AI applications in automatic weapons - a double-edged sword

While autonomous weapons are increasingly used, scientists and human rights advocates are calling for stronger regulation of their use.

A soldier from the Australian 1st Armoured Regiment remotely controls a weapons system (Photo: Australian Defence Force).

The world is turning to autonomous weapons powered by artificial intelligence in armed conflicts, while scientists and human rights advocates are calling for stronger regulations to guide the use of these weapons systems.

Last month, the Australian Defence Force unveiled a new weapons suite that includes drones, robotic combat vehicles, unmanned tanks and mine-clearing robot dogs. These are semi-autonomous weapons that still require some human control.

Lieutenant Colonel Jakes Penley (left) and Lieutenant Colonel Adam Hepworth (right) at Puckapunyal Military Base, Australia (Photo: ABC Shepparton: Charmaine Manuel).

The Australian Army sees the potential for artificial intelligence (AI) to be used in military settings, including administrative and combat roles, but it is important to maintain human oversight of decision-making, said Lieutenant Colonel Adam Hepworth, director of autonomous systems and robotics.

He said all systems used by the Australian military had to pass the testing process in accordance with domestic obligations and international law.

However, some leading scientists say stronger regulatory frameworks are still needed to control the use of these military systems.

Leading AI professor Toby Walsh at the AI ​​Institute, University of New South Wales, Australia, says artificial intelligence is a double-edged sword (Photo: TU Berlin/Press/Christian Kielmann).

According to Professor Toby Walsh at the AI ​​Institute, University of New South Wales, Australia, the use of AI and automated tools is a double-edged sword.

On the positive side, some applications have many benefits, such as mine-clearing robots. 'No more people should be put at risk of losing their lives or limbs as a result of mine-clearing,' he says. 'This is a job that is perfectly suited to robotics.'

We could use robot dogs to clear mines (Photo: ABC Shepparton: Charmaine Manuel).

" But at the same time, I think there are cases where leaving decision-making to algorithms will lead us to very dark, very dangerous outcomes," Professor Walsh added. According to him, the software used in weapons is very easy to steal, copy or influence.

Machines cannot determine the value of human life.

Australian Human Rights Commissioner Lorraine Finlay said the use of autonomous weapons posed challenges to the principles set out in international human rights law. She said the Geneva Conventions' control system was flawed because autonomous weapons were not designed to learn from each mission and technology was constantly evolving.

There are concerns about whether machines can truly deliberate, because they do not understand the value of human life.

"It's not enough to simply say that there are humans involved somewhere in the whole process of using these weapons, but to clearly define their place in that process, what their authority is, whether they are the key decision makers or whether this is delegated to machines," said Ms. Finlay.

Drones on display in Puckapunyal (Photo: Australian Defence Force.)

There are no specific regulations on lethal autonomous weapons, and many scientific and human rights opinions argue that such regulations are needed.

In November 2023, Australia voted in favour of a UN resolution urging the international community to consider the challenges posed by autonomous weapons systems. 'Now is the time to address these issues and ensure that safeguards are put in place immediately,' Finlay said.