The impact of autonomous technologies can be felt virtually across all spheres of human activity. The trend does not replace humans all together but makes the system’s functioning and processes significantly shift towards unmanned and increasingly autonomous platforms. This delegation of control to such systems is continuously increasing alongside improvements in enabling technologies. As John Markoff notes, Boeing 777 pilots only spend about seven minutes out of every flight manually flying the plane, while Airbus pilots manually fly only about three and a half minutes of every flight. The global market for robotics and systems utilizing AI is expected to reach US$153bn by 2020.
Therefore a host of countries, including China, India, Canada, Denmark, Finland, France, Italy, Japan, Mexico, Singapore, South Korea, Sweden, Taiwan, UAE, US, UK, and the EU have released strategies to promote and use the development of AI. In 2019, the US published its Department of Defense AI Strategy, which aimed to accelerate the integration of AI across the US armed forces. And unsurprisingly, Russian Military Industrial Committee approved the plan to have 30% of Russian combat power consist of entirely remote-controlled and autonomous robotic platforms by 2030, keeping in line with the country’s industrial goals.
The relationship between economic and military strengths is the most evident security concern which arises from the development of AI, because the obvious decline in demand for labor and the changing dynamics between capital and labor is causing population size to become relatively less important as an instrument of national power. Greg Allen and Taniel Chan argue that this “race” in AI would not just lead to military, information, and economic superiority; but also to serious changes in national security strategies, organizations, and resource allocations. According to Bard’s 2019 Drone Data Book, already the number of countries with military drone programs has jumped to 95, with Turkey to even begin operating kamikaze drones on its Syrian border from 2020. The Blowfish A2, an autonomous helicopter drone outfitted with a machine-gun, was just in the news as the new sales pitch from China to Pakistan and some Middle Eastern countries.
The adoption of AI into defense related systems by such multitude of countries cannot be wishfully undone or terminated, therefore the technology community should pursue the more viable technology management goals and problems, which brings us to the subject of multinational interoperability.
Interoperability is generally defined as a characteristic of a product or system, whose interfaces are completely understood, to work with other products or systems, at present or in the future, in either implementation or access, without any restrictions. There are a number of states developing AI-enabled capabilities that have expressed an interest in maintaining interoperability with their allies and partners. 2019 US Department of Defense AI Strategy, the EU’s 2019–2020 Work Programme for the European Defense-Industrial Development and the 2019 Work Programme for the Preparatory Action on Defense Research strongly reference interoperability in AI-enabled capabilities.
If we conceptualize a coalition or alliance as a “decision-unit” — the ability of heterogeneous (of separate types and national ownerships) autonomous systems and their parent organizations to cooperate seamlessly in pursuit of common objectives will be critical to achieve the requisite international posturing and military/diplomatic effects in an AI-centric operational environment. BAH’s enormous ten-year $17.5 billion contract with the American DISA to provide a “globally accessible enterprise information infrastructure” can be (maybe) seen as a precursor and preparation for such interoperable multinational cooperation and struggle for information dominance. With its capacity to be constructive or cause disturbances — the interoperability in context of autonomous weapons and systems could be competing states’ trump card in the multipolar world’s emerging future.
While the operating models for countries trying to increase synergy with their allies consists of methods such as multilateral exercises etc, these are primarily more directed towards training and performance evaluation than towards multinational interoperability. And these measures are already hindered in the desired goals by differences in policy, technical standards, organizational cultures, mutual trust between allies, technological inequalities, as well as by volatility in international politics. The ongoing integration of autonomous systems into the mix further increases the complexity of the situation.
Legislation and regulation gaps often occur with societal transitions to new technologies which leave decision-makers in companies and governments wanting. From the vantage point of international security, if the present price decline and capability growth of robotic and AI systems persists, we will have to develop mechanisms which will:
1. Maintain a high degree of adaptable multinational interoperability, and
2. Mitigate the geopolitical risks and socio-technical vulnerabilities that these systems and their integration opens
It is time that the larger technology community, and especially cybersecurity community, starts looking at hazards as control problems and not as human or machine failures. For the present circumstances, as well as Artificial Intelligence’s still embryonic ability to overcome geographical limits and influence physical spaces and political behavior, pressure us to take a more holistic view of AI safety, one which is well integrated with the dimension of international security environment.