to justice.

Milicent Ochieng, Research Scientist at Microsoft, delivered a powerful talk on the limitations of current large language models (LLMs) when applied to multilingual and culturally diverse settings, particularly African languages. She highlighted findings from MEGA and IrokoBench benchmarks, which revealed that despite models like GPT-4 performing well in English, they significantly underperform in low resource languages. Her call to action urged the AI community to build models that reflect the full spectrum of global languages and cultures.

Dr. Joseph Muguro, Research Affiliate at Center for Data Science and Artificial Intelligence (DSAIL), delivered an informative session on use of Natural Language Processing (NLP) and Artificial Intelligence (AI) to enhance road safety in Kenya. He highlighted alarming trends in road traffic accidents linking them to human error, poor infrastructure and data collection gaps. He showcased the potential of AI tools like vision-language models (vLLMs) and dashcam analytics for detecting traffic offenses in real time.

Mr. Alfred Kondoro also had an insightful presentation on building AI agents that reflect African voices and values within the education sector. He highlighted that students are active users of AI tools but they face challenges such as poor internet access and tools that do not reflect their culture. He emphasized on the importance of developing AI tools that are built using local data and designed to fit African context.

Another engaging talk delivered by Mr. Jay Patel, CEO for Jacaranda Health, that focused on real world application of AI In maternal healthcare. He showcased how their AI-powered platform supports expectant and new mothers by providing timely personalized health information through mobile messaging in high resource languages such as English, Swahili, Yoruba and Hausa. He also shared plans to expand the platform’s reach by incorporating African low-resource languages to better serve underrepresented communities.

Spotlight on Innovation: African Research project Showcases

After a series of insightful keynote presentations, IndabaX Summit transitioned into African Research Project Showcases. This session provided researchers with an opportunity to present their projects, which involved different domains such as healthcare, agriculture and language technologies.
Kelvin Mwita, Nurse, at a local clinic in Nyeri, showcased a project titled The Digital Stethoscope, drawing from his firsthand experience as a health provider in a rural healthcare setting. He highlighted the critical gaps in Kenya healthcare systems such as manual recording, outdated tools and lack of AI training from the frontline workers. His call to action encouraged the need for investment in public health dataset and digitizing health applications so as to improve health care delivery.

Robison Otoch, presented an impressive project on Building a Pretrained NLP Diagnosis Chatbot. He demonstrated how he leveraged publicly available dataset from Kaggle and Hugging face to fine tune a model capable of responding to health related questions. His step-by-step approach included data preprocessing, model fine-tuning, and deployment using a Flask API NLP Emergency Diagnosis Chatbot.

Mary Kariuki, Research Intern at Center for Data Science and Artificial Intelligence (DSAIL), Presented her ongoing work on Building a Health Dataset for Low-Resource Kenyan Languages. Her project aimed to address the language gap in healthcare by collecting and preparing high quality data for Kenyan local languages, which will support the development of a multilingual large language model that can improve access to health information for underserved communities in Kenya.

Edward Gichura, presented his innovative project on Predicting and Managing Maize Pests and Diseases through Weather Intelligence. He demonstrated how he used weather data from Open Weather API to predict pest and disease occurrences using machine learning models such as Random Forest and XGBoost. His results showed strong potential for building early warning systems that can support small scale farmers in making timely decisions.

Finally, there was also an ideathon, where participants had the opportunity to propose innovative project ideas they aim to undertake after the summit.

Highlights and Connections

1. Learning from experts: We were privileged to learn from fellow AI researchers and professionals during various sessions and workshops on how AI can solve real world problems in different domains.
2. Project acknowledgment: Project from DSAIL by Mary Kariuki was featured as one of the best during the summit at African Research Showcase.
3. Networking: We had an opportunity to connect with fellow researchers, industry experts, and innovators from across the region.