Isabel Hou, Deputy Minister, Ministry of Digital Affairs (MODA), Taiwan

1) How do you use your role to ensure that technology and policy are truly inclusive?

In my role, ensuring the "inclusiveness" of technology and policy is one of the core missions of digital governance.

This involves three levels: alignment with human-centric values, the popularisation of digital literacy, and mechanisms for substantial participation. 

First, regarding policy design, we adhere to the fundamental principle of being "human-centric." This means that the AI development we promote must robustly serve humanitarian values and safeguard human autonomy.

Even if AI can provide precise data analysis, policy formulation or critical judgments should still be the responsibility of human decision-makers to ensure that people's freedoms and rights are not overly interfered with by algorithms. 

Second, we are committed to deepening digital citizenship literacy.

I believe that in this era, everyone needs to know what they want to achieve and actively explore to find the right tools. The AI literacy courses we promote include basic knowledge and understanding (e.g., understanding how AI works and its limitations), application and operational skills (e.g., learning to use various AI tools), and ethical reflection (e.g., paying attention to data bias and privacy protection).

By elevating the AI literacy of the entire population, we can ensure that the ubiquity of technology does not create a new digital divide, allowing more people—not just a minority of "AI elites"—to utilise these tools. 

Finally, in practical application, we pay attention to the "the strong get stronger, the weak get weaker" effect of AI.

We promote collaboration among large, medium, and small enterprises to allow AI to bring substantial social benefits to disadvantaged groups in specific fields, such as improving the quality of healthcare or solving labor shortages in small service industries. 

2) What’s a moment in your career when you saw firsthand how technology or a new policy changed a citizen’s life for the better?

Over my twenty-five-year career, I have experienced massive technological advancements, from the internet and mobile platforms to the current era of AI.

"Democratising technology, technologising democracy" has been the central theme of my efforts.

I have always positioned myself as a "translator," hoping to make technology understandable to non-tech people, make law understandable to non-lawyers, and enable people from different backgrounds to collaborate and co-create through tech tools to solve real human needs. 

Since joining the g0v (gov-zero) community in 2012, I have deeply felt the power of "collective intelligence."

For example, the Cofacts project combines open data and AI to assist in analysing potential misinformation reported by the public, connecting it with suitable professional editors to make information more transparent and credible. 

I was even more impressed by the birth of the "Mask Map" during the early stages of the pandemic in 2020. After the government opened up pharmacy inventory data, community engineers created various applications in a very short time.

An elderly person once told me that he was originally anxious about not being able to buy a mask, but with his grandson's help, he learned to use the Mask Map on Line and finally bought one successfully. That sense of "peace of mind" is the most direct value of technology. 

3) What was the most impactful project you worked on this year, and how did you measure its success in building trust and serving the needs of the public?

This year (2025), the most influential project I am focusing on is driving Taiwan's AI governance model and risk framework.

I believe that the advent of the AI era is like a new wave of industrial revolution, and the government must possess "agile" adjustment capabilities to respond to this century's changes. 

With the upcoming passage of the "AI Basic Law," the government will establish an AI risk framework.

Our governance model will be "Japanese in spirit, but more flexible," distinguishing itself from the market-oriented model of the US, the strict regulations of the EU, and the centralised control model of China.

I expect that by the end of next year (2026) at the earliest, Taiwan's AI governance will demonstrate concrete results. 

Regarding measuring effectiveness and building trust, I focus on the following points: 

I. Ensuring Human Autonomy and Transparency: Policies require that AI decisions must be verifiable and human-controllable, ensuring citizens' right to know and choose regarding how AI is used.

In financial or smart taxation applications, we must provide citizens with control over their personal data and analysis results, and offer mechanisms for manual processing or review when necessary. Disclosing transparency is the foundation for building public trust in AI. 

II. Cross-Domain Collaboration and International Alignment: We emphasise strengthening internal and external government connections, promoting public-private partnerships, and deepening international alignment.

One measure of success is whether we can successfully establish a cross-agency digital governance model and establish dialogue and cooperation mechanisms with major economies in fields like AI, cybersecurity, and data governance, allowing Taiwan's experience to contribute to the international community. 

III. Meeting Industry Needs: Our goal is to promote the adoption of AI across all trades and industries to drive industrial upgrading.

Success is measured by the experience of "teaming up to fight monsters" (collaborative group efforts), helping those already taking action to find markets and assisting them in "going to sea" (expanding internationally).

This is not just a technical success, but a success in market and global positioning. 

4) What was one unexpected lesson you learned this year about designing for real people? This can be about a specific project or a broader lesson about your work.

The most unexpected lesson was this: The empowering effect of AI depends entirely on an individual's domain knowledge and the depth of their critical thinking. 

In the AI era, I often mention the formula "N × X = Z". Here, N represents your original domain knowledge capability, X is the AI empowerment multiplier, and Z is the final capability enhancement.

This made me profoundly realise that if your N is zero, then no matter how large X is, the result is still zero. 

