Introduction
Dealing with existential risks requires international cooperation. Naturally, one might expect scholars of international relations (IR) to provide the best answers regarding how states can cooperate to protect humanity’s long-term potential. Yet reading Tody Ord’s new book The Precipice as a PhD student in IR, I am surprised how little attention my field has paid to the existential threats Toby raised in the book, such as global disease, climate change, and risks from artificial intelligence (AI).
Although mainstream IR often overlooks existential risks, it does offer insight into how to make international cooperation easier. In particular, IR theory’s emphasis on the importance of national interests offers us a realistic view of international behavior. Isaac Asimov, a universalist and humanist, once dismissed decisions based on the national interest as “emotional” reactions on “such nineteenth century matters as national security and local pride.” My view is exactly the opposite: We should work with states as they are, not what we wish them to be.
This blogpost is my attempt to bridge the IR and EA communities. On the one hand, I argue that IR’s state-centrism and obsession with national interests have restricted the field from engaging with issues that concern humanity’s long-term survival. On the other, IR theory is critical for understanding how we can implement Toby’s “grand strategy for humanity” and other schemes for international cooperation against existential risks.
IR Theories and International Cooperation
I want to start with a brief overview of IR theories’ views on international cooperation.
There are three main schools of thought in contemporary IR. Realism, which is the most enduring tradition in IR thought, holds a pessimistic view of how the world works and believes that there is little hope for long-lasting, genuine cooperation.
Such pessimism originates from realism’s basic assumptions about the international system. Realists believe that states operate under “anarchy,” the fact that there is no higher authority in the international arena. As such, states must be responsible for their own survival, maintain sufficient military capabilities, and behave strategically in international competition. The competitive nature of international politics means that states can never be certain of others’ intentions, let alone trust that they will act benignly in the long-run.
These assumptions lead to the realist concern over relative gains, i.e., how much more or less they will gain from a collective action compared to their partners. For example, if two states agree to reduce their nuclear arsenal, each state will think “How much does the other get from our arms control agreement? Am I making a greater sacrifice?” According to realists, such concerns are logical because states are positional and care about maintaining an advantageous relative position. This, however, creates a significant barrier to international cooperation.
A second barrier lies in the issue of cheating. Realists tend to conceptualize collective actions as Prisoner’s Dilemmas, where each side has an incentive to cheat and let the other be the sucker. For example, even if states agreed to some AI security measures, they should still try to cut corners to get ahead of their competitors. For realists, in international cooperation, cheating is a feature, not a bug.
While we (members of the EA community) might dismiss realists as overly pessimistic and disagree with their basic worldview, the realist school does offer one key insight into international cooperation: national interests often stand in the way of collective action.
The question of collective action is exactly what another IR theory -- neoliberal institutionalism -- intends to tackle. Beginning their analysis with the Prisoner’s Dilemma, liberals believe that the repeated iteration of the Prisoner’s Dilemma game is the key to facilitating long-term cooperation. During these iterations, institutions provide many mechanisms for punishing cheating behavior and incentivizing cooperation. These mechanisms include tit-for-tat retaliation and opportunities to “link” issues together in a negotiation. For instance, a pair of states in an arms control negotiation might link strategic arms reduction with trade -- if you default on your promises to limit arms, we will impose higher tariffs on your goods.
In addition, neoliberal institutionalists believe that states cooperate for absolute gains or shared interests. Especially in non-security related areas, such as climate change, international finance, and global health, the common interest is often the main driver of collective action. In other words, liberals believe that long-term cooperation is possible despite states’ self-interest, as long as there are institutional mechanisms for punishing defection and sufficient interest in the common good.
A final IR theory I want to mention is constructivism, a school of thought that has gained prominence over the past thirty years. Unlike realism and liberalism, which make many ontological assumptions (e.g., states are interested in survival, states are rational actors), constructivism emphasizes that international politics is socially constructed. For example, constructivists believe that “international anarchy” is not a given state but a condition created by shared ideas, narratives, and patterns of state interaction. By using the language of “national sovereignty” in international forums, rejecting the rule of international courts, or constructing narratives that challenge the authority of international institutions (e.g., the UN, WHO, or WTO), states “construct” the condition of international anarchy.
Constructivists also believe that ideas, identities, and narratives influence how easy it is to cooperate. The prominent constructivist scholar Alexander Wendt argues that there are “three cultures of anarchy,” a Hobbesian culture where states identify each other as enemies, a Lockean culture where states see each other as rivals, and a Kantian culture where states view each other as friends. Needless to say, the ease of interstate cooperation depends on the dominant culture in the international sphere. If everyone were to think like a realist, cooperation would be very hard. In contrast, if everyone were to think like an altruist, cooperation would be much easier.
Similarly, the “national interest” is also socially constructed. Nineteenth-century great powers often see their colonies as integral to their “national interest” but few states today believe in such blatant colonialism. Similarly, eighteenth- or nineteenth-century statesmen would have been bewildered by contemporary foreign policy makers’ interest in promoting democracy and human rights. While realism and liberalism hold static views of the states’ interests, constructivism believes that the so-called “national interest” can be shaped by changes in social norms and values.
