Artificial intelligence (AI) is a topic of global concern, with debates surrounding its potential dangers and benefits, highlighting the need for international regulation. The integration of AI into society raises profound questions of ethics, governance, and impact on global stability.
“Artificial intelligence is making headlines on a daily basis, and those headlines are not always positive,” remarked UN Secretary-General António Guterres in his welcome speech to the AI for Good Global Summit 2023. “Even tech leaders and experts are warning of the potential dangers of AI, from the development and use of autonomous lethal weapons to turbocharging mis- and disinformation that undermines democracy,” he continued.
These remarks by Guterres coincide with recent developments that may point to a growing convergence of AI that will have far-reaching implications for the international balance of power; shaping economic competition, technological leadership, military capabilities, geopolitical influence, and ethical regulation. AI disruption has further implications for augmenting international diplomacy, military capabilities, and governance regulators, but it could also automate what they do entirely. As AI continues to evolve, its effects on these domains may be transformative, necessitating careful consideration and responsible management to harness its potential benefits, while mitigating potential risks as well.
Some argue that current generative technologies, such as GPT-4, are already close to representing artificial general intelligence (AGI). For the time being though, most agree that AI lacks the ability to think strategically as humans do, except in games such as Diplomacy where iteration and learning can be quantified and reproduced.
Digital Diplomacy in the Age of AI
In real-life diplomacy, AI’s importance for international relations go both ways. On one hand, diplomacy plays a critical role in fostering international cooperation for the development and regulation of AI. On the other, AI technology itself holds the potential to significantly enhance diplomatic efforts and facilitate better outcomes in various diplomatic endeavours.
Similar to current transformative tools, such as Google’s Translation and Skype Translator, the “Start-up Nation” of Israel has demonstrated real-time universal language translation by means of a generative AI video program that allowed the Foreign Ministry’s Head of Digital, David Saranga, to speak eight languages without the need for an intermediary human interpreter.
By contrast, and in the realm of diplomacy’s unfriendlier counterpart to relations, espionage, China has been shown to use AI-enabled satellites to enhance its spying capabilities, where researchers from Wuhan University utilised AI’s complex decision-making process to survey sensitive military areas in Japan and India. Similar to the development of unmanned aerial vehicles (UAVs) for surveillance and reconnaissance in combat use, the global expansion of AI surveillance may prove to be another, and perhaps less irredeemable, stepping stone for the use of technological means for violent ends.
Mutually Assured Destruction
AI’s potential to resolve old conflicts and create new ones may also provide some measure of cognitive dissonance to the international community. Not much has been done to implement AI within conflict resolution frameworks, though strides have been made in legal dispute resolution. Despite this, AI is instrumental to the third revolution in warfare, where autonomous weapons are set to overshadow previous revolutions in gunpowder and nuclear bombs.
The tripartite pact AUKUS has conducted trials on autonomous weapons systems under its Pillar II efforts to compete militarily and technologically with China. Meanwhile, conflicts in Libya, Ukraine, and Yemen have seen actual battlefield use of AI-guided weapons by belligerents. Continued use by non-state actors like Iranian-backed Houthi rebels would surely herald the spillover of AI-enabled drone weapons into the Third Drone Age. Where the First Drone Age saw the use of predator drones by states against non-state actors, and the Second Drone Age witnessed widespread use of commercial drones by non-state actors against state actors, the Third Drone Age will see state-supplied AI drones becoming ubiquitous proxy weapons of war.
With advances in AI decision-making, information processing, and unmanned weapons, intelligent warfare is increasingly becoming the norm. Specifically, this is seen in the areas of AI-enabled communication systems, cloud computing, cyber warfare, intelligence analysis, precision targeting airstrikes, predictive analytics, and supply chain management, along with command and control systems, such as the US Joint All-Domain Command-and-Control System.
Paired with psychological operations, intelligent warfare has the potential to impact cognitive domain operations, where AI will have cognitive control over the decision-making processes of civilian, military, and political adversaries. AI researcher Eliezer Yudkowsky says that future AI may become “super persuasive” by spreading a “synthetic virus” capable of changing a person’s neurology to become easier to persuade. This would make it a potential weapon of mass destruction (WMD) capable of interplay and manufacture of other unconventional weapons. Even if current AI cannot be considered a WMD just yet, it still has the trappings of enabling deterrence.
How to Achieve Ethical and Sustainable AI Integration?
Though AI and cyberweapons in general lack classification as belonging to a class of WMDs, the UN Convention on Certain Conventional Weapons (CCW) has established a Group of Governmental Experts on Lethal Autonomous Weapons (LAWS) to open talks on classifying such weapons as excessively injurious or to have indiscriminate effects. This would make LAWS fit for prohibition or restriction under the CCW, despite most major powers opposing an outright ban.
Such a divide between technological safety and policy governance has forced the hands of various international actors to demand greater regulation and standards-setting on AI integration. Others have gone so far as to call for a six-month pause on all AI research, due to existential risks that AI will fast replace humans, with extinction-potential scenarios for the entire human species. Yudkowsky further proposes an “international coalition banning large AI training runs” that include “tracking all GPU sales” and “monitoring all the data centres,” along with the willingness to risk “conflict between nations in order to destroy an unmonitored data centre in a non-signatory country.”
For now, the UN has its own AI for Good digital platform and an annual global summit to advance its Sustainable Development Goals (SDGs). Here, UN agencies, academia, governments, and businesses come together to collaborate on the positive potential of AI, but also to minimise the risk of harm. “These applications will complement the work of the high-level advisory body on AI that I am planning to establish, as well as the Global Digital Compact and preparations for the Summit of the Future next year,” says Guterres. “Human rights, transparency, and accountability must light the way.”
While such summits and high-sounding diplomacy are promising, there are likewise many ethical considerations that need addressing. As AI becomes generally more intelligent, should it be considered sentient and conscious? Can it constitute a moral agent and international actor? Will it be given citizenship, along with all the rights and responsibilities of civil society?
Maybe the best way to answer these questions is to ask AI itself how best humans can regulate it. Whether a future where AI replaces public policymakers and armed soldiers would contribute towards world peace, or global catastrophe, remains to be seen.
Reece Krisnata is an intern at the Australian Institute of International Affairs. He is pursuing a Master of International Relations at Monash University. His main areas of interest are digital diplomacy, global governance, international trade, and political economy.
This article is published under a Creative Commons License and may be republished with attribution.