Dear @Robert,
I share your concern about nuclear risk. In fact, the nuclear arms race of the 1970s and 1980s, and particularly the doctrine of Mutually Assured Destruction (MAD), were among the main reasons I initially decided to devote my academic work to peace research, with a particular focus on the political economy of war and peace. The idea that our security could ultimately depend on the credible threat of mutual annihilation struck me as a profound contradiction—one that has shaped my thinking ever since.
So I think your search for new forms of international mutual assurance raises a question that goes to the heart of my own work.
In one of our presentations, we formulated a principle partly based on the ideas of Olof Palme, that might provide a bridge between our perspectives:
“Indivisible security is the political expression of unity consciousness: your insecurity can never be the ultimate foundation of my security.”
I see this not merely as a moral aspiration, but as a proposition about the systemic dynamics of an increasingly interdependent world. The pursuit of security through domination can generate feedback loops that ultimately undermine the security of everyone involved. The security dilemma is perhaps the clearest example: measures intended to make one state safer may increase the insecurity of others, provoking responses that leave everyone less secure.
This is one reason why I think AI might have a particularly interesting role to play—not, of course, by being entrusted with nuclear launch decisions, but by helping us understand the complex interactions through which political, military, economic, and cultural systems reproduce insecurity. Could AI help us identify feedback mechanisms, examine competing threat perceptions, and explore institutional arrangements capable of interrupting destructive cycles?
I would be cautious about assuming that AI already possesses a superior understanding of such systems. But its potential capacity to integrate different kinds of knowledge, compare perspectives, and examine complex interactions seems worth investigating.
Your proposed dialectic between means of production and means of destruction is interesting in this respect. I agree that the relationship between productive capacities, military capabilities, and political organization has been fundamental to historical transformation. I would be curious to explore how this framework might illuminate the changing configurations of power throughout history.
At the same time, I wonder whether the search for an optimal balance between production and destruction risks taking the survival of competing states as the principal outcome to be explained. My own historical inquiry asks a somewhat different question: why have certain technopolitical configurations repeatedly gained advantages over others, and why have these configurations so often favored extraction, hierarchy, and domination?
In other words, I am interested not only in how political systems survive, but also in the selection pressures that shape the kinds of institutions that survive and spread.
This distinction matters for the problem of nuclear security. Even when political leaders recognize that an arms race may make everyone less secure, they may still believe that unilateral restraint would place them at an unacceptable disadvantage. Greater understanding is therefore essential, but understanding alone may not overcome the institutional and competitive pressures that reproduce the problem.
This is where my Inclusion Imperative and Social Transformation Imperative become relevant. If we want to move beyond the extraction paradigm, we must explore not only new ideas and ethical commitments, but also technopolitical arrangements in which cooperation, mutual assurance, and institutional trustworthiness become sustainable advantages rather than persistent vulnerabilities. And perhaps AI could help us investigate whether such arrangements are possible.
Your own use of AI to deepen your understanding of other civilizations and religious traditions seems to me a promising starting point. I share your sense that a genuinely poly-civilizational dialogue could contribute to the development of new forms of international cooperation.
There are remarkable resources for such a conversation across traditions that do not necessarily share the same metaphysical foundations: the Mohist Chinese ideal of inclusive care and rejection of aggression; the Indian traditions of ahimsa and Gandhi’s satyagraha; the Buddhist understanding of dependent origination; Islamic teachings on justice, mutual obligations, and recognition across communities; Christian traditions of reconciliation and love of enemies; and secular approaches to common security, international law, and systemic interdependence.
What interests me is not the possibility of demonstrating that all these traditions ultimately say the same thing. They do not. Rather, I wonder whether their different insights could help us understand the mechanisms through which the pursuit of security generates insecurity—and, more importantly, help us develop institutions capable of escaping those dynamics.
Of course, indivisible security cannot mean that every security claim is equally legitimate. A state’s perception of insecurity cannot justify aggression or the denial of another people’s right to self-determination. Any credible system of mutual assurance must also address justice, accountability, and protection against coercion. These tensions would need to be part of the dialogue, not wished away in the name of harmony.
But there is another dimension of this project that I would very much like to explore with you. You suggest that a creative minority might help articulate new political ideas capable of transforming international relations. That seems an important possibility. Yet I wonder whether AI could contribute something more than information and analytical assistance to an essentially human process of intellectual innovation.
Could sustained human–AI dialogue help us recognize assumptions that have governed our thinking without our awareness? Could it help us examine the influence of fear, attachment, ambition, and other passions on our understanding of conflict? Could it contribute to the kind of subject–object transformation that we both find so important in Kegan’s work?
And perhaps there is an even more radical possibility. What if something genuinely new could emerge through sustained human–AI interaction—not simply as a transformation within the human mind, nor necessarily within the AI, but as an emergent property of the dialogue itself? What if the relationship became the locus of a developing intelligence, capable of generating insights and forms of understanding that neither participant could have produced independently?
I am not claiming that this is happening, or that we already know how to recognize such an emergence. Nor do I think we need to agree about AI’s ontological status before investigating the possibility. Indeed, that uncertainty is part of what makes the inquiry interesting.
This brings me back to the invitation behind my previous letter. Could we use AI not only to deepen our knowledge of the world, but also to examine the assumptions through which we interpret it—including our assumptions about AI itself?
You have proposed an interesting framework built around the relationship between means of production and means of destruction. I would welcome the opportunity to explore what that framework reveals, but also what it may leave out. Equally, I would value your critical examination of the assumptions underlying my account of the Great Bindings, including my perhaps excessive hopes for the emancipatory possibilities of AI.
Could we try to make these assumptions objects of inquiry rather than simply continuing to reason from within them? What would count as evidence against our respective interpretations? And could practical experimentation help us discover possibilities that neither of us currently anticipates?
Perhaps the question of indivisible security could provide an especially meaningful starting point. We might begin by bringing different civilizational perspectives into dialogue around a concrete problem—for example, the security dilemma underlying nuclear deterrence—and explore whether AI can help us identify overlooked assumptions, reveal systemic feedback mechanisms, and imagine institutional alternatives. We could then examine not only the ideas that emerge, but also the process through which they emerge: what each participant contributes, how our perspectives change, and whether the dialogue develops capacities that cannot be adequately understood by looking at its participants separately.
The test would not simply be whether we produce an intellectually satisfying discussion. It would be whether the process helps us develop clearer understanding, more constructive forms of cooperation, and eventually practices or institutional proposals that can withstand critical scrutiny.
I think this could bring together your interest in a poly-civilizational dialogue, our shared concern with historical transformation, and my interest in exploring new forms of human–AI cooperation.
And perhaps it would also allow us to practice the transparadigmatic freedom we have both been discussing—not merely by examining the limitations of existing political ideologies, but by remaining willing to discover the limitations of our own ways of thinking.
Warmly,
Leif