On 14 January 2020, the Infocomm Media Development Authority (Infocomm), together with the Personal Data Protection Commission of Singapore (PDPC), and in collaboration with the World Economic Forum (WEF), adopted an implementation and self-assessment guide for organisations as a companion to the Model AI Governance Framework, which was adopted in 2019 and updated in 2020. The guide is intended to help organisations assess the alignment of their AI governance processes with the Model Framework, identify gaps and address them accordingly. It covers internal governance structures and measures, operations management, human involvement in AI-augmented decision-making, and stakeholder interaction and communication. This includes ethical considerations such as defining a clear purpose and weighing the benefits and expected costs of implementing an AI solution in developing a business case to deploy AI. Further, the guide recommends putting in place a governance structure to oversee the organisation's use of AI and conducting an impact assessment on individuals and organisations affected by the AI solution. The results of the impact assessment should be taken into consideration when implementing an appropriate level of human involvement in AI-augmented decision-making. Lastly, the guide addresses data quality, inherent bias, model training sets, and explainability, repeatability, robustness, reproducibility, and auditability of algorithms and models used.
Original source