Machines which decide who lives and dies

Located in the centre of a picturesque Northern Italian city, the University of Mantova welcomes hundreds of students on a daily basis. Today, however, the University’s Aula Magna hosts Mariarosaria Taddei, professor of Digital Ethics and Defence Technologies at the Oxford Internet Institute, and Dario Bassani, editorial director of cultural platform Lucy sui mondi, for a far-reaching debate on the use of AI in modern warfare.

There is a difference between automated and automatic weaponry, Taddei points out. In traditional armed conflict, when soldiers lower their weapons, their humanity leads to fewer victims, to the lesser evil. However, compassion has no place in code-driven reasoning. Moreover, AI’s rapid and independent learning and evolution can produce unpredictable, often undesirable outcomes. In a conflict situation, stakes are incredibly high: human lives, including those of civilians, are at risk. If deemed necessary for military purposes, AI can and has targeted even kindergartens, as demonstrated by American AI drones operating in Iran.

Accountability presents another challenge. AI's opaque chain of decision-making renders determination of responsibility almost impossible if wrongful conduct occurs. Where no actor can be held accountable, warfare is ethically doomed from the start. It is inherently atrocious.

While private enterprises advance AI at remarkable speed, international law struggles to keep pace, leaving the use of force unchecked and unbalanced. The risks extend beyond foreign battlefields: in a regulatory void, governments can use AI against their own citizens. “The likelihood of atomic war is lower than a drone, led by AI, targeting you, even wrongfully.” While it is not possible to stop the evolution of AI altogether, international limitations are essential to ensure its military use is progressively decreased.

Because unfortunately, the ultimate danger is a battlefield where machines, not humans, decide who lives and who dies.