Thursday, May 16, 2024
HE the Minister: We Envision "Fanar" as High-Accuracy Arabic LLM Capable of Processing and Understanding Natural Arabic
- Fanar is a world-class Arabic GenAI built on Arabic LLM, and able to understand the complex nuances of the Arabic language.
- Fanar will be provided with high accuracy dataset comprising at least 300 billion words, which will allow Fanar to produce highly accurate texts.
- Fanar to boost Arabic language representation in global AI arena and advance scientific cooperation in Arab world, driving innovation in the field.
- HE the Minister of Communications and Information Technology: There exists a considerable gap in Generative AI capabilities between Arabic and English languages concerning contextual comprehension, linguistic precision, and content depth and fluidity. Fanar project aims to bridge this disparity.
His Excellency Mr. Mohammed bin Ali Al Mannai, Minister of Communications and Information Technology, took part in a discussion titled "Artificial Intelligence: Regulation & Innovation" during the Qatar Economic Forum 2024. Addressing the session, he highlighted "Fanar" the Generative AI Arabic Language Model (LLM), a result of strategic collaboration between the Ministry of Communications and Information Technology (MCIT), the Qatar Computing Research Institute of Hamad Bin Khalifa University (HBKU) – Qatar Foundation Member, and other partners.
Fanar stands as an Arabic generative AI model adept at navigating the intricate linguistic landscape of Arabic. It will be fueled by vast, precise Arabic content, enabling it to evolve and enhance within linguistic frameworks, and ultimately produce culturally resonant Arabic content. This initiative is a cornerstone in the development of Arabic LLMs, enriching digital experiences for institutions and Arabic speakers alike.
His Excellency emphasized the transformative potential of Fanar, stating, "Fanar will transform GenAI-generated Arabic content in terms of accuracy and nuanced understanding, significantly enhancing translation, media, and academic research capabilities." He mentioned Qatar Computing Research Institute's remarkable achievements in creating this expansive linguistic model, marked by exceptional accuracy. Efforts are underway to translate these achievements into a pragmatic project, mindful of the Arabic language's specificity and rich cultural heritage.
His Excellency emphasized, "There exists a considerable gap in Generative AI capabilities between Arabic and English languages concerning contextual comprehension, linguistic precision, and content depth and fluidity. Fanar project aims to bridge this disparity. Fanar will also offer balanced perspectives, safeguarding Arab culture from any adverse effects."
His Excellency highlighted the fundamental role of data accuracy and training in determining Fanar's content quality. The Ministry and its partners are committed to providing high-quality Arabic information and texts, providing Fanar with around 300 billion words for robust linguistic model development.
Furthermore, His Excellency noted His Highness Sheikh Tamim bin Hamad Al Thani, the Amir of Qatar's allocation of a 9 billion QR incentive package to support comprehensive digital transformation, along with a billion-dollar fund to support regional entrepreneurs and emerging companies in line with the goals of Qatar’s National Development Strategy 3. This underscores Qatar's commitment to fortifying its regional digital leadership and driving digital transformation in the Arab world.
From his side, Dr. Ahmad M. Hasnah, HBKU President, commented: “This initiative demonstrates that HBKU employs a proactive approach towards building an inclusive research and education ecosystem in Qatar that safeguards its cultural values. Our collaboration with MCIT underscores our commitment to implementing and integrating cutting-edge AI research and technology into the social and economic development of our economy and society.”
Upon project completion, Arab users will access accurate, cutting-edge content surpassing current challenges in generative AI applications. Innovative solutions will enrich their online experience, providing enhanced access to knowledge and advanced content in an innovative, high-quality manner. Users will benefit from precise, useful content tailored to their needs, elevating their digital experience.
Ahmed K. Elmagarmid, Executive Director, Qatar Computing Research Institute (QCRI) also stated: “Our collaboration with MCIT will see the curation of Arabic-language data that encapsulates the Qatari people’s shared heritage and traditions combined with an innovative application of LLM technology. Arabic language technologies have long been one of QCRI’s research focus areas, and we are eager to utilize our expertise in service of the protection and growth of our mother tongue in the age of AI.”
Fanar's capabilities extend to assisting students, researchers, and the public by providing accurate information and streamlining tasks, saving valuable time and effort. In addition to generating high-quality Arabic text, Fanar facilitates the development of Arabic chatbots and virtual assistants for companies and institutions of all sizes, ensuring culturally appropriate responses.
Furthermore, Fanar offers a comprehensive set of services, including translation, summarization, and creative writing, empowering companies and institutions to effectively engage their Arabic-speaking audience. By enhancing the Arabic user experience through accurate and culturally appropriate responses, Fanar drives broader technology integration into the daily lives and business activities of Arabic speakers.