Attention to racial, social, and economic inequalities has forced all industries, including financial services, to focus on addressing disparities within organizations and in business operations. Algorithms and AI models, which are increasingly being used for their power to better predict risk, recommend future courses of action, or help with market growth, may potentially also introduce bias risk raising important questions related to ethical use, trust, fairness, and accountability.
Our Point of View paper, Algo risk: Mitigating potential bias, discusses how developing or enhancing the appropriate compliance controls and governance frameworks around algorithms and AI models can help to mitigate differences across access, treatment, and impact while also bolstering corporate compliance and ESG initiatives. These efforts should include:
- Defining principles and metrics
- Evaluating the coverage and suitability of the underlying data
- Conducting monitoring, testing, and reporting
- Aligning ethical and responsible use with the company culture.