2025.12.20
NewsCIEL Deputy Director Yueh-Ping Yang Invited to the Conference “Major Constitutional Challenges in the AI Era”
On December 20, 2025, CIEL Deputy Director Professor Yueh-Ping Yang was invited to attend the conference “Major Constitutional Challenges in the AI Era,” hosted by the ROC Constitutional Law Society. Professor Yang delivered a speech entitled “AI Discrimination and Explainable AI: Focusing on AI Credits,” offering an in-depth analysis of how AI-driven decisions affect financial fairness and how law and technology can work together to respond to these challenges.
Professor Yang first introduced the provisions on AI fairness set out in the Financial Supervisory Commission’s Guidelines for Artificial Intelligence (AI) Applications in the Financial Industry. He emphasized that “fairness” does not mean uniform equality or the absence of differential treatment, but rather differential treatment with legitimate justification. The key issue lies in whether such differential treatment is grounded on reasonableness and whether it results in systematic adverse impacts on specific groups with reasonable grounds.

Professor Yang further noted that while AI entails risks of discrimination, a more comprehensive understanding is that AI “may be more fair than humans, yet also more discriminatory.” The fundamental risk of AI stems from its “black box” decision-making process, and Explainable Artificial Intelligence (XAI) can play a critical role in addressing this challenge. Through techniques such as SHAP, LIME, and counterfactual explanations, financial institutions can analyze the key variables used in AI-driven decisions and the influence of individual variables on outcomes, thereby assessing whether protected characteristics are being used and whether such use can be justified.
In conclusion, Professor Yang proposed the “relative black box theory,” arguing that the integration of AI with XAI does not necessarily exacerbate discrimination, but may instead facilitate more objective solutions. Nevertheless, this approach requires supporting legal frameworks, including clearly defined protected groups in relevant legislation, permitting the use of protected characteristics for testing purposes, and empowering regulators to establish acceptable thresholds for differential treatment.
CIEL will continue to closely monitor the interaction between the evolution of AI technologies and the rule of law, promote financial fairness and justice in the digital age, and contribute to the development of a more resilient and transparent AI regulatory framework in Taiwan.







