The TH-Ai Passport project shows distorted Thai language usage and actual users fewer than registered. After one day of launch, there were 100,000 users, while registrations reached 1.5 million.
On 31 Aug 2023, the Ministry of Digital Economy officially launched the TH-Ai Passport (access to Pro-level AI). Registration had opened on 19 Aug 2023. The project consolidated over 30 AI models on one platform to enhance AI understanding and encourage the public to use AI practically, aiming to support up to 5 million Thai citizens aged 15 and above.
On 1 Sep 2023, Rakchanok Srinok, a party-list MP from the People's Party, posted that the original TH-Ai Passport contract was a lump sum payment of 1.6 billion baht regardless of user count, whether 10 or a million users. The People's Party questioned this, prompting the Ministry of Digital Economy to revise the contract to pay-per-active-user—about 25 baht per account. Currently, about 1.5 million people have registered, but only about 100,000 use it daily, meaning the government pays approximately 2.5 million baht per month, down from the original 1.6 billion baht.
Thairath Online's special team asked Dr. Tamthee Sukchotirat, an expert in Big Data and government technology, who said that with over a million registered but only 100,000 users on the first day, that is about 10%. To fairly assess the project, it should be compared to active user rates on similar private or government platforms in Thailand. The key will be to monitor usage after one week to see if user numbers increase or decrease. Initially, daily, weekly, monthly, or quarterly monitoring is needed, and relevant agencies must watch this closely as it is quite important.
Dr. Tamthee noted that active user rates depend on definitions and platform types. For example, news platforms measure daily users since people consume news daily, while travel platforms may count users who purchase flights or hotels once or twice a year as active users.
Dr. Tamthee pointed out that agencies managing the project should categorize users by frequency—daily, every three days, or weekly—to observe how often users return. They must also assess whether usage problems or unmet expectations exist, since the platform is labeled Pro, promising Thai users savings by avoiding multiple subscriptions. Users who tried the Pro version expect Pro-level experience—no system crashes, errors, delays, and accurate image generation.
Dr. Tamthee posted on Facebook: “The government’s highest-valued generative AI project, but the Thai language image generation is distorted?”
With mixed feedback, the official page admin has worked on improvements, such as focusing on image generation models that perform well. Initial measurements over the first 22 hours showed about 3.2% user praise for model diversity. Technically, the government is evaluating whether the variety of models is truly necessary and how to fix any models that produce distorted outputs.
User feedback falls into four main groups:
1. Issues with images (most frequent).
2. Video needs (resolution and length).
3. Data analysis.
4. Question-and-answer accuracy.
Knowing these directions allows project managers to adjust accordingly.
“On the first launch day, comments seemed disabled, which I believe is not a good approach for a public project today, as it prevents the project owner from knowing public opinions. If comments were open from the press conference, they might have received abundant feedback to improve the project.”
Dr. Tamthee said value means what is gained compared to money spent. Many agencies question the revised pay-per-active-user contract, noting that 'per user' payment may lack detail—such as whether payment continues after registration regardless of usage, or if low usage months lead to no payment. Ideally, payment should depend on actual use and user frequency to make contracts fair and protect taxpayers’ money.
At the press conference on 28 Aug 2023, Dr. Tamthee presented survey results showing the public wants civil servants to use AI to improve government work efficiency, speed, and transparency. The Ministry of Digital Economy plans to coordinate with other ministries, which is seen as a positive direction. Widespread use by civil servants would allow citizens to gauge government efficiency improvements.
Dr. Tamthee also emphasized the importance of data processing within the country. The terms of use and privacy policy, section 1.3 of chapter 3, clearly state that prompts—commands, texts, images sent to AI—are personal data. This conflicts with claims that personal data will not be processed overseas. Since government work is classified at a national level, if this project does not store and process data domestically as intended, and civil servants send data abroad for processing, this raises security concerns.
The privacy policy section 4.1 states the project prohibits AI model providers from using user data to train their models under any circumstances. Providers may refuse or cancel service based on this clause. If used by government or private sectors, the data is also company confidential. The government should develop and process AI models domestically, which can be done in various ways—for example, allocating budget to enhance AI models already developed by the government or engaging in government-to-government agreements to host the latest AI versions locally and develop further. This ensures budget funds are used for development, not just renting, benefiting the nation most without contradicting any terms of reference.
A positive aspect of the project is training and skill development, where users earn points beneficial for their organizations, depending on use cases. However, feedback on performance superiority compared to older platforms has not yet been seen.