This went far beyond my previous understanding of tool usage in the legal or tech worlds.

For example, while AI art tools are powerful, friends without a background in art design cannot reach the level of those with the necessary domain knowledge and aesthetic judgment, no matter how they use AI. 

At the same time, I have more deeply realized the importance of "Verify, don't blindly trust."

Content generated by AI may be very fluent, but it can "spout nonsense" (hallucinate)—this is seen as a feature of the model, not a bug. For instance, there was a case abroad where a lawyer used ChatGPT to find legal precedents and submitted fake cases, resulting in a fine.

This serves as a strong reminder that AI will not take responsibility for you; you must verify it yourself. For professionals, our core expertise—verification and auditing capabilities—has ironically become even more important and valuable because of this. 

Furthermore, in the process of interacting with generative AI, I found that to give precise instructions, I became more and more like a machine, while the machine started acting more and more like a human.

This shift in the human-machine relationship is something I find truly fascinating. 

5) We hear a lot about AI. What's a practical example of how AI can be used to make government services more inclusive and trustworthy?

If AI applications in government services are established on principles of responsibility and human autonomy, they can greatly enhance inclusivity and trustworthiness. 

Here are a few practical examples: 

I. Enhancing Inclusivity: 

A. Simplifying Administrative Processes and Improving Healthcare: When designing and deploying AI systems, we should evaluate whether they bring actual well-being to the public, such as simplifying administrative processes or improving the quality of healthcare.

Currently, the top five AI applications introduced by Taiwanese enterprises and government departments include process automation/optimisation, employee personal assistants, and customer service bots.

This helps reduce repetitive work for civil servants, making services more efficient and allowing resources to be invested in more complex areas requiring human touch. 

B. Bridging Language and Cultural Gaps: By promoting "Sovereign AI," we can train models that better understand Taiwanese culture and terminology, even including all of Taiwan's national languages, such as Amis and Atayal.

This ensures that AI services do not produce bias or omissions due to the influence of dominant cultural data.

For example, the Ministry of Digital Affairs has provided computing platforms to support industry players in developing multi-language direct translators (such as Taiwanese-to-Indonesian direct translation), which is a massive aid for cross-cultural communication and inclusion. 

C. Applications for the Elderly: Starting from "human-centric" design, we explore how to combine AI technology with the actual needs and challenges of elderly care environments, creating more humanised and intelligent solutions to improve the quality and efficiency of elderly care while respecting user needs. 

II. Enhancing Trustworthiness: 

A. Risk Analysis and Management: AI can be used for budget assessment and fiscal risk monitoring to help the government allocate resources and ensure long-term fiscal stability.

In financial services, AI applications include fraud analysis and investigation, which help combat digital scams. 

B. Maintaining Human Decision-Making as the Final Gatekeeper: We must maintain humans as the final gatekeeper for decisions.

When the public sector uses generative AI, the information produced must undergo objective and professional final judgment by the case officer; it cannot replace the officer's independent thinking, creativity, and interpersonal interaction. 

C. Fairness and Non-Discrimination: Policies should consider fairness and non-discrimination issues, establishing a risk classification framework for AI applications and decisions.

At the same time, the government should establish application accountability mechanisms and improve the verifiability and human controllability of AI decisions. 

To subscribe to the GovInsider bulletin, click here.

6) How are you preparing for the next wave of change in the public sector? What new skill, approach, or technology are you most excited to explore in the coming year?

The impact of AI gives me a sense of urgency. I believe the AI era requires the government to possess "agile" adjustment capabilities.

As the Deputy Minister, I am preparing for the next wave of change through three core aspects: institutional design, talent cultivation, and infrastructure. 

I. Institutional Design: I am most looking forward to seeing concrete results from the AI risk framework by the end of next year.

We are pushing for regulatory adjustments following the "AI Basic Law," which will be a significant milestone in Taiwan's digital governance.

This involves assessing risks, planning response measures, and setting usage norms or internal control mechanisms. 

II. Talent and Methods: I will continue to play the role of "translator," acting as a bridge between policy and public opinion, using clear language to let society understand policy directions.

In terms of methodology, I look forward to exploring cross-domain collaboration, using my past industry experience of "teaming up to fight monsters" to promote public-private partnerships and international layouts.

Simultaneously, we must face the challenges of a declining birthrate and industrial structural adjustments by promoting "balanced development" of software and hardware, especially in AI cybersecurity and dual-use (military-civilian) applications. 

Regarding New Skills and Technologies, I am most interested in compute-saving technology and responsible AI development:  

I. Compute-Saving Technology: Although we are committed to increasing computing power, resources are ultimately limited.

The development of DeepSeek has given us a revelation, showing the possibility of using "brainpower to supplement computing power." It is necessary for our R&D personnel to study and learn from this open-source technical approach.  

II. Responsible AI Use and Development: We must continuously explore how to use AI more responsibly, including selecting the right tools, using AI correctly, using AI content prudently, and actively providing feedback to AI.

This is a process of continuous optimisation. 

