SEOUL (Reuters) — A worldwide summit in South Korea on Tuesday introduced a “blueprint for motion” to control accountable use of synthetic intelligence in the military, with extra sensible pointers than an analogous doc final yr, however nonetheless legally nonbinding.
It was not instantly clear what number of of the 96 nations that despatched authorities representatives to the summit, together with the United States and China, are endorsing the doc.
The Accountable AI in the Military Area (REAIM) summit in Seoul, the second of its form, follows one held in Amsterdam final yr. At the moment, round 60 nations endorsed a modest “name to motion” with out authorized dedication.
Authorities representatives talking at a roundtable on Tuesday mentioned this yr’s “blueprint” was extra action-oriented.
That is in maintaining with advances in discussions about AI threat, and developments in military use of AI equivalent to Ukraine’s rollout of AI-enabled drones.
“We’re making additional concrete steps,” Netherlands Protection Minister Ruben Brekelmans informed Reuters. “Final yr … was extra about creating shared understanding. Now we’re getting extra in direction of motion.”
Audio system together with Radha Plumb, the chief digital and synthetic intelligence officer in the U.S. Division of Protection (third from left), talk about accountable AI use in a plenary session at the REAIM summit in Seoul on Sept. 9.
© Reuters
This consists of laying out what sort of threat assessments needs to be made, vital circumstances equivalent to human management, and the way confidence-building measures will be taken in order to handle these dangers, he mentioned.
Amongst the particulars added in the doc was the want to stop AI from getting used to proliferate weapons of mass destruction (WMD) by actors together with terrorist teams, and the significance of sustaining human management and involvement in nuclear weapons employment.
South Korean officers mentioned the doc covers a lot of the similar floor as rules laid out elsewhere, equivalent to the U.S. authorities’s declaration on accountable use of AI in the military launched final yr.
However the Seoul summit — co-hosted by the Netherlands, Singapore, Kenya and the United Kingdom — goals to make sure ongoing multistakeholder discussions not dominated by a single nation or entity.
The venue and timing for the subsequent summit remains to be being mentioned, officers mentioned.