纵观人类历史，技术进步已经转化为军事实力。在多数情况下，那些能更快、更有效地将新技术融入本国军队的国家都获得了相比对手的显著优势。对人工智能来说，情况可能也是如此，美中两国目前正陷入一场争夺全球人工智能优势的竞争中，而这场竞争很可能决定未来的全球格局。

美国坚信这一点，这从美国国防部副部长凯瑟琳·希克斯在今年8月28日的讲话中可见一斑。她的讲话之所以意义重大，主要原因在于其对美国军方关于人工智能和自主系统，以及技术创新的战略思考提出了很有价值的见解。

希克斯发表演讲的核心内容是美国国防部打算建立一支“数据驱动、人工智能赋能的军队”。人工智能在过去几年里非常流行，其实几十年来，各个大国一直都在研究如何将人工智能应用到军事领域。从2014年美国宣布第三个“抵消战略”开始，它就一直在为将人工智能纳入军事领域打基础。美国人工智能国家安全委员会2021年的报告或许最能说明问题。报告说，国防部远未做好“人工智能方面的准备”。报告还敦促国防部在2025年前大幅增加投资，“将人工智能技术融入作战的方方面面”。希克斯的讲话也是基于这一思路。

希克斯宣布了“复制器”无人机项目。她称该计划是国防部的一项新举措，旨在快速研发和部署“可蜂拥打击敌人的低成本海陆空无人机群”。她表示，这是一个“大赌注”，可以抗衡中国最显著的优势——即有能力把大量武器平台和人员带上战场。美国国防部希望充分利用“各领域低成本可消耗的自主系统——这些系统成本较低，火线上的人员更少，而且可以在更短的时间内对其进行更新或改进”。

这一计划将把重点放在“小型、智能、廉价、数量众多”的平台上。希克斯说，“复制器”项目的近期目标是让美军“在未来18到24个月内，在多个领域部署数千个规模的低成本可消耗自主系统”。

首先，自主系统的规模非常大，它们将适用于各个领域。鉴于美国目前是全球的技术中心，美军广泛使用自主系统可能会迫使其他国家也采用此类系统，以保持战略均势。自主系统可能也会扩散到美国的盟友和战略伙伴。

其次，更重要的是未来18到24个月这一既定时间表。这相当令人担忧，尤其是考虑到人工智能伦理和监管方面的问题愈发严峻。尽管美国声称要在“复制器”项目中对人工智能采取“负责任与合乎伦理”的态度，但这一既定时间表让人难以相信这些说法。同样值得注意的是，美军很可能已经为这一项目努力了相当长的时间，因此会有具体的规则来降低将人工智能纳入军队的风险。然而这些规范和法规将如何具体影响一场危机，还得另当别论。

最终，未来的战争将具有数据驱动和人工智能赋能的特征，而且在许多方面已然如此。我们必须更好地了解将人工智能融入自主军事系统的潜在危险。鉴于人工智能的快速发展以及各个大国对其军事应用的重视，将人工智能纳入各国军队不过是早晚的事情。希克斯在讲话中提到了“复制器”项目对美军速度和规模的影响，这很可能就是未来战争的特点：快速进行，人类作战人员与许多自主系统并肩作战。虽然对某些人来说，这似乎是一种更有效的作战方法，但自主系统带来的战争升级风险也可能太大了。

The future of warfare will certainly be data-driven and AI-enabled, and, in many ways, it already is.

Shayan Hassan Jamy

Throughout human history, technological progress has translated into military prowess. In most instances, the states that incorporate new technologies more quickly and effectively into their respective militaries have gained a significant advantage over their adversaries. The same is likely to be true for artificial intelligence (AI), with the United States and China currently locked in a competition for global AI superiority. This competition for AI and technological supremacy could very well dictate the future global landscape. 

Although China might disagree with the existence of such a technological competition, the United States firmly believes in it. This was evident in a speech by U.S. Deputy Secretary of Defense Kathleen Hicks on August 28, 2023. Deputy Secretary Hicks’ speech was significant for several reasons, primarily because it gave valuable insight into the U.S. military’s strategic thinking about China, AI and autonomous systems, and technological innovation.

