人工智慧應用的社會影響評估

Although the substantial impact of AI on business and society is conspicuous, there is lack of research discussing how to measure such impact. Hence, this proposal aims to bridge the research gap by applying the approach of Social Return on Investment (SROI) to measure the social impact caused by AI applications. SROI is an outcome-based method developed from cost-benefit analysis and social accounting which is focusing more on the created values than financial benefits. It is also based on the Theory of Change and extensive stakeholder engagement with assigning monetary values to the amount of positive or negative change created. A social impact measurement framework with value map will be developed for AI applications. The framework contains steps of scope and stakeholder identification, mapping outcomes, evidencing and valuing outcomes, establishing impact, calculating SROI, and sensitivity analysis. Potential stakeholders, well-defined outcomes, outcome indicators, financial proxies, uncertain factors such as deadweight, attribution, displacement, and drop off during the measurement procedures for AI applications are addressed in this research as well. Case studies on real-world AI applications will also be conducted to demonstrate the accountability and feasibility of the proposed framework.