Technology
Concerns Raised Over AI-Generated Text Errors in Thailand's Passport System
Opposition MP Pavut Pongvitayapanu has criticized the government's 'TH-AI Passport' project, citing the use of low-cost AI models as the cause for garbled, nonsensical text in generated images.
On September 2, 2026, at the Thai Parliament, list-MP Pavut Pongvitayapanu of the People's Party addressed technical issues regarding the government's 'TH-AI Passport' initiative. According to Pavut, the system has been producing images containing unintelligible, 'alien-like' text. He attributed these errors to the government's decision to prioritize low-cost AI models over more sophisticated, language-proficient alternatives like Google’s Nano or ChatGPT.
Pavut argued that the government’s attempt to minimize operational costs by selecting cheaper, less capable image-generation engines has compromised the quality of the output. He suggested that if the project had utilized more advanced models, the linguistic accuracy would be significantly higher. The MP urged the government to involve external experts in the inspection and acceptance process of such technology and called for greater transparency regarding the project's data and implementation.
For travelers and residents, this development highlights potential reliability issues with government-integrated AI services. While the project budget was reportedly reduced from 1.6 billion baht to reflect actual usage, the technical shortcomings raise questions about the efficacy of the current system. It remains to be confirmed how the government will respond to these criticisms and whether they will adopt more advanced AI models to rectify the text generation errors in future updates.