GENEVA, Nov 18 — With the future of lethal autonomous weapons systems under debate, a Red Cross expert is sounding the alarm over “killer robots” that “reduce human beings to data”.
Gathered for a conference this week at the United Nations in Geneva, countries are mulling whether to seek binding rules for such weapons.
Laurent Gisel, who heads the weapons unit at the International Committee of the Red Cross (ICRC), said there was an urgent need to establish such rules before unpredictable and unmanageable weapons are deployed in large numbers.
“The debate on autonomous weapons is very often associated with drones,” he told AFP in an interview, stressing though that “autonomy is a capability that can be integrated into any type of weapon, including a missile or a bomb”.
Warfare is undergoing a radical transformation as armies seek to capitalise on artificial intelligence to boost their fighting abilities.
While the concept of weapons capable of reacting without human intervention is not new, the impact of AI and the speed of development has spurred calls for regulation.
Autonomous weapons “have existed for decades”, Gisel said, pointing for instance to antipersonnel mines, and to warship missile defence systems capable of automatically triggering a response to detected threats.
When faced with an incoming missile travelling at very high speed, “soldiers oversee the system, but they don’t approve each individual strike”, Gisel said.
In this context — on the high seas and in response to missiles with identifiable signatures — he said the risks remained relatively limited.
‘Loitering munitions’
But rapid technological developments are bringing new risks.
“What concerns us in particular are the developments we are seeing on today’s battlefields,” Gisel said, pointing to the emergence of offensive weapons designed to “seek out targets”.
He cited the dangers of so-called “loitering munitions”, which are capable of searching for an enemy radar and then diving onto it to destroy it.
Because such weapons can operate autonomously, there is a risk that they might detonate on a radar located on infrastructure also used by civilians, he explained.
That is at odds with international humanitarian law (IHL), he warned.
IHL, enshrined in the Geneva Conventions of which the ICRC is considered the key guardian, requires the parties to a conflict to distinguish between civilians and combatants, and between military targets and civilian infrastructure.
Reducing humans to data
Gisel stressed that the goal is not to ban every weapon that uses some form of automation.
The ICRC is calling for a ban on two categories of autonomous weapons: so?called “unpredictable autonomous weapons” and those designed or used to directly target humans.
The latter in particular raise ethical questions, according to the organisation.
“These weapons reduce human beings to data, which is captured by sensors and analysed by algorithms,” Gisel said.
“It is this machine process that ultimately decides who lives and who dies.”
The expert also pointed to risks linked to how situations on the ground can evolve.
Between the moment when a weapon is activated and the moment it fires, a combatant might surrender or become wounded — changing their status to one protected under international law.
But the ICRC warns autonomous weapons may not be capable of taking that change into account and responding appropriately.
And it raises fears about autonomous weapons falling into the hands of rights abusers, with disastrous effect.
“Should anti-personnel autonomous weapons be developed, they would certainly be used by terrorist groups, criminal organisations and states seeking to repress their own populations,” Gisel warned.
He also highlighted how autonomous AI-based software had repeatedly exhibited unexpected behaviours well beyond the parameters established by its designers.
“Just imagine the risk,” he said, “if such unpredictable agents were integrated into a weapons system”. — AFP