Beyond the Tech: The AI Policy We Seem Not to Talk About Enough

Pretext

This blog builds on a perspective I shared at the Roundtable on Equitable and Responsible AI Access for All hosted by Microsoft on April 29, 2025. Policy discussions often center on infrastructure, regulation, and innovation — but I believe we also need to ask how people, especially those early in their careers or outside tech, can meaningfully engage with AI. This piece offers a human-centered lens on what truly inclusive AI policy requires.

Why Thailand’s Future with AI Depends on Human Capacity, Not Just Technology

As Generative AI (GenAI) reshapes industries, public services, and global competition, many policy conversations in Thailand still focus on infrastructure, regulation, and AI access.

But in my view, access is only the starting point. The deeper challenge is:

Who knows how to think with AI — and who doesn’t?

The real value of GenAI isn’t in using it like a search engine or chatbot. I believe its strength lies in becoming a Thinking Partner — a tool that helps us frame questions, explore ideas, challenge assumptions, and refine reasoning.

I believe this is not just a digital divide.

It’s a Thinking Divide — and it may quietly shape how competitive, inclusive, and resilient Thailand becomes in the years ahead.

What Makes GenAI Different — And So Valuable

GenAI models like GPT-4o are trained on a wide range of public knowledge — from science and law to business and culture.

I believe, GenAI today is like having the Library of Alexandria at your fingertips — not because it holds all the answers, but because it offers access to a wide pool of collective knowledge that humans can explore, interpret, and learn from.

This includes examples such as:

• National policies, government reports, and legislative summaries
• Research papers from scientific and engineering fields
• Medical publications and clinical best practices
• Business strategy documents, market analyses, and consulting reports
• Procurement guidelines, compliance manuals, and legal contracts
• News articles, educational content, speeches, and even cultural writing

In short, these models reflect a wide cross-section of how people around the world write, think, and make decisions.

GenAI offers access to a vast pool of collective knowledge — but it takes human skill to ask the right questions, interpret the output, and apply it in context.

Why Human Skill Is Now the Limiting Factor in AI Value

LLMs don’t work like search engines. They simulate thinking — generating text based on what language is likely to come next.

That means value depends on how well you prompt, and how well you evaluate the result.

It requires:

• Framing clear, specific questions
• Testing from multiple angles
• Recognizing quality and credibility
• Applying judgment and iteration

A Real Example: Prompting Nuclear Energy

Shallow prompt:

"What are the pros and cons of nuclear power?"        

• Output is generic.

Deeper prompt:

"Explain nuclear power trade-offs from (1) a climate activist and (2) a national security advisor."        

• Output shows tension, policy trade-offs, and deeper insight.

Same tool. Different prompt. Completely different value.

Shallow prompt:

"Write a marketing plan for a new product."        

• Output is generic, with vague steps and no clear context.

Deeper prompt:

"I’m a junior marketing analyst at a mid-sized FMCG company. Help me draft a 30-day product launch plan for a new beverage targeting urban millennials in Thailand. Include digital channels, messaging ideas, and basic metrics to track."        

• Output is tailored, actionable, and context-aware — offering not just content, but guidance.

Same tool. Different prompt. Much higher impact.

Three Structural Risks to Thailand’s AI Future

Thailand can benefit greatly from GenAI — if we prepare our people to use it wisely. But there are risks:

1. Language Disadvantage

Most top AI models are trained in English. Thai responses are often lower quality.

Thai users start with a knowledge handicap.

2. Narrow View of AI Literacy

Training coders alone is not enough.

We need AI for the many — teachers, civil servants, SMEs, students.

3. Weak Prompting Skills

Many treat AI as a shortcut. But without context, weak prompting leads to shallow or misleading results.

Without skill, we risk overconfidence in flawed outputs.

What a Human-Centered AI Policy Might Look Like

From what I’ve seen, Thailand’s AI strategy could evolve by investing in cognitive infrastructure — preparing people not just to access AI, but to think with it.

This isn’t just about training developers. It’s about helping more people — students, professionals, entrepreneurs — build the mindset and confidence to work with AI.

1. Government’s Role: Building Thinking Capacity

Ideas worth exploring:

• Make Critical Thinking a Core Competency
• Frame AI as a Thinking Tool
• Support Diverse Entry Paths

If AI shifts how people learn and work, then how we teach people to think becomes a high-leverage policy move.

2. Enterprise’s Role: Developing AI-Ready Talent Through Practice

Ideas for companies to consider:

• Mentorship for Early-Career Talent
• Cross-Team Prompting Practice
• Normalize AI Debriefs
• Rethink Readiness Metrics

Enterprise capability in the AI era may depend less on access to tools, and more on how people think while using them.

Final Thought

AI is moving fast — but human readiness is not.

As someone at the intersection of technology, business, and society, I don’t claim to have all the answers. But I do believe that the defining factor won’t be who has AI. It will be who knows how to think with it.

Thailand has an opportunity to lead by making AI readiness a broad, inclusive effort.

Let’s build not just better tools — but better thinking.

Because a truly inclusive AI future won’t be built on code alone. It will be built on the quality of human thinking we pair with it.