AI is on track to ‘eliminate’ human decision-making in warfare

In a briefing at UN Headquarters in New York, UNIDIR Director Robin Geiss asked journalists a pointed question: “Will we let machines target and kill humans?he asked, warning that AI is increasingly “on track to override human decision-making” in the sector.

The UN Secretary General has repeatedly urged stakeholders to ensure we do not cross this “moral red line.”

Master and commander

AI is now increasingly being given responsibility in military affairs, from intelligence analysis and target identification to drone operations and orchestrating cyber attacks.

Most importantly, this compromises human judgment, including in the nuclear field, Geiss stressed.

The shift is dramatically increasing the pace of battlefield operationsshortens the time between identifying and attacking a target and allows troops to attack an unprecedented number of targets every day.

At the same time, many of these developing AI systems are opaque, and the head of UNIDIR warns that they are opaque increasing the risk of errors, miscalculations and unwanted escalations.

As AI tools become easier to use, it will also be easier for non-state actors to engage in increasingly sophisticated military operations, he said.

Geopolitical convergence

Mr. Geiss is aware of that Geopolitical competition and tensions are “serious obstacles” to regulating AI engagement in the military sector.

But he also emphasized that concerns about these risks are increasingly being felt – regardless of the political outlook – including among the world’s major countries.

In September, for example, the United States proposed an AI “notification mechanism” with China, so that the world’s two largest military spending countries could warn each other about AI incidents that could threaten national security.

Repeated calls from the UN Secretary General for stronger regulation of military AI and a legally binding agreements regarding lethal autonomous weapons systems are also “on the rise [finding] echoes in the international community,” said the Director of UNIDIR.

Multilateral response

UNIDIR is working actively to create some of the necessary foundations for multilateral cooperation on this issue.

On Thursday, UNIDIR Center of Excellence on AI, Peace and Security will launch a new multi-stakeholder forum on AI in international peace and security as well as in the military field – the first and only such forum in the UN system so far, Geiss said.

This forum will focus on three areas:

1. Develop a framework for responsible industrial behaviorwith practical and voluntary guidelines that companies should follow at the design and development stage of military AI tools.

2. Develop a “AI safety lexicon” to clarify key concepts such as autonomy, safety or securitywhich are currently used in very different ways by different stakeholders – creating ambiguity that hinders trust and collaboration.

3. Addressing the risk of AI abuse by non-state armed groups to support operational planning, implementation and logistics.