7) What advice do you have for public sector innovators who want to build a career focused on serving all citizens?

My advice for new-generation talent can be summarised as: Cultivate the courage for "Curiosity, Domain Knowledge, and Team Collaboration." 

I. Cultivate Continuous Curiosity and Domain Knowledge: Technology changes too fast; this is an era of lifelong learning. Success in AI depends on your domain knowledge (N).

If you want to innovate in the public service sector, you must first have "problem awareness" regarding the social issues you care about.

Do not just be satisfied with the superficial answers AI gives you, but know how to verify and validate its output. 

II. Actively Participate in Communities and Practice: "Learning AI isn't like learning history; it's more like learning to swim"—you have to get your hands dirty.

I highly recommend everyone get in touch with the g0v community. Even if you don't know how to code, you can participate in open-source projects and find like-minded partners to work with.

Innovation in public service is rarely a "one-man project"; it requires team building, complementing each other's skills to solve real problems together. 

III. Focus on Solving Real, Hard Problems: I quote Academician H.T. Kung's advice: Solve difficult problems, don't look for easy ones.

You need imagination, skills, experience, and courage. Think about how to use AI to solve practical problems that impact society, such as flood warning systems or auxiliary tools for environmental law enforcement. 

IV. Strengthen Basics and International Language Skills: Although generative AI can assist with translation, mastering English is still very important, as the latest knowledge and most educational materials are still primarily in English. 

8) Who inspires you to build a more inclusive and trustworthy public sector?

The force that motivates me to devote myself to this field comes from "a passion for the intersection of technology and law" and "inspiration from the community." 

I started as a lawyer, but in 2000, I stepped into the tech world by participating in the open-source community.

In this process, I learned a lot of tech knowledge and began focusing on digital governance issues like innovative tech, open data, and open government, along with their legal regulations.

This cross-disciplinary experience cultivated my specialty in "explaining law to non-lawyers" and "explaining tech to non-tech people." 

Furthermore, in discussions on AI governance, I deeply resonate with former White House OSTP Director Arati Prabhakar's emphasis on "ensuring that the risks and benefits of AI are balanced," and Melinda Gates' statement that "when women's voices are absent, technology cannot truly serve everyone."

These viewpoints reinforce my belief: We must ensure that technological development is democratised, truly realises diverse participation, and injects humanitarian values into AI design and application. 

It is precisely this enthusiasm for digital governance legislation, participation in open-source communities, and persistence in human autonomy and multicultural values that inspired me to join the administrative team, hoping to push Taiwan's AI development onto the right path. 

9) If you had an unlimited budget, what would your dream project be?

If I had an unlimited budget, my dream project would be to establish a "Sovereign AI Data Ecosystem" and a "High-Inclusion Language Model" centered on Taiwan's culture and languages. 

Currently, the Chinese training data for international Large Language Models (LLMs) is mostly Simplified Chinese, with Traditional Chinese data accounting for less than 0.1%.

This leads to a risk: if we fully adopt these models, Taiwan's terminology, imagery, society, culture, and values face the danger of being "ignored" or even disappearing. 

Therefore, my dream project would concentrate on: 

I. Systematically Building a High-Quality Traditional Chinese Corpus: Systematically organising and establishing text datasets with unique Taiwanese characteristics, incorporating Taiwan's diverse languages such as Indigenous languages and Hakka. This is not limited to text but includes multi-modal image or audio-visual data. 

II. Training a High-Inclusion Sovereign AI Model: Using these high-quality, culturally value-laden data to train an open-source model that "understands Taiwan better," maintains the country's multicultural values, and can simultaneously contribute to the world. 

III. Perfecting Data Governance Laws and Mechanisms: Although we are pushing for the draft "Data Innovation and Utilisation Development Act," if I had unlimited resources, I would thoroughly resolve legal barriers to data acquisition and copyright disputes, building a complete framework where both public and private sectors can use data safely, legally, and efficiently. 

This would make Taiwan a global model for responsible AI development and cultural inclusivity, ensuring our digital future is defined by us, not dominated by powerful external cultures. 

10) Outside tech, what excites you the most?

Setting aside the tech field, what excites and fills me with passion the most is my family, and extending from that, next-generation education and humanistic care. 

I am a mother of three. I am deeply aware of the anxiety parents and students feel amidst the AI wave. Therefore, how to help the next generation find their own direction in this era filled with "fog" is something I care about deeply. 

I believe the value of education lies in cultivating children to become "Brave Ones" who explore the future. This includes cultivating their curiosity, imagination, and the confidence and courage to face the unknown. 

Although my profession spans law and technology, I always emphasise the importance of humanities and social sciences.

I encourage everyone to pay attention to cross-disciplinary collaboration and integration beyond just human-machine collaboration, and to apply our wisdom and compassion to creating places that are beneficial to the world. 

This adherence to humanitarian values, respect for multiculturalism, and expectation that the next generation can navigate the future autonomously and critically are the greatest driving forces supporting my continued dedication to public service.