Artificial Intelligence and Arms Control

Shutterstock

When analyzing the history of arms control, it becomes clear that arms control among states is the exception rather than the rule. Many reasons contribute to the difficulty of establishing such control. First, weapons that provide significant military value are hard to relinquish, as states are often reluctant to give up something that grants them unique capabilities or decisive battlefield advantages.

The desirability of arms control is often determined by a weapon’s value, weighed against its perceived horribleness. And even if states finds it desirable to regulate a weapon practical realities might make it unsustainable.

An example is is submarines. During the early 20th century, submarines were seen as an unethical weapon. In the 1899 Hague Convention, Russia proposed a ban on submarines, but this was rejected. In 1907, nations met to codify the laws of maritime warfare, establishing a series of provisions relating to the treatment of hospital ships, merchant vessels, and prisoners. The 1907 convention did not discuss banning submarines further, but it did establish maritime laws that proved problematic for submarine operations.

The primary issue with submarines was their ability to comply with maritime law concerning attacks on merchant ships. Much of naval warfare aims to disrupt the opponent's merchant vessels to impede the war effort. According to maritime law, sinking merchant ships without providing safe passage for the crew or commandeering the ship was illegal.

Submarines, however, relied on stealth for their combat effectiveness, and surfacing to inspect a merchant ship exposed them to detection and potential counterattacks. Furthermore, a merchant ship could potentially ram a submarine, an action that would have been suicidal against a warship but feasible against a submarine. As World War I broke out the parties initially complied with maritime law, but as the hostilities escalated there were more violation, until Germany declared unrestricted submarine warfare against merchant ships.

Rather than trying to regulate specific actions, another approach might have been to ban submarines entirely. The simplicity of this approach, combined with a lack of complex rules to follow, has historically made it the most successful method of weapon restriction in warfare. Bans on landmines, cluster munitions, chemical weapons, and biological weapons are generally considered successful. The drafts of these treaties suggest that the creators knew outright bans could help stigmatize a weapon. And complex exceptions that were necessary for states to reach agreement were pushed to the fine print.

The challenge when considering AI in warfare is its broad and general application. An outright ban on AI in military operations would be akin to declaring "no industrialization" in militaries during the turn of the 20th century. Therefore, the key to regulating AI in warfare isn't an absolute prohibition but rather a strategic understanding of its potential implications and applications. One must thoroughly understand its capabilities and design appropriate regulations that would prevent misuse while still allowing for technological advancement.

To mitigate the uncertainty surrounding the use of AI in warfare, there are several proactive steps that policymakers, scholars, and civil society members can adopt. These steps include establishing dialogues at various levels to enhance understanding of the technology. Academic conferences, peer exchanges, bilateral and multilateral dialogues, and discussions in international forums are valuable for grasping how this technology could be deployed in warfare. Analysis of potential arms control measures must be tightly linked to the technology itself, and these dialogues must include AI scientists and engineers to ensure that policy discussions are grounded in technical realities.

The industrial revolution unleashed the wrath of industrial scale warfare that the world never previously witnessed. Large efforts were made to regulate the new weapons. Leading military powers at the time met to discuss arms control in 1868, 1874, 1899, 1907, 1909, 1919, 1921, 1922, 1923, 1925, 1927, 1930, 1932, 1933, 1934, 1935, 1936, and 1938. Though many of these efforts for arms control failed or faltered in wartime, their frequency highlights the significant effort and patience needed for even modest success in arms control.

Artificial intelligence offers an unprecedented leap in warfare capabilities, much like the industrial revolution. However, just like the industrial revolution's horrific consequences on the battlefield, so too could the unregulated application of AI in warfare. Much as those generations grappled with their new reality, we too must grapple with ours.

Once again we stand at a crossroad. Arms control is not a new challenge but rather an evolving one. It has been a part of our global history for centuries, with a track record of both failures and successes. Each new technological advance brings with it new ethical, political, and military dilemmas, and AI is no different. As the latest development that could potentially revolutionize warfare, AI presents us with a unique opportunity to learn from past mistakes and victories.

In conclusion, the complexities of AI in warfare and the critical need for effective arms control cannot be understated. The challenge lies in striking a balance between the desire to gain military advantage and the need to preserve ethical standards and humanitarian principles. While the path to this balance may not be clearly defined, we must commit ourselves to this essential endeavor. After all, one does not need to see the top of the staircase to take the first step.