Go To Main Content

As the wave of generative AI sweeps across the globe, how can government departments embrace technological innovation while upholding safety and ethical bottom lines? The Taiwan E-Governance Research Center (TEG) held its "Annual Digital Governance Research Results Presentation" on May 6, 2025, with the first session focusing on an in-depth discussion of the "Artificial Intelligence Governance Framework for Governments." The session was chaired by Keng-Yu Yang, Deputy Director General, Department of Digital Government, Ministry of Digital Affairs (moda), and featured Hsien-Lee Tseng, Deputy Director of the E-Governance Research Center, who shared the latest research findings. Together with experts from industry, government, and academia, they analyzed Taiwan's AI governance blueprint.

 

International Perspective: From Risk Management to Impact Assessment

Deputy Director Hsien-Lee Tseng pointed out that global AI governance has formed three major camps: the EU, emphasizing human rights and regulations; the US and UK, pursuing extreme development and low regulation; and China and Russia, guided by national sovereignty. It is noteworthy that during the Trump administration, the US shifted its governance mindset from "risk" to "impact," treating AI as a general technology application subject to accountability. The research team observed that Taiwan's current AI governance is moving towards a balanced approach that emphasizes human rights protection, digital rights, and public sector transformation.

 

Taiwan's Practice: The Current Status and Challenges of AI Applications in the Public Sector

The research team used GPT to collaboratively analyze over 100 AI application projects in the Taiwanese government, finding that AI has penetrated diverse scenarios such as fire image recognition, sewage pipe inspection, and even tap water leak detection. However, the public sector also faces six major gaps in deploying AI, including funding, regulations, and personnel qualifications.

To assist organizations in more systematically adopting AI, the Digital Governance Research Center has proposed "Eight Key Elements of AI Governance":

  1. Awareness: Recognizing that AI can improve efficiency and effectiveness.
  2. Literacy: Enhancing colleagues' ability to distinguish between AI illusions and information security risks.
  3. Model Evaluation: Establishing an internal evaluation mechanism.
  4. Compliant Use: Ensuring compliance with laws and regulations.
  5. Formal Deployment: Implementing practical applications.
  6. Education and Training: Continuously strengthening colleagues' operational capabilities.
  7. Effectiveness Evaluation: Tracking the tangible impact of AI.
  8. Continuous Iteration: Using a PDCA (Plan-Do-Check-Act) cycle to refine processes based on feedback.
 

Experts Suggest: Give AI a Touch of "Humanity" and "Rule of Law"

Experts offered valuable suggestions for this framework from different perspectives:

1. Humanity and Categorization:

Chih-Chien Jau, Director of the Department of Statistics, Ministry of the Interior, suggested that the governance framework should be more "human," using easily understandable categorization methods (such as data collection, analysis, application, and provision), and shared 90 AI application experiences compiled by the Ministry of the Interior.

2. Technology and Education:

Jyi-Shane Liu, Professor & Dean, College of Informatics, National Chengchi University, likened AI to a "new Manhattan Project," emphasizing that AI's impact on civilization is overwhelming, and that AI-literate talent should be systematically cultivated through "AI+X" credit programs.

3. Law and Defense:

Ken-Ying Tseng, Partner, Lee and Li, Attorneys-at-Law, emphasized from a legal perspective that confidentiality is paramount, especially when using generative AI. She stressed the strict prohibition of inputting confidential information and the importance of national security and personal data protection.

 

Future Outlook: Sovereign AI and Public Value

Looking to the future, Deputy Director Hsien-Lee Tseng stressed the importance of "Sovereign AI," including the establishment of a Traditional Chinese corpus and avoiding vendor lock-in. The Ministry of Digital Development is also actively drafting the "Basic Law on Artificial Intelligence" and promoting AI evaluation centers, aiming to achieve a balance between innovation and trustworthiness.

This presentation not only showcased fruitful research findings but also conveyed a core message: AI is not only a technological breakthrough but also an opportunity to transform public governance models. The ultimate value of AI governance lies in freeing civil servants from repetitive tasks and allowing them to engage in more valuable decision-making and innovation.

Attachments