LIRNEasia concluded its first workshop (30 March, 24’), its first and part of a larger series of initiatives on ‘Artificial Intelligence for Social Good’; intended to raise interest and awareness on the potential of AI to benefit society at large. The event featured a keynote speech by Dr. Romesh Ranawana, Chairman of the national AI strategy committee (Sri Lanka, 2024-28). Dr. Ranawana outlined Sri Lanka’s tactical roadmap for AI development, including its synergies with existing digital policies, the challenges still ahead. Looking towards the next 2Qs, we can expect the strategy draft to launch in May followed by the establishment of the national AI center.
Dr. Ranawana reiterated that the strategy “[…] must continuously evolve, be responsible, inclusive, and collaborative—beyond the paradigms of traditional software development; we should consider developing AI solutions that not only address problems but also ensure their sustainability […]”. He noted that public awareness could play a key role in driving legislative and corporate action towards the responsible development and deployment of AI technologies towards prioritizing societal benefits and ethical considerations.
Education as always, was a critical highlight, “[…] we need to enhance AI literacy in all education levels, with concrete steps to cultivate the necessary talent. In the next two years, Sri Lanka should take steps to train data engineers, AI engineers, retrain public servants including teachers, add new degrees, and make changes to existing business degree curricula”.
On the technical end, the question of open data continues to be a challenge—with reference to the lack of available data and collaboration infrastructure required to build and train machine learning systems for use-cases in public sector governance. “An AI strategy cannot be like other strategies, where you have a plan for the next 10 years and then follow it […]”.
Merl Chandana, lead for LIRNEasia’s Data, Algorithm and Policy team, drew attention to the interdisciplinary nature of addressing societal challenges, calling for the integration of vulnerability assessments into adaptive plans and the importance of collaboration with [development] partner organisations.
One example concerned the outdated delineation of ‘urban’ and ‘rural’ in Sri Lanka and its inability to reflect the evolving demographic landscape. Important, because these classifications have impact budget allocations, project implementations, and poverty assessments and other public policy. This was followed by snippets on LIRNEasia’s work to map urban areas more accurately analysing satellite imagery with computer vision, and re-examining these definitions, and ensuring they correspond to ‘ground truth’, as defined by UN Habitat Global.
Furthermore, Merl delved into another initiative done by LIRNEasia regarding heat vulnerability, highlighting its significance in understanding the impacts of heat exposure on across demographics. He spoke about how factors such as pre-existing medical conditions, age, and access to resources contribute to individuals’ susceptibility to heat-related complications. Through comprehensive data gathering and analysis at the smallest administrative unit level, LIRNEasia is working to prioritise interventions for the most vulnerable populations.
Dr. Kasun Amarasinghe, Senior Research Scientist at Carnegie Mellon University, highlighted the diverse applications of AI across various sectors. His presentation showcased how AI-powered diagnostic systems have demonstrated remarkable potential to revolutionise patient care in the healthcare sector. For instance, AI algorithms can analyse medical images such as X-rays and MRIs with incredible accuracy, assisting radiologists in detecting and diagnosing conditions. Amarasinghe also discussed the role of AI in the education sector, citing the example of chatbots like “Rori,” deployed on platforms like WhatsApp to facilitate efficient information dissemination. Furthermore, he elaborated on how AI can aid in the research sector by identifying urban and rural patterns through the analysis of aerial photographs and heatmaps, addressing challenges in accurately categorising areas due to governmental amendments in location-based hierarchy levels. All examples were studies in which Dr. Kasun was directly involved in or adjacently through his colleagues at Carnegie’s DSaPP lab.
The hands-on session in the afternoon led by Dr. Kasun was an interactive coding session based on a social-good project conducted by Carnegie’s DSaPP. The session’s objective was to help students understand the complications of deploying machine learning models to situations affecting citizens’ lived experiences. This included temporal constraints, addressing bias, trade-offs between precision and recall, and the explainability of the models.
Please note that more workshops and speaker sessions (via Zoom) are upcoming. We encourage you to stay tuned for updates on them by following our social media channels.
