News
MODA inaugurates AI evaluation centre
The Digital Ministry has inaugurated its AI Product and System Evaluation Center, targeting language models for its initial assessment.
The centre will scrutinize ten critical aspects, including safety, fairness, and privacy. This initiative aligns with European and American AI standards and is supported by global tech giants like Meta and Microsoft.
The evaluation framework, developed in collaboration with institutions like the Industrial Technology Research Institute, will establish free verification services, with plans to gradually introduce charging.
Digital Minister Audrey Tang emphasized the need for AI products to align with societal expectations and local cultures.
The centre’s approach will not just test language models against a fixed set of questions but will probe to identify potential vulnerabilities in a dynamic and challenging manner. By 2025, Taiwan aims to align its AI evaluation technology with international standards, with domestic products passing these evaluations by 2026.
