Reflections on the AfricAI Conference

The AfricAI Conference 2023 allowed researchers, industry executives and policymakers from across Africa and around the world to explore the boundless possibilities of artificial intelligence (AI) on the continent. As one of the presenters in the workshop by the African Observatory on Responsible AI, in particular, Meta Bursary Researchers: Insight into AI Ethics, AI Policy/Governance through an AfricanLens, I showcased my research on using privacy-preserving synthetic mobile money transaction datasets for financial fraud detection. The conference provided a platform for knowledge exchange and opened doors for collaboration and innovative solutions.

Presentation

Overall, the conference was an enriching experience that expanded my understanding of AI's potential in Africa. The Meta session explored smart townships, AI and education for children in Africa, the use of automated decision-making in migration management and my own research topic, using AI in financial fraud detection. Here are key points I noted from my audience at the conference.

·      the need to contextualise AI for policy making, drawing inspiration from our presentations.

·      the need to involve relevant stakeholders and policy makers during AI development.

·      the need for further discussions with individual researchers on the uptake of the findings, privacy and how nontechnical people on the continent could join AI research.

Lessons Learned

Industry executives, researchers and individuals working in the telecommunications and financial sectors engaged with me to further discuss the details of my research. These one-on-one discussions and other engagements at the conference underscored three key lessons for me.

Collaborative Spirit

Generally, the conference fostered an environment in which researchers, practitioners, and policymakers could share their insights and experiences. This exchange of knowledge enabled me to gain a broader perspective on the challenges and opportunities facing AI in Africa and emphasised the importance and benefits of collaboration.

Impactful Applications

The conference showcased various applications of AI in sectors such as healthcare, agriculture, and languages.Witnessing the transformative impact of AI in these domains was truly inspiring. It reinforced the notion that AI has the power to address some ofAfrica's most pressing challenges and shone a spotlight on solutions that have been developed locally.

Privacy-Preserving Approaches

Through discussions and presentations, I learned that privacy-preserving techniques are gaining traction in AI research. As privacy concerns continue to grow globally, AI systems must incorporate respect for individuals' privacy while still achieving accurate and meaningful results. This aspect aligns with my work in the development of trustworthy AI for financial crime prevention. I am passionately working in this domain and aim for leadership in the application of AI for financial services in the sub-Saharan context. Meanwhile, the discussion around privacy led to ideas on potential directions my research could take especially to benefit society. Below is a summary of the ideas generated.

Ideas on the Usage of Research Findings

1. Mobile  Money Crime (MoCrime) Vaccines

Development  of machine learning (ML) models for financial fraud detection using  high-quality synthetic datasets.

2. Simulation  Platform as a Regulatory Sandbox Environment

Measurement  of the true cost of fraud through simulation and evaluation of existing  regulatory policies and the evolving regulatory dynamics in mobile money  service.

3. PolicyBriefs

Issue  financial policy briefs to mobile money service providers, regulators and  forward-looking financial institutions.

Further Engagements: TheTrustworthy AI Consortium for Mobile Money Financial Crime Prevention

1. Consortium item: Trustworthy  AI

Details: Based  on fairness, transparency, robustness, stakeholder benefits and safety  including human oversight.

2. Consortium item: Focus

Details: Financial  crime prevention using high-quality synthetic data to train AI.

1. Consortium item: Purpose

Details: To promote collaboration, education and access to cutting-edge AI/ML solutions  to financial crime challenges in the sub-Saharan region.

4. Consortium item: Lead

Details: The  AI in Finance research group at the University has developed a product that  allows junior researchers and industry to generate simulations of mobile  money financial crime, facilitating use of privacy-preserving synthetic  datasets to train AI.

5. Consortium item: Stakeholders/ Participants

Details: Junior  researchers, Service providers, financial institutions, Government bodies,  Payment aggregators, Regulators (Central banks), Trade bodies, Low  Enforcement Authorities (LEAs), Consultants and Distinguished academics.

6. Consortium item: Requirements

Details: Members  to participate in discussions around Trustworthy AI for Finance, collaboration,  and financial crime vaccine trials.

7. Consortium item: How  to join?

Details: Express  interest and get notified to join the consortium

8. Consortium item: Potential  challenges

Details: Low  uptake of machine learning models for financial fraud prevention on the  continent. Shortage of talent and lack of awareness in the industry.

Sessions Attended

In addition to presenting my own research, I attended several other thought-provoking sessions.

Lacuna Fund - Datasets, Use Cases, and Sustainability

Training high-value, high-impactAI systems for addressing persistent problems on the continent requires large amounts of datasets. Datasets that could aid crop disease detection, precision farming, and smart irrigation were presented that could promote food security and revolutionise agriculture in Africa.

Inclusive AI - Open Language Data for Sustainable Development

This session highlighted potential contributions of AI in preserving African languages. Given that the continent has many ethnic groups, the development of AI systems that take into account the existence of various languages has the potential to ease communication, enable patients to communicate with doctors and improve overall service delivery. Openness in the development of AI in this domain could eliminate biases.

Healthcare-Related Sessions

These sessions educated attendees on issues of AI-powered diagnostics, personalised medicine and telemedicine initiatives that are improving access to quality healthcare, particularly in remote areas. The potential to save lives and enhance healthcare outcomes through AI was remarkable. These sessions had more work in progress and experimentation.

Emerging Opportunities

My presentation at the AfricAI conference opened doors to new opportunities. Industry executives and policymakers expressed a keen interest in collaborating with me to develop trustworthy solutions for financial crime prevention. Such collaboration has the potential to drive real-world impact and address the pressing need for secure financial systems across Africa.

The interactions with fellow researchers from different African countries were invaluable. They shared their experiences at graduate school, which will significantly enhance and propel myPhD study to the next level. Most of the junior researchers were finalists in their PhD programmes.

Conclusion

The AfricAI conference, the first international conference I have attended in person, underscored the immense potential of AI to reshape Africa's future. I am grateful for the connections made, the lessons learned, and the inspiration gained. I look forward to participating in future international confer