When it comes to artificial intelligence, sometimes less is more. This week, the Mohamed bin Zayed University of Artificial Intelligence, in partnership with Abu Dhabi-based technology group G42, unveiled a new AI model, one that actually boasts about being smaller than its peers.
K2 Think, which is designed to tackle complex reasoning tasks, features 32 billion parameters – internal variables that help the model perform better as it learns. This makes it much smaller than some other AI models, such as OpenAI and DeepSeek, which often exceed 200 billion parameters. However, Prof Eric Xing, president of MBZUAI, explains that K2’s smaller size means users will get “a much faster compute and also a much less expensive cost of generating the results”. In short, it may work harder than its rivals – but it will also work smarter.
In the race to stay ahead of the technological curve, such innovation not only gives developers an important edge, it reveals how nurturing home-grown talent and developing one’s own expertise pays off. Last month, the UAE was listed among the world’s top three AI superpowers in a report by US data centre company TRG – an achievement that is inherently linked to the Emirates’ demonstrable investment in the people and technology fuelling the country’s AI revolution.
A closer look at K2 reveals where that revolution may be headed. As a kind of reasoning engine, it is quantifiably different to what has come before. Although large language models have been popular for chat-style applications, the ability to reason through complex problems is the next frontier for AI, with potential applications in research, finance, logistics and engineering. Prof Xing, speaking to The National this week, said K2 could eventually serve as the backbone of business tools across these industries, and MBZUAI is already examining how other fields – such as health care and genomics – could benefit from the K2 approach.
By not only building its own AI platforms but pushing the envelope of what the technology can do, the UAE is showing that it has much more to contribute to this field beyond building some of the world’s largest data centres. Intriguingly, K2 is open source; this means its data, training recipe and deployment code are available, giving researchers the chance to study how reasoning emerges and adapt the model. Even robust criticism will help it improve and learn. In doing so, the UAE is informing the global conversation on AI, not hiding its knowledge in silos.
K2 and its place in the UAE’s dynamic technology scene are also a glimpse into how the country’s economy is developing. Last week, preliminary estimates released by the Federal Competitiveness and Statistics Centre found that non-oil economic activities contributed 77.3 per cent of total real gross domestic product. For context, non-oil contributions to GDP stood at 71.3 per cent in 2020 and have gradually increased over the past five years.
Creating and embracing leaner, smarter AI looks set to add impetus to an economy that is diversifying and a society that is quickly putting this technology to good use, whether it be in the classroom, on the roads or in hospitals. Silicon Valley may still command much of the world’s attention when it comes to tech breakthroughs but the UAE is showing that it is possible to break new ground anywhere.