TII Unveils Falcon-Emirati AI Models to Enhance Arabic Language Technology

Abu Dhabi:The Technology Innovation Institute (TII), the applied research arm of Abu Dhabi's Advanced Technology Research Council (ATRC), has launched three new artificial intelligence models aimed at integrating Emirati language and culture into AI technologies. These models are designed to provide developers, researchers, and everyday users with tools that incorporate Emirati Arabic and multimodal functionalities.

According to Emirates News Agency, the new Falcon-Emirati model is a 7-billion-parameter language model specifically built to capture the vocabulary, expressions, idioms, and cultural contexts of Emirati Arabic. It was introduced alongside Falcon-ASR, a 1.6-billion-parameter model for converting spoken language into text in various languages, and Falcon-OCR-Arabic, which is designed for extracting Arabic text from images and documents.

Faisal Al Bannai, Advisor to the UAE President for Strategic Research and Advanced Technology Affairs, emphasized the importance of embedding local language into future technologies, highlighting the pride and freedom in developing such capabilities within the UAE. This initiative aims to create a more seamless interface between people and AI across different content forms, thereby strengthening the presence of Arabic in AI applications.

Falcon-Emirati enables interaction using the local dialect, while Falcon-ASR facilitates multilingual transcription services, supporting applications such as customer service and accessibility tools. Dr. Najwa Aaraj, CEO of TII, stated the significance of sovereign AI capabilities reflecting daily language usage, ensuring AI systems understand the unique aspects of Emirati communication.

The Falcon-Emirati model was developed using native content and scored highly on an Emirati Arabic benchmark, outperforming other models. Falcon-ASR, despite its compact size, delivered superior accuracy in speech conversion, exceeding the performance of larger models. Falcon-OCR-Arabic expands into optical character recognition, aiding in document digitisation and information extraction.

Dr. Hakim Hacid, Chief Researcher of TII's Artificial Intelligence and Digital Research Centre, noted that these advanced capabilities empower regional developers and researchers to innovate further, enhancing Arabic AI's progression across language, speech, and document understanding. This launch is a significant step in advancing TII's sovereign AI capabilities.