🔗 Share this article

Around the globe, states are investing massive amounts into the concept of “sovereign AI” – creating domestic AI models. Starting with the city-state of Singapore to Malaysia and the Swiss Confederation, states are vying to build AI that comprehends native tongues and cultural nuances.

The International AI Arms Race

This initiative is part of a wider global contest spearheaded by major corporations from the America and China. Whereas firms like a leading AI firm and a social media giant allocate massive capital, middle powers are also placing their own bets in the AI landscape.

Yet with such tremendous amounts in play, is it possible for less wealthy nations attain meaningful gains? According to a specialist from an influential research institute, “Unless you’re a wealthy state or a major firm, it’s a significant challenge to create an LLM from scratch.”

Security Issues

Numerous countries are hesitant to depend on overseas AI models. In India, for instance, Western-developed AI systems have sometimes fallen short. An illustrative case featured an AI assistant deployed to teach students in a remote village – it spoke in English with a strong US accent that was hard to understand for local students.

Additionally there’s the defence aspect. For the Indian defence ministry, using certain foreign systems is considered inadmissible. As one entrepreneur explained, There might be some arbitrary training dataset that might say that, such as, a certain region is not part of India … Utilizing that certain AI in a security environment is a big no-no.”

He continued, “I have spoken to people who are in defence. They aim to use AI, but, forget about specific systems, they are reluctant to rely on US systems because information might go abroad, and that is totally inappropriate with them.”

National Efforts

Consequently, a number of countries are funding local ventures. An example such effort is underway in India, in which a firm is working to build a national LLM with government support. This initiative has dedicated about $1.25bn to artificial intelligence advancement.

The expert foresees a system that is significantly smaller than premier systems from US and Chinese firms. He explains that the country will have to compensate for the financial disparity with talent. Based in India, we don’t have the advantage of investing billions of dollars into it,” he says. “How do we compete with such as the $100 or $300 or $500bn that the US is investing? I think that is the point at which the core expertise and the strategic thinking plays a role.”

Native Focus

In Singapore, a state-backed program is funding machine learning tools educated in local regional languages. These languages – including Malay, the Thai language, the Lao language, Bahasa Indonesia, Khmer and additional ones – are commonly underrepresented in Western-developed LLMs.

I wish the experts who are creating these national AI systems were aware of the extent to which and the speed at which the cutting edge is moving.

An executive engaged in the program notes that these tools are intended to complement bigger systems, as opposed to replacing them. Systems such as a popular AI tool and another major AI system, he states, frequently find it challenging to handle native tongues and cultural aspects – communicating in awkward the Khmer language, for example, or recommending non-vegetarian meals to Malay users.

Developing native-tongue LLMs enables national authorities to include cultural nuance – and at least be “knowledgeable adopters” of a powerful system built overseas.

He adds, I am cautious with the concept sovereign. I think what we’re attempting to express is we want to be more adequately included and we want to understand the capabilities” of AI platforms.

Cross-Border Partnership

Regarding states trying to carve out a role in an growing worldwide landscape, there’s an alternative: team up. Analysts affiliated with a prominent university have suggested a government-backed AI initiative allocated across a group of middle-income countries.

They call the initiative “a collaborative AI effort”, in reference to the European successful initiative to build a rival to a major aerospace firm in the 1960s. This idea would involve the formation of a government-supported AI organization that would pool the assets of several states’ AI projects – including the UK, Spain, the Canadian government, Germany, Japan, Singapore, South Korea, the French Republic, the Swiss Confederation and Sweden – to establish a competitive rival to the American and Asian giants.

The primary researcher of a paper setting out the proposal states that the concept has attracted the attention of AI leaders of at least three nations so far, along with a number of national AI companies. Although it is currently centered on “developing countries”, less wealthy nations – the nation of Mongolia and Rwanda included – have also shown curiosity.

He elaborates, “Nowadays, I think it’s simply reality there’s diminished faith in the assurances of the present White House. People are asking like, should we trust such systems? In case they choose to