The obligations of power

Could you briefly talk about your path, and emphasize two or three pivotal moments that brought you to where you are today?

Rangan: I grew up happily in India receiving much love and learning from my parents, siblings, and teachers. By the time I was a teen, one theme spoke to me: The desire to serve.

I trained and came close to joining the military in the hopes of serving the nation that way. At the very same time an opportunity came up to travel to and stay in Canada on an exchange program for several months. For a variety of reasons, I was encouraged to and took up that opportunity. The exchange experience with a group of peers was transformative and sparked a curiosity about the larger world outside.

The second pivot was my move to the United States as a graduate student. I studied computer science at a time when computers were still a new frontier, their use moving beyond specific technical applications to offices around the world. Programming gifted me an invaluable lesson: humility. Most of the time when the program does not work, it’s a programming error. Things are intricate and interconnected and to make an operating system work as intended is a demanding and delicate task. You must work logically and carefully, and inspect your work, every time. To work properly, software, like life itself, requires frequent updates.

The third pivot was my PhD studies at Harvard. My curiosity had remained and grown and I was drawn to the idea of taking a few years to try and understand how the social, economic, and political world evolved and worked. I found myself drawn to questions about the history and workings of prosperity, policy, and power all in the international context. I had great mentors and advisors. My adult self is a reflection of that education and those teachers.

After completing my PhD in Political Economy, I joined INSEAD and have been studying, learning, and teaching here for thirty years now.

We are very good at evaluating leaders on the metrics that were visible at the time, and not great at passing judgment calls on the obligations they had but didn't see.

Merkel spent 16 years at the helm of German politics and was celebrated around the globe as one of the most positive figures. And yet, she called the internet an uncharted territory in 2013 – in a year when Apple launched its fifth model of iPhone. She decided to shut down nuclear power plants, cutting off German options for a diversified and resilient energy policy.

Clinton’s presidency celebrated the US as the only superpower, and yet his administration might not have seen how their policy towards the collapsing Soviet Union and Yugoslavia planted seeds of conflicts the world today struggles to resolve.

What is the right framework for distinguishing leadership that looked great in its moment from leadership that actually holds across time?

Rangan: You put your finger on a certain deficiency in how we teach and evaluate leadership. We teach and evaluate it as performance — did the leader achieve stated objectives? Did they do their job? But what about their role? Given the power and discretion leaders have, their moral role also matters. The logic and standards are different.

A pertinent insight is articulated by Amartya Sen, who explains that moral obligations arise from three distinct sources. A conventional source is from special relationships such as friend or parent. A second source of obligation arises from the need to compensate for harm that one may have caused another. Then there is a third and less conventional source: viz., the “obligations of power.” In this logic, can implies ought. The fact that a leader may have the capacity to help creates a sort of obligation to do so.

Clinton and Merkel did well on the grounds of performance—both achieved most of what they set out to do. But historians may ask whether these leaders might have managed the great power of the West (at that time) with greater magnanimity.

Dominique Moïsi argues in The Geopolitics of Emotion: How Cultures of Fear, Humiliation, and Hope are Reshaping the World (2008/2009) that international politics can be interpreted through three dominant emotions: hope, humiliation, and fear. The triumph of the West (especially economically) could have created an annuity of peace and prosperity more widely.

I summarize outcomes this way: O = P*I. Outcomes (O) are the product of Power (P) and Interest (I).

If great power is coupled only to self-interest, then moral obligations to others are limited. But if great power integrates a broader interest then we can experience such outcomes as a lasting peace and wide prosperity. In the economy, this means orienting the power of business not only to profit but also to people and planet.

When we recruit leaders, we are always confronted with the gap between knowing what good leadership might look like and actually identifying it before it is given the reins of power. Self-awareness and humility are differentiating variables at the elite level — not IQ, not technical expertise, not even track record, because track record is often produced in contexts that may not fully transfer.

But here is my frustration: humility is also one of the most readily performed qualities in an interview setting. Every board candidate will tell you they know their blind spots. How do you actually interview for these qualities, especially at the board level?

Rangan: Humility and caring are indeed hard to assess.

I could imagine using 3 questions to assess this, and I want to explain the logic behind them, because the logic matters more than the questions themselves.

The first is: What concerns do you have regarding the broader society we live in? Not in your industry or your company. In society. Most candidates will answer this question by almost immediately redirecting back to their sector. That redirection speaks louder than many words. It tells you that the person's moral imagination is essentially bounded by their formal job. The people who give insightful answers are ones that appraise the world around us with welfare concerns of a broader nature. They appreciate and