.png)
Preeminence of Thucydidean Leviathans: Why Is It Impossible to Hold Great Powers Accountable?
Accountability presents a paradox in human affairs. Despite being one of the most important concepts in the legal and philosophical world, it is one of the least enforceable aspects of international relations. It is a “chameleon-like” concept that is described differently across disciplines. Accountability has multiple dimensions, including moral responsibility, which seeks to identify the causes of “states of affairs” to allocate commendation or blame.
Illustration by The Geostrata
However, one significant conceptualisation asserts that holding a party liable is not analogous to embodying the belief that the entity is responsible. Another conceptualization of accountability for public administration, pioneered in the 1880s, stated: “the more power is divided the more irresponsible it becomes.” This was likely intended to implement stricter controls on citizens.
Accountability is also tied to the extent to which states follow the expressed needs of their citizens irrespective of obligations and incentives of “authoritative exchange and control.” Accountability then becomes the “exchange of reasons for conduct” targeted towards “verbally” closing the chasm of action and expectation. Scholars posit that accountability depends on the voluntary disclosure of conduct by sharing information about task performance, results, and processes.
This conduct is theorised to be interrogated by a “forum” intended to find or judge the explanation or legitimacy of the course of action. The judged entity could then become socially blameworthy, which in most extreme cases could lead to scapegoating and the establishment of a “convenient other” that can exacerbate anger and aggression.
However, the core of the concept is belief, which leads to the development of commonalities of thought and ultimately transforms into calls for accountability. Notably, accountability also changes according to expectations, which is particularly important as global consensus on belief is essential for achieving accountability on universal issues such as technology. Today’s society exists in a state wherein the ability to develop belief not only as an individual but as a society is being challenged.
This development can be attributed to attempts at cognitive and information warfare that reduce the ability to form beliefs, which are a “mental architecture of how we [humans] interpret the world.” Additionally, AI has led to a fall in traffic for fact-checking and independent media websites. It has become harder to verify and then to achieve accountability, especially in cases of state involvement.
DIMENSIONS OF ACCOUNTABILITY GAPS IN THE DEVELOPMENT OF AI AND AUTONOMOUS WEAPONS
A vital aspect of accountability is that one can be “accountable for a state of affairs” without being responsible for it. For example, situations wherein a company can be held accountable for actions that are caused by decentralisation or technology devised by the firm. This makes attribution difficult due to the complex chain of command and dynamic hierarchical structures.
Currently, this becomes notable for understanding how accountability can be attributed in cases where states are involved. The majority of developments in frontier AI and autonomous weapons are being conducted by great powers, including the US, China, and Russia. This advancement route has directly contributed to low accountability, as great powers remain extremely reluctant to implement any barriers due to rising inter-state competition. Great powers are actively working to dilute legislation on AI that could lead to greater accountability not only domestically but globally.
This was witnessed when US and Russian diplomats reportedly removed a “range of provisions designed to safeguard the use of artificial intelligence in weapons” from a treaty intended for lethal autonomous regulation.
The non-binding treaty was between the member states of the Convention on Certain Conventional Weapons. Reports indicate that the US pushed for flexibility, which extended to the exercise of human judgment. The US and Russia prefer national guidelines over international legally binding rules. A watered-down treaty is likely to negatively influence norms and ultimately be counterproductive to achieving accountability, especially in cases where civilians were harmed.
Meanwhile, China depends on “developer obligations, state-backed standards, security assessments and outside testing” for AI. Therefore, there doesn't seem to be any accountability mechanism in place. Similarly, the US White House Accord on Superintelligence outlines “sensible safeguards,” but there is no compulsion for adoption by companies. There is a clause for an “independent committee” which mentions “issues” for remedy but contains no details regarding the exact purview of the range of concerns.
THE FUTURE CHALLENGES AND OPPORTUNITIES FOR ACHIEVING ACCOUNTABILITY
This calls for a new model of accountability, especially for great powers. Currently, it is difficult to allocate responsibility to states due to the Articles on State Responsibility and State immunity barriers. Furthermore, reports indicate that it is challenging to find evidence for the “forum” due to classified documentation which limits accountability.
This was highlighted in the case of ANI News vs OpenAI case which was fought in Indian courts. The company stated that it was not under the jurisdiction of Indian courts, as it has no servers or offices in the country, and it is subject to US law and is required to preserve data for training materials for concurrent American litigation. This highlights that individual states and their local courts are responsible for regulating, licensing, or supporting private companies that operate AI and autonomous weapons.
Therefore, one way to better regulate accountability would be to reprimand or penalise states whose companies are involved in the development of AI or autonomous weapons that cause civilian harm. A more effective framework geared towards accountability would call for the imposition of responsibility on states where AI intellectual property is stored and developed. Especially if the state’s law enforcement agencies refuse to regulate technology despite knowing the potential for its harm.
Analysts have stated that state responsibility for “deployment consequences and manufacturer liability for systemic failures creates powerful incentives for rigorous testing, ethical design, and restrained use while guaranteeing redress for victims.”
This would require better record-keeping, investigations, and allocation of resources from international agencies and organizations. This will also require developing countries to coalesce together and ask for new governing bodies aimed at writing rules and achieving greater accountability for technologies that will have an outsized impact on the future, such as AI and autonomous weapons.
It also highlights a dilemma, as the most powerful states, which have usually led the charge, are working on closed models that prioritise opacity. This leads to problems around transparency that is critical for accountability. Without transparency, the ascertainment of accountability is nearly impossible, as it requires demonstration that work has been conducted in accordance with agreed rules and standards and reporting fairly and accurately on performance results vis-à-vis mandated roles and/or plans.
A new paradigm shift is required that prioritizes global responsibility and accountability. The prioritisation must drive common beliefs that then propel global commitments and assist the most vulnerable of countries and communities. This is likely to be extremely taxing amid rising polarisation and people losing a sense of global “imagined community,” which is described as a transnational community with individuals who experience a shared sense of destiny and are able to express and communicate public policy concerns.
In this regard, we appear to be living in the times of the Peloponnesian War, which immortalised the Melian dialogue, including the phrase “the strong do what they can and the weak suffer what they must.” This remains pertinent as only certain elements of small states can be held accountable for their actions but not larger ones. Great or middle powers can dodge responsibility for significant developments such as AI which is then assigned to economic entities such as firms, even when they knowingly enable, authorize, export, financially support, or fail to regulate a company whose system causes foreseeable civilian harm.
Arushi Singh has a master's in Geopolitics and International Relations. She has domain experience in the geopolitics of West Asia, the geopolitical implications of great power competition in North Africa, Russia’s foreign policy orientations, and emerging technologies. She has served as a researcher at the Consortium of Indo-Pacific Researchers and a geopolitical risk analyst for a private consulting firm based in the Middle East.
The article published here is the author's own scholarly work and does not necessarily reflect the views of The Geostrata or Carnegie Council for Ethics in International Affairs. The author certifies that the article is original in content, unpublished, and has not been submitted for publication anywhere else. Further, the facts and figures quoted are duly referenced with a hyperlink to the original source, whenever needed, and are presumed to be correct.
BY ARUSHI SINGH
TEAM GEOSTRATA
Comments