China has rejected a US proposal to limit the role of artificial intelligence (AI) in the deployment of nuclear weapons read more
Sixty countries gathered in South Korean capital Seoul to declare that humans and not artificial intelligence (AI) should control nuclear weapons.
The countries including leading nuclear powers like the United States and United Kingdom. The only major exception was China.
At the Responsible AI in the Military Domain (REAIM) Summit, around 60 of the 100 countries in attendance adopted the ‘Blueprint for Action’, which declared that it was essential to “maintain human control and involvement for all actions…concerning nuclear weapons employment”, according to AFP.
The declaration — a non-binding agreement — comes as China has repeatedly refused to rule out the AI control of nuclear weapons. The Seoul declaration, which China did not join, said “AI applications should be ethical and human-centric”, as per the news agency.
What’s ‘Blueprint for Action’ for military use of AI?
This was the second edition of the summit. In the first edition, around 100 countries agreed on a much more modest call to action. It was co-hosted by the United Kingdom, the Netherlands, Singapore, and Kenya.
While the ‘Blueprint for Action’ published this year was broader, the signatories fell from 100 to 60. The declaration billed itself as the “most comprehensive and inclusive platform for AI in the military domain”, as per AFP.
The declaration acknowledged there was a long way to go for nations to keep pace with the development of AI in the military domain. It noted that they “need to engage in further discussions…for clear policies and procedures”.
Dutch Defence Minister Ruben Brekelmans told Reuters that while the last year’s summit was about creating a shared understanding, this year’s summit was about “getting more towards action”.
This includes laying out what kind of risk assessments should be made, important conditions such as human control, and how confidence-building measures can be taken in order to manage risks, said Brekelmans.
Even as the declaration sought to stop the AI control of nuclear weapons, it did not outline any sanctions or other punishments in case of a violation, as per AFP.
China has so far refused to rule out AI control of nukes
China has consistently used to rule out the AI control of its nuclear weapons.
So far, the militaries around the world are using for reconnaissance, surveillance, and analytical purposes, but there remains a broad consensus that the decision to launch nuclear weapons will always rest with humans.
The AFP reported that there are plans that AI could be used in the future to pick targets autonomously but most nations still agree that the decision to launch should rest with humans — but not China.
In June, the White House said that China had rejected a US proposal to limit the role of AI in controlling the launch of nuclear weapons.
In addition to revealing China’s rejection of the US proposal, Tarun Chhabra, the Director of Technology at the White House National Security Council, said that it was a long-stated US policy that AI should be completely out of nuclear bombs-related decisions.
“Our position has been publicly clear for a very long time: We don’t think that autonomous systems should be getting near any decision to launch a nuclear weapon. That’s long-stated US policy,” said Chhabra.
While it is not known how China may use AI in the launch of nuclear weapons, Soviet Union used to have a ‘Dead Hand’ autonomous system for launching nuclear weapons with the idea that an autonomous system would launch weapons in case leaders and officials are incapacitated or killed. There have been reports that Russia continues to have the system even today.
