In Abu Dhabi, proponents of open-source AI development make their case

A recent gathering of some of the most influential names in artificial intelligence has solidified the UAE's place as a centre for open-source AI development, according to an AI researcher in the country.

“The message this summit sends is that AI in general is something for everyone, for humanity. It shouldn’t be controlled by a handful of people,” said Hakim Hacid, chief researcher of Technology Innovation Institute's artificial intelligence and digital science research centre, which organised the inaugural open-source AI summit in Abu Dhabi.

Mr Hacid said that TII was among the first to make a big push for open-source AI development with the release of its Falcon 40B large language model last year.

“We have been extremely open, with our models you can take them, you can build your whole business or ecosystem around that as you wish,” he told The National, which Mr Hacid said allowed for TII to engage with global AI experts to ensure the future of AI development is collaborative.

“It makes Abu Dhabi and the UAE an explicit hub for AI, but in particular, open-source AI development,” he said.

AI experts from technology heavyweights such as Google DeepMind, Meta, Cerebras and Amazon Web Services delivered presentations and speeches during the two-day summit at the St Regis Saadiyat Island resort in Abu Dhabi.

Proponents of open-source AI models often say the open-source mentality democratises artificial intelligence, whereas closed-source models only allow for those with access to larger computing infrastructures to develop the technology.

“People are struggling to get compute, without compute you don’t have AI at the end of the day,” Mr Hacid said.

“We believe that there are brilliant people out there who have brilliant ideas, and to implement them they need to compute. That’s why the foundation of open-source AI development is so important.”

Those who push for closed models, however, often claim there are potential security vulnerabilities posed by open-source AI development, a lack of quality assurance and in turn, a potential lack of technical stability.

In many ways, the debate is yet another version of an age-old argument that's surrounded technology companies since the 1980s.

Back in July, Meta chief executive and Facebook founder Mark Zuckerberg sent ripples throughout the tech world when he wrote a lengthy post where he argued that open-source AI was the path forward.

“AI has more potential than any other modern technology to increase human productivity, creativity, and quality of life – and to accelerate economic growth while unlocking progress in medical and scientific research,” he said.

“Open source will ensure that more people around the world have access to the benefits and opportunities of AI, that power isn’t concentrated in the hands of a small number of companies, and that the technology can be deployed more evenly and safely across society.”

At the Open Source Summit, Mr Hatim echoed those sentiments, reflecting on the upside of open source research and development in the context of artificial intelligence and large language models.

“It means that you have access to the whole system you are using, you know what’s happening on the inside, you know where you data is going, what is done with your data," he said, comparing it to non-open source AI implementations.

"You don’t know what’s happening in the background, you don't know where you data is going, and if you want to adapt that AI to try things related to your specific problems, you cannot," he added.

Yet some of the world's most popular implementations of artificial intelligence such as OpenAI's ChatGPT, are not open-source, although with more AI players on the scene, OpenAI's dominance is far from secured.

Speaking at the Open Source AI Summit, Natalia Vassilieva, vice president and chief technology officer of Cerebras Systems, a Silicon Valley-based AI hardware design firm said that open-source AI development was critical for achieving a high rate of progress with large-language models, the backbone technology that makes AI possible.

“We are in this together,” she said, explaining the upside to having more people tinkering with AI development.

“We often need to train customers that open-sourcing models is ultimately good to them, it takes some explaining.”

The Open Source AI Summit is the latest prominent technology gathering held in the UAE amid crescendoing AI advancements.

Several years before AI was the technology topic du jour fuelling unprecedented investment and research, the UAE had sought to be a leader in artificial intelligence research.

In 2019, the country announced the Mohamed bin Zayed University of Artificial Intelligence, one of the world’s first dedicated AI educational institutes.

Sam Altman, founder and chief executive of OpenAI, whose ChatGPT help AI go mainstream, told the 2024 World Government Summit in Dubai that the UAE would be well-positioned to lead discussions about a hypothetical global AI watchdog system.

The country also recently became the first in the Arab world to join the Hiroshima AI Process Friends Group, which is dedicated to AI safety and security.

According to the Stanford Institute for Human-Centred AI, among 36 countries around the world evaluated for AI competitiveness, the UAE ranked fifth, ahead of countries including France, South Korea, Germany, Japan and Singapore.