Singapore proposes a UN framework convention on AI safety
PHILADELPHIA -
asserted
Singapore → propose → safety
As the US and China accelerate their artificial intelligence push, it is too late to ask for a safety pause, but what would help is the creation of institutions that can set safeguards and standards, Singapore Foreign Minister Vivian Balakrishnan told the United Nations in a speech on Sept 26.
asserted
Balakrishnan → accelerate → Sept
“We must collectively formulate new rules and be prepared to create new institutions, if necessary, for new technologies,” Balakrishnan said while delivering Singapore’s national statement at the General Debate of the UN General Assembly (UNGA) in New York.
asserted
Balakrishnan → formulate → York
“We must explore all options, including the idea of a UN Framework Convention on AI Safeguards,” he said in a 20-minute speech that advocated revitalising multilateralism to tackle acute geopolitical challenges.
asserted
that → explore → challenges
A framework convention, like the one adopted by the UN on climate change, could be in the form of a broad international treaty that establishes foundational principles and cooperative institutions and provides the architecture for international action.
uncertain
that → adopt → action
A similar call was made by Turkish President Recep Erdogan who used his address at the UN to stress the need for a common international legal framework for AI.
asserted
who → make → AI
Balakrishnan said new institutions would also be necessary amid loud calls for AI safety from top AI tech leaders, safety researchers and lawmakers who warn that AI capabilities are advancing faster than the human ability to safely manage them.
asserted
capabilities → say → them
“We may also need an international institution that can perform for AI some of the functions that currently exist for institutions that perform this function of setting common technical standards,” the minister said, citing the examples of the International Telecommunications Union (ITU), which manages global telecommunications and digital networks and the International Atomic Energy Agency (IAEA) which oversees the peaceful and safe use of nuclear technology.
uncertain
which → need → technology
Balakrishnan said the risks arose from three possibilities – the loss of control of autonomous systems; the misuse by rogue actors to create bio weapons or other weapons of mass destruction, and pervasive political and socio-economic disruption.
asserted
risks → say → destruction
“But the genie is already out of the bottle,” he added, in a reference to the clamour for safety as well as the nascent efforts between the US and China to set up a dialogue on the issue.
asserted
he → add → issue
“It is too late to call for a halt, especially given the superpower rivalry and the huge financial incentives for companies who already have frontier models,” he said.
asserted
he → call → models
He used the analogy of a car and traffic lights to bring the point home.
asserted
He → use → point
“To drive a high-performance car quickly and safely, you need a powerful engine, but equally, you need good brakes,” he said.
asserted
he → drive → brakes
“And especially if you are going to go beyond your neighborhood or cross boundaries, you need effective shared rules, traffic lights, insurance, norms of behaviour, and seat belts,” he added.
asserted
he → go → behaviour
“Our safeguards must keep pace,” he said, calling for rigorous testing and evaluation before deployment.
asserted
he → keep → deployment
“We need clear limits on what autonomous systems can do, and mechanisms to intervene when they behave in ways we do not care,” he said.
asserted
he → need → ways
He also said these mechanisms had to be universally recognisable.
asserted
mechanisms → say → ?
“Since the risks from AI do not remain confined behind borders, it means we also need rules of the road that will form a shared global understanding of how we will collectively confront this challenge, just as all of us instinctively understand traffic lights the same way, whether we’re here in New York, or in Beijing, Moscow, or Singapore,” he added.
asserted
he → remain → Beijing
Above all, human beings must ultimately remain in charge.
asserted
beings → remain → charge
“Do you want the nuclear button to be in the hands of a human being or an AI?” he asked.
asserted
he → want → being
Any durable international approach must also give all states a meaningful stake, he added, not only those with the “most advanced models, the deepest pockets or the largest set of armaments.”
asserted
he → give → armaments
Singapore is contributing to ongoing efforts too, he added.
asserted
he → contribute → efforts
“This week, we joined the call for greater safety and security of frontier AI models,” he said.
asserted
he → join → models
Over 20 countries and the European Union signed a declaration led by Finland and Norway on the sidelines of the UNGA in New York.
asserted
countries → sign → York
It calls for binding international government oversight over frontier AI models, rather than corporate self-regulation
Global cooperation over AI must begin now, Balakrishnan said.
asserted
Balakrishnan → call → AI
“Rules do not stop us or slow us down.
asserted
Rules → stop → us
They give us the assurance and the confidence so that we can move faster and more effectively for the benefit of all our peoples.
asserted
we → give → peoples
”
Why China is dogged in pursuit of AI
asserted
China → dog → AI
During a Sept 25 conversation with former Australian prime minister Kevin Rudd, who is now global president and chief executive of Asia Society in New York, Minister Balakrishnan zeroed in on the US-China battle over AI.
asserted
Balakrishnan → zero → AI
China sees the AI race through a deeply historical lens, he said.
asserted
he → see → lens
Having missed the 20th century industrial revolution, it is especially keen not to miss the boat a second time.
asserted
it → miss → boat
“Because China missed it, China underwent a century plus of humiliation and unequal treaties.
asserted
China → miss → humiliation
This time, China has absolutely no intention of missing it again,” Balakrishnan said in a wide-ranging, hour-long discussion which challenged the ways in which the US seizes the competition from China.
asserted
US → have → China
“You need to understand it is not just a matter of competition with America, being number one or number two, and I am projecting into China’s motivations, but China says: ‘this is my turn, my time in the sunshine.
asserted
this → need → sunshine
Why should anyone get in the way, especially when I have the human talent, the brain power?’,” he said.
asserted
he → get → talent
“So can you understand from China’s perspective, ‘I am not going to make the strategic mistake made a century and a half ago.
asserted
I → understand → mistake
This is a strategic opportunity and I do not want anyone to get in my way.
asserted
anyone → want → way
And all I am asking for is a level playing field, and access to technology, and to get ahead as quickly as I can’,” he said.
asserted
he → ask → technology
Rudd pointed out that China was heavily using state subsidies to win the AI race, a part of the frequent western criticism of the Chinese economy where domestic consumption remains weak and growth comes from exports.
asserted
growth → point → exports
Balakrishnan agreed on the pervasive use of subsidies but said they did not fully explain the country’s successes with AI.
asserted
they → agree → AI
…and 17 more, not listed.
The Aporia
