Every prompt typed into foreign artificial intelligence platforms carries an invisible bias: exposure to the cultural values, language quirks, and political perspectives of the countries where those models were trained. For the Philippines, exclusive reliance on external systems like ChatGPT, Claude, or foreign-built platforms risks gradually altering local language norms and tilting public understanding on critical national issues.
To mitigate these risks, the Department of Science and Technology (DOST) is building local large language models (LLMs) trained on local languages and sovereign data, including a public service chatbot called “Itanong.”
A Sovereign Counterweight to Foreign Bias
During a Senate budget hearing for DOST’s proposed 2027 spending plan, Senator Bam Aquino warned that heavy reliance on foreign AI systems exposes Filipinos to subtle cultural assimilation.
“As you keep on using the AI, you’re training the AI, but the AI is also training you,” Aquino said, noting that generated outputs inevitably reflect the worldview of the model’s creators.
To illustrate the geopolitical risks, Aquino pointed out that a Chinese-developed AI would yield a fundamentally different perspective on the West Philippine Sea dispute compared to a locally built system. Without sovereign alternatives, Filipino users risk absorbing external perspectives on national issues.
Meet ‘Itanong’: The Local AI Built for Taglish
Rather than competing directly with trillion-parameter global AI models, DOST Secretary Renato Solidum Jr. explained that the country’s national initiative focuses on domain specificity, localized challenges, and sovereign data protection.
Developed by the DOST’s Advanced Science and Technology Institute (ASTI), the agency showcased Itanong, a specialized AI chatbot designed to operate in a private data space.
Key details of the local LLM roadmap include:
- Language Support: Built to process queries fluently in Filipino, English, and Taglish, with plans to expand into less widely spoken Philippine regional languages.
- Public Service Integration: Designed for eventual deployment across e-government platforms to help citizens understand administrative requirements, permits, and public services in natural local phrasing.
- Data Sovereignty: Powered by localized information, including DOST’s Analytics and Technology Solutions Platform, which digitizes and geographically tags data for secure spatial analytics.
How the National AI Framework Is Divided
The push for localized AI coincides with the Senate committee referring DOST’s proposed P30.87 billion budget for 2027 to the plenary for approval.
To execute the country’s broader National AI Strategy Framework, Solidum outlined a multi-agency division of labor across government:
- DOST: Leads core research and development, platform operations, and local AI talent development.
- Department of Information and Communications Technology (DICT): Oversees underlying compute infrastructure, cloud hosting, and nationwide connectivity.
- Department of Economy, Planning and Development (DepDev): Drives policy formulation, AI ethics, and regulatory governance.
- TESDA, DepEd, and CHED: Manage workforce upskilling, educational integration, and technical training.
This article is published on BitPinas: DOST Builds Homegrown AI ‘Itanong’ as Senate Flags Cultural Risks in Foreign Models
What else is happening in Crypto Philippines and beyond:
