Facebook
People take part in a demonstration as part of the "Stop Killer Robots" campaign organized by German NGO "Facing Finance" to ban what they call killer robots on March 21, 2019 in front of the Brandenburg Gate in Berlin.

In November 2020, a top Iranian nuclear scientist was killed when his car came under machine gun fire. The attack sparked outrage and confusion, particularly over who pulled the trigger. 

But, according to Iranian officials and news reports, no one did.

The assassination was reportedly carried out by an Israeli remote-controlled machine gun that used artificial intelligence (AI) to target and kill the scientist, Mohsen Fakhrizadeh. Secrecy continues to surround the attack, but revelations of a sophisticated “smart gun” have added fuel to the ongoing debate over the morality and practicality of AI weaponry.

In this lesson, we’ll explore the debate and detail efforts by governments and international organizations to regulate AI’s use in conflict. 

What is artificial intelligence, and how is it used militarily?

AI refers to machines or systems that can make decisions, perform tasks, and—in theory—improve performance without human involvement. It exists in everyday technologies including customer service chatbots and email spam filters.

AI can also enhance military capabilities, including many nonlethal functions such as systems that record and analyze data from aircraft sensors to better predict engine failures. In fact, the U.S. Department of Defense has over six hundred AI programs in use or in development, with funding for AI hitting $874 million in 2022.

U.S. Secretary of Defense Lloyd Austin delivers remarks at the National Security Commission on Artificial Intelligence Global Emerging Technology Summit on July 13, 2021 in Washington, DC.

Additionally, AI is used in lethal autonomous weapons systems (LAWS), which can locate, target, and kill without human involvement. LAWS, for example, could be affixed to an autonomous drone or tank and use recognition technology to identify enemy targets. When a target, say, Mohsen Fakhrizadeh is located, a weapon could be programmed to fire.

LAWS have sparked controversy, with critics arguing that human beings should never entrust computers with the decision to kill.

What are opponents of AI weaponry saying?

For many, LAWS cross an ethical red line.

Critics argue that LAWS remove nuanced decision-making in life-or-death scenarios and could be prone to programming mistakes or hacks. For instance, detractors fear autonomous weapons could malfunction, fail to distinguish between civilian and military targets, or attack disproportionately—all of which could harm civilians and violate international law. They also argue that LAWS create accountability issues because no individual decides when to kill. 

In recent years, numerous Nobel Peace Prize winners, dozens of governments, the Vatican, and over 180 nongovernmental organizations have called for a ban on LAWS. Many others have called for strict limits and regulations around LAWS. In fact, in 2017, over one hundred AI industry leaders—including big names like Elon Musk—published an open letter urging world leaders to regulate AI weaponry.

These can be weapons of terror, weapons that despots and terrorists use against innocent populations, and weapons hacked to behave in undesirable ways.

—open letter from AI industry leaders, 2017

What are advocates of AI weaponry saying?

Proponents argue that LAWS can boost a military’s strength, creating stronger deterrents that prevent conflict. And if war does break out, advocates maintain that AI can make fighting more efficient and targeted, removing human error and limiting loss of life.

Take the killing of Fakhrizadeh. His wife walked away unharmed from the attack despite sitting inches away from her husband. Iranian investigators attributed the shooting’s pinpoint accuracy to the weapon’s advanced facial recognition capabilities.

This photo released by the semi-official Fars News Agency shows the scene where Mohsen Fakhrizadeh was killed in Absard, a small city just east of the capital, Tehran, Iran, Friday, Nov. 27, 2020.

Moreover, advocates claim that LAWS are necessary for defense because other countries are developing them. The U.S. military, for instance, notes AI advances in China and Russia as partial motivation for pursuing AI technology. Proponents also contend that banning LAWS could inhibit AI research more broadly, limiting the development of new technologies that could help civilians and society at large.

Finally, advocates maintain that AI weaponry can be developed ethically. To this end, the U.S. Department of Defense and the North Atlantic Treaty Organization have recently created principles they say will facilitate such responsible development, such as mandating robust accountability and oversight processes and consciously addressing race, gender, and other biases in AI programming.

What is the state of AI in combat?

Today, LAWS are not in widespread use. But their development is increasing.

Experts believe several countries possess or are developing LAWS, and numerous private defense manufacturers now sell them. For instance, Israel’s leading weapons manufacturer, Israel Aerospace Industries, has reportedly exported AI weapons to Chile, China, India, South Korea, and Turkey.

China’s military and defense contractors are also making inroads in AI development. In 2019, Chinese manufacturer Ziyan caught the world’s attention when it released AI drones that could autonomously swarm and attack targets. China has reportedly exported the autonomous drones and other weapons systems to the Middle East.

Certain countries have begun to use AI weapons in combat. In 2021, the United Nations reported that Libyan forces used an armed AI drone in an offensive during the country’s civil war the previous year. The drone—a Turkish-made LAWS known as a Kargu-2—attacked retreating militia forces without human involvement.

The incident in Libya sparked scrutiny and furthered the debate over the need for regulations on LAWS. However, as is often the case, weapons technology has advanced much more quickly than diplomacy can respond.

Efforts to regulate AI weaponry

In 2014, the UN Convention on Certain Conventional Weapons (CCW), which tries to ban or restrict an array of weapons systems, met for the first time to discuss LAWS. So far, the CCW has failed to garner consensus on the issue.

Dozens of governments have supported a global ban on LAWS. In fact, Pakistan, which has endured sustained U.S. drone strikes [PDF] for decades, was the first country to call for their prohibition.

Meanwhile, countries such as the United States, Russia, and others with powerful militaries have opposed a ban. China has supported a ban on using LAWS but not a ban on their development.

In 2019, the CCW made some progress in establishing guardrails around LAWS. The body adopted eleven guidelines, including steps to prevent terrorist groups from acquiring such weaponry and rules prohibiting LAWS from having human-like traits so the line between human and machine is never blurred.

However, the CCW failed to take any significant further steps during their last meeting on the subject in December 2021.

The AI future is here

AI’s use in combat remains in its early stages, and much of the AI technology militaries are developing is for nonlethal use. So, the world is unlikely to witness a state-sanctioned robot war anytime soon. 

But the reality of AI weapons is no longer constrained to science fiction hits like I, Robot or Black Mirror. The technology exists today, and its development and use are accelerating.

Referenced Module