Thailand’s AI drive constrained by talent and data access gaps

Thailand’s adoption of artificial intelligence (AI) remains at an early stage, held back by weak domestic technology development, limited access to data and a persistent shortage of skilled workers, according to the National Electronics and Computer Technology Centre (NECTEC).

Dr Chai Wutiwiwatchai, director of NECTEC, outlined the challenges during a special address titled “Thailand’s AI Vision: Thailand’s AI Strategy for Sustainable National Development” at the AI for Public Sector: AI-Driven Government Services seminar on Thursday (August 6).

AI is a crucial tool for improving Thailand’s economy and people’s wellbeing, he explained. Since its first phase began in 2022, the national strategy has focused on fostering innovation that benefits society, with wider economic gains expected to follow.

Rank rises but momentum slows

Citing an Oxford Insights index covering about 200 countries, Chai noted that Thailand ranked around 60th in 2020 and climbed to around 30th after the Cabinet endorsed the national AI plan. However, the country ranked 34th in 2025 and has made limited further progress.

“The key observation is that we are weak at developing our own technology. We are users, but rarely developers,” he noted, describing this as a major factor preventing Thailand from moving higher.

Although the country has made significant improvements in data and infrastructure, access to data remains restricted.

“We have good data and a lot of it, but access to it and the ability to use it productively are still limited,” he added.

Between 2020 and 2022, education received the largest share of AI research and development investment, followed by healthcare and agriculture.

A survey of nearly 600 organisations, about 20% of them public-sector bodies and 80% private companies, found that fewer than 20% had adopted AI at organisational level in 2024.

Chai believes the integration of AI into organisational workflows remains at an early stage in 2026, even though individual use is likely to be much higher.

Sectors moving at different speeds

Introduced in 2022, Thailand’s national AI plan has sought to build the country’s AI ecosystem through legislation, research and development, workforce training, infrastructure investment and incentives encouraging organisations to improve efficiency through AI.

The strategy covers 10 sectors, but their levels of readiness vary. Finance and logistics are moving fastest and require less direct state support because they have sufficient momentum to develop independently.

Industry, healthcare and education remain at an early stage but could become important new growth engines.

Agriculture and tourism, however, require a stronger push. The agriculture sector lacks sufficient participation from a younger generation willing to adopt digital technology and AI, while tourism data collection remains inadequate to support clear and practical innovation.

Building AI skills from school to work

Workforce development is a central part of the strategy, spanning basic education, higher education and post-study training. It also covers different skill levels, from professionals using AI in their work to developers and specialist researchers, alongside wider AI literacy and the safe use of the technology.

AI content has already been introduced at upper-primary, lower-secondary and upper-secondary levels. The courses remain optional, with the next goal being to make them compulsory.

The Ministry of Higher Education, Science, Research and Innovation is also promoting the use of AI tools in teaching and learning, alongside improved teaching methods and training for lecturers.

Large-scale boot camps are available for people seeking to move into AI-related careers. They accept around 10,000 participants each year, with more than 2,000 completing the programmes annually.

Local language models for government

Thailand can use foreign generative AI services, but their adoption by government agencies raises concerns because much of the information handled by the public sector is sensitive and should not be exposed to overseas cloud systems.

The national plan therefore supports local cloud infrastructure and Thai-developed open-source large language models to strengthen data security and keep sensitive public-sector information within domestic systems.

Thailand released its first major open-source Thai-language LLM in 2022, and the technology has since been continually improved and fine-tuned for different sectors. AI industry groups and private-sector developers have adapted the open-source model, while NECTEC developed Pathumma LLM for government agencies.

Pathumma now supports an agentic chatbot for the Office of the Public Sector Development Commission. The system is designed to provide a one-stop enquiry service capable of retrieving information from chatbots operated by different government agencies.

Medical AI needs wider data and faster approvals

Thailand has invested substantially in medical AI research and development, but many projects have yet to reach commercial deployment.

One challenge is the need for comprehensive datasets that can produce models capable of working effectively across different hospitals. Wider data sharing is therefore needed to support clinical testing that meets Thai Food and Drug Administration requirements.

To address these barriers, a medical AI consortium involving major hospitals has been pooling data, training specialists and accelerating model development. More than 10 models have been developed through the consortium, and at least two have secured Thai FDA approval over the past two years.

Standards bridge ethics and law

Thailand’s approach to responsible AI has three levels: existing ethics guidelines, technical standards and a future AI law being drafted by the Ministry of Digital Economy and Society in collaboration with the Electronic Transactions Development Agency.

Chai described standards as a practical middle ground between voluntary guidance and full legislation.

NECTEC has established an ISO 17025-accredited laboratory to test AI software in three areas: AI integrated with Internet of Things devices, known as AIoT; healthcare AI; and biometrics.

For AI products not yet covered by formal standards, benchmark.ai.in.th uses datasets and competitive testing to assess performance and identify vulnerabilities, following an approach similar to that used by the National Institute of Standards and Technology in the United States.

Testing for trustworthy large language models has been brought forward as a priority in 2026.

Chai concluded that the first four years of the national plan had produced three clear lessons: Thailand must continue developing its workforce, concentrate resources on high-impact projects rather than applying AI indiscriminately, and prepare the infrastructure needed to support those priorities.