Ahead of COP28: UAE unveils bilingual AI language model with 1.4 mn climate-related instructions

Jais Climate uses climate information and sources in both English and Arabic
Ahead of COP28: UAE unveils bilingual AI language model with 1.4 mn climate-related instructions
New model is dedicated to climate and sustainability topics

Just days before the COP28 conference in the UAE, Jais Climate, the world’s first bilingual large language model focused on climate intelligence, was revealed at the Mohamed Bin Zayed University of Artificial Intelligence in Abu Dhabi.

Developed through a collaboration between the university and Core42, a subsidiary of G42, an Abu Dhabi-based technology holding company, the large language model (LLM) is specifically tailored to address climate-related topics, sustainability, and the COP28 conference.

Jais Climate incorporates 1.4 million climate-related instructions, making it the largest instruction-based bilingual data set dedicated to climate and sustainability topics. 

Read more: Generative AI in MENA to hit $23.5 billion by 2030

The use of LLMs has stimulated the expansion and fascination surrounding artificial intelligence, as they generate content by processing vast volumes of data. This capability empowers users to obtain answers and automate tasks significantly faster than ever before.

According to Andrew Jackson, Core42’s executive vice president and chief AI officer, Jais Climate provides an opportunity to enhance understanding of climate-related matters both leading up to and during COP28.

The climate conference is scheduled to take place in Dubai from November 30 to December 12.

climate UAE

Bilingual solution

Initially, there were criticisms surrounding the language bias of LLMs, particularly their heavy reliance on English, which some critics argue may introduce biases and influence answers and outcomes.

In August, the introduction of Jais, an open-source bilingual Arabic-English model, successfully addressed these concerns by incorporating Arabic into the AI mainstream. Its developers have highlighted its superior accuracy compared to other Arabic LLMs.

Similar to Jais, Jais Climate was developed using climate information and sources in both English and Arabic. This enables easy access to climate data for over 274 million Arabic speakers and 1.4 billion English speakers globally, as stated by MBZUAI and Core42.

Both LLMs, Jais and Jais Climate, derive their names from Jais, which is a nod to the UAE’s tallest mountain peak located in Ras Al Khaimah.

Jais Climate will continuously receive updates with new information, documents, and developments throughout the duration of COP28. Its primary focus will be on science-based climate fact-checking and addressing any potential misconceptions.

For more news on sustainability, click here.

The stories on our website are intended for informational purposes only. Those with finance, investment, tax or legal content are not to be taken as financial advice or recommendation. Refer to our full disclaimer policy here.