“Robots are not, yet, autonomous killing machines. Even the rule briefly considered in San Francisco, would have kept a human operator in the decision loop.”
Branka Marijan on the risks of weaponizing robots in Newsweek
Among the challenges that AI poses to global governance, its application in warfare and defence is the most complex, as Bessma Momani, Aaron Shull and Jean-François Bélanger wrote in the introduction to the Ethics of Automated Warfare and Artificial Intelligence essay series. The war in Ukraine has become AI-enabled weapons’ latest proving ground, including with advanced autonomous aerial and marine drones.
At the Responsible AI in the Military Domain (REAIM 2023) summit in The Hague, Branka Marijan and others discussed “killer robots” that still have human operators, but perhaps won’t for long if governance can’t keep up. There’s a real temptation to unleash these machines, including in law enforcement capacities, which Marijan has said would be a huge mistake. As David Evan Harris argued in The Guardian, the race to the bottom on AI safety must stop to prevent real harm to people.