Taiwan’s Trusted AI Dialogue Engine (TAIDE) Opens for Beta Applications

The TAIDE project, initiated by the National Science and Technology Council (NSTC), has reached a new development milestone. This project is a generative AI dialogue engine specifically designed for Taiwan. Initial capabilities of this generative AI dialogue engine will include Chinese-English translation, automated summarization, email drafting, and article generation based on specific prompts. For many public sector entities, this will be a major asset, significantly enhancing work efficiency and reducing repetitive, tedious tasks.

However, the TAIDE project aims for more than just assistance with basic tasks. The project team also hopes to collaborate with various public sector departments to explore more application possibilities. This includes adding more tasks and features, as well as improving the overall performance and efficiency of the engine.

The launch of this project will bring a brand-new transformation to public services in Taiwan. Public sector work will become more intelligent and convenient. Whether in document processing or citizen interaction, this generative AI dialogue engine will become a powerful and practical tool.

One of TAIDE's core advantages is its adaptation to Taiwan's linguistic characteristics; it is a generative AI model fine-tuned using a massive amount of Traditional Chinese data. It incorporates elements unique to Taiwan, such as language nuances, values, and customs, allowing it to communicate and interact more effectively with local users. The model has undergone diverse training, particularly using a variety of Traditional Chinese materials, enabling it to perform better across different subject areas. Furthermore, as training continues to expand, TAIDE's capabilities will keep improving, making it more powerful and better suited to the needs of local users.

Government ministries will be invited to sign up starting in early September to ensure broader participation and benefit from the project’s outcomes. From mid-September, the TAIDE project team will open applications for trial use, providing test accounts to interested users. Through these accounts, users can apply and experiment with the models and functions provided by the TAIDE project. In October, as the TAIDE project enters its implementation phase, the team plans to hold online meetings in mid-to-late October for exchange and discussion with central government departments. Additionally, workshops will be held during this period to introduce the TAIDE models and provide usage tutorials to agency personnel.

The TAIDE project team firmly believes that this model has immense potential for application across central and local government departments. Therefore, they hope to fully introduce the value and functions of the model to agency personnel through online meetings and workshops. This will not only help them understand how to use the model but also stimulate their interest in data analysis and AI technology.

Through these activities, the team aims to attract more attention and participation from government departments, thereby achieving the long-term goals of the project, enhancing government data analysis capabilities, and promoting more efficient and intelligent government operations.