An international treaty restricting the use of artificial intelligence (AI) in military decision-making and autonomous weapons is now “urgent and essential”, the Oireachtas AI Commission will consider.
Bonnie Docherty, a Harvard law lecturer and expert on military weapons, will tell the committee that the use of autonomous weapons systems in peacetime is likely to violate international human rights law, in that humans can be “arbitrarily deprived of life” by systems operating without human reason.
Docherty will explain to the committee, which has been considering the impact of AI on international defense for several weeks, that autonomous weapons and AI are problematic for several legal and moral reasons, and that they “could lower the threshold for war and lead to an arms race.”
“Machines are incapable of understanding the true value of human life because they are not living things themselves. Furthermore, by relying on algorithms that reduce humans to data points, machines will instrumentalize and dehumanize their targets,” she plans to say.
She would say that autonomous weapons risk becoming discriminatory by default, given that “algorithmic biases can disproportionately and negatively impact already marginalized groups and discriminate against people based on categories such as race, gender, and disability,” as has been seen with other AI technologies such as Company X’s Grok chatbot.
She added that the various problems plaguing the use of AI in international defense are exacerbated by an “accountability gap” for the harm that could be caused by autonomous weapons.
Ms. Docherty would argue that while it is difficult to hold human operators or companies responsible for the “unexpected behavior of machines that we don’t understand,” it is equally legally problematic to hold individual programmers and developers responsible.
“To adequately address this mounting problem, legally binding measures are needed, essentially banning autonomous weapons systems that operate without ‘meaningful human control’ or systems that target humans,” she says.
Given the rapid development and change in AI technology, she would argue that informal consultations between countries should lead to full formal consultations between countries as soon as possible.
“Given the speed of technological development and the seriousness of the threat these systems pose, it is urgent and essential to move diplomatic negotiations into formal treaty negotiations,” Ms. Docherty said.
“Killer Robot”
Separately, Rosanna Fani, UNESCO’s AI advisor, will tell the committee that AI in war is not limited to the cultural concept of “killer robots.”
“AI powers many non-dynamic processes, such as intelligence analysis, decision support systems, and logistics operations,” Fani says.
“These uses still have significant ethical implications and raise important governance considerations,” she added.