Existential Risks and the Problem of the National Interest
As we talk more about the different IR theories, a clear pattern emerges: IR scholars pay very close attention to the concept of national self-interest. Whether it is the realists, the liberals, or the constructivists, IR theorists have very carefully examined the effects of the national interest on international cooperation.
But what are the potential problems with this focus on national interests?
First, IR scholars, and perhaps the entire foreign policy establishment, tend to concentrate on the national interest but overlook the fact that humanity has a shared future. Even when IR researchers and policy-makers discuss existential risks, they tend to do so within the framework of national interests. For example, Robert Jervis’ thesis on the nuclear revolution argues that the advent of nuclear weapons makes war “unwinnable” because the costs for fighting a nuclear war are prohibitively high for both sides. What is at stake in this argument is the human and economic costs for each belligerent, not humanity as a whole. However, by stepping out of their national-interest-centric worldview, decision-makers might be more likely to consider policies that protect humanity’s long-term survival.
Second, the field of IR and foreign policy decision-makers have a rather narrow definition of national interests. They are primarily concerned with traditional security and economic issues, such as increasing national military and economic power, maintaining security alliances, and gaining favorable positions in international trade. These issues bias decision-makers to focus on the short-term and disincentivize long-term planning. As for now, the UK is perhaps the only country that has seriously discussed legislation to safeguard the welfare of future generations.
Third, the state-centrism that IR theory exhibits downplays the importance of non-state actors. Admittedly, states and governments are critical actors in international cooperation but non-state actors could also contribute to international governance. For instance, the most recent development in U.S.-China AI cooperation is the track-II diplomatic talks between Brookings president John Allen and former Chinese foreign affairs vice-minister Fu Ying. (1)
Lessons from IR Theory
Although I have listed the deficiencies of IR theory, I nevertheless believe that it offers insights into how cooperation against existential risks can be more effective. Drawing from the three different schools of thought and also considering their drawbacks, I want to offer a couple of suggestions for EA-aligned researchers working on mitigating existential risks:
- Try to come up with policies or technical solutions that curtail states’ economic and military capabilities as little as possible. It would be very politically difficult to implement policies that require large sacrifices (e.g., abolishing nuclear weapons).
- Think about shaping the international norms surrounding your issue area. Norms shape how states conceptualize their national interests and their cost-benefit calculations when pursuing a policy.(2)
- Consider facilitating cooperation between non-state entities. As states face the burden of upholding the national interest, their behavior is often more constrained than individual companies, labs, scientists, and non-profit organizations.
Despite spending most of this blog post talking about the self-interestedness of states, I want to end on a more optimistic note: There are many reasons to believe that humanity’s moral circles will further expand in the future and that state decision-makers will attach greater weight to more distant entities. The effect of reasoning,(3) the improvement of human welfare,(4) and the increased contact of individuals from different backgrounds all induce us to expand our circles of moral concern. (5) Perhaps in the future, the so-called “national interest” will be less of a barrier to international cooperation than it is today.
(Many thanks to Amy Labenz for suggesting the idea of writing this forum post to me and Aaron Gelter for his wonderful edits and suggestions! Aaron has helped me make the post much more readable. If you’re interested in the topic of international cooperation against global catastrophic risks, please check out my EAGxAsia-Pacific talk.)
(1) Track-II diplomacy refers to the diplomatic contact between non-state actors such as think-tanks, former government officials, and private enterprises.
(2) The emergence of a moral taboo against the use of chemical weapons led to the signing of the Chemical Weapons Convention, which in turn changes how states view chemical weapons and imposes a high cost on states that violate the norm of non-chemical weapons use. Similarly, the moral taboo against nuclear weapons use might be a reason behind the 70 years of nuclear peace.
(3) Steven Pinker and Peter Singer have both indicated that reason promotes the expansion of one’s moral circle. By asking “what is the appropriate limit of my moral circle of concern,” individuals include more and more entities in their moral circle.
(4) The World Value Survey (WVS) finds a shift from “survival values” to “self-expression values” as countries develop economically. The former emphasizes physical and material security, while the latter emphasizes issues such as environmental protection, minority rights, and gender equality.
(5) This is known as the “contact theory” in social psychology.
I agree that IR is important for EA. I would be particularly interested to hear your opinion on collapse of civilization scenarios here.
Here are a couple of thoughts from an IR perspective:
Case 1: 50% population loss, no infrastructure damage, no climate change (e.g. a limited pandemic)
Case 2: 90% population loss, infrastructure damage, and extreme climate change (e.g. nuclear war that caused nuclear winter)
Case 3: a catastrophe causes the deaths of 99.99% of people (leaving 800 thousand survivors), extensive infrastructure damage, and temporary climate change (e.g. a more severe nuclear winter/asteroid impact, plus the use of biological weapons).
As you can see, the higher the percentage of deaths, the less IR theory has to say about the situation. I think the central insight is that the number of survivors determines people's identities (whether they think of themselves as part of a nation or as part of humanity as a whole). This, in turn, affects whether there will be internal competition over resources among different groups. In other words, we cannot simply assume that the survivors can effectively allocate resources among themselves. Internal resource competition can lead to excess/unnecessary deaths. Paradoxically, the percentage of such excess deaths might be Case 1 > Case 2 > Case 3.
I am very uncertain about these predictions but would love to hear your thoughts!