At the core of Deputy Secretary Hicks’ speech was that the U.S. Department of Defense (DOD) aimed to have a “data-driven and AI-empowered military.” Although AI has gained mainstream popularity within the past few years, great powers have been looking into the military applications of AI for decades now. From 2014 onwards, when the United States announced its Third Offset Strategy, it has been building the foundation for incorporating AI into its military. The 2021 report by the U.S. National Security Commission on Artificial Intelligence (NSCAI) was perhaps the most telling. The report stated that the DOD was far from “AI-ready” and urged it to heavily increase investment by 2025 and “integrate AI-enabled technologies into every facet of war-fighting.” This same line of thought informed Deputy Secretary Hicks’ speech.

Deputy Secretary Hicks announced the “Replicator Initiative,” which she described as a new DOD initiative to develop quickly and field “swarms of low-cost air, land, or sea drones that could swarm an enemy.” She called it a “big bet” that could counter China’s most significant advantage—the ability to bring a mass of platforms and people to the battlefield. The DOD hoped to leverage “attritable, autonomous systems in all domains—which are less expensive, put fewer people in the line of fire, and can be changed, updated, or improved with substantially shorter lead times.” 

The initiative would focus on platforms that are “small, smart, cheap, and many.” The immediate objective of the Replicator Initiative is for the U.S. military to “field attritable autonomous systems at scale of multiple thousands, in multiple domains, within the next 18 to 24 months,” Hicks said. This statement deserves thorough analysis. 

Firstly, the scale of the autonomous systems is enormous and will apply to various domains. With the United States currently the technological hub of the world, the widespread use of autonomous systems by the U.S. military would likely force other states to adopt such systems to maintain strategic parity. Autonomous systems would likely proliferate to U.S. allies and strategic partners as well.

Secondly, and more importantly, is the stated timeline of the next 18 to 24 months. This is rather alarming, particularly given that issues surrounding AI ethics and regulation have gathered momentum recently. Although the United States claims to follow a “responsible and ethical” approach to AI in its Replicator Initiative, the specified timeline makes these claims hard to believe. However, it’s also important to note that the U.S. military has likely been working on this initiative for quite some time, so it would have specific rules to reduce the risks of incorporating AI in the military. How AI norms and regulations would affect a crisis, however, is a debate for another day.

Even if the United States had been planning such an initiative for years, it now feels confident enough to announce and implement it. Ukraine has acted as a testing ground for using drones and autonomous systems on the battlefield and has clearly demonstrated their power. Russia and Ukraine regularly deploy drones in military operations. The Royal United Services Institute (RUSI) estimates that Ukraine has lost a staggering 10,000 unmanned aerial vehicles (UAVs) monthly. These drones are helpful for intelligence, surveillance, and reconnaissance (ISR) purposes, as well as for direct targeting of the adversary’s military and civilian infrastructure.

Deputy Secretary Hicks also directly mentioned China as the sole target for the Replicator Initiative. She added: “We must ensure the PRC leadership wakes up every day, considers the risks of aggression, and concludes, ‘today is not the day’—and not just today, but every day, between now and 2027, now and 2035, now and 2049, and beyond.” She also mentioned that “all-domain, attritable autonomous systems (ADA2) will help overcome the challenge of anti-access, area-denial systems (A2AD). Our ADA2 to thwart their A2AD.” This is a critical point. China’s A2AD strategy focuses on the South China Sea. The United States stating that it would use drones to counter China’s A2AD strategy indicates that it is willing, directly or indirectly, to intervene militarily in the region.

China, on the other hand, holds an entirely different understanding of AI than the United States does. Although China aims to become the global leader in AI by 2030, it has so far remained characteristically secretive about its military incorporation of AI. However, this has not stopped the United States from viewing China’s AI progress as a major challenge to its global leadership.

Ultimately, the future of warfare will be data-driven and AI-enabled, and, in many ways, it already is. However, we must better understand the potential dangers of integrating AI into autonomous military systems. Given the rapid pace of advancements in AI and the importance given to the military applications of AI by major states, the incorporation of AI into the militaries of major states is a matter of when not if. Deputy Secretary Hicks’ speech mentioned the impact of the Replicator Initiative on the speed and scale of the U.S. military. That will likely be the character of future warfare: it will be fought rapidly, and human combatants will operate alongside many autonomous systems. Although this might seem to be a more effective method of warfighting for some, the risk of escalation from autonomous systems might be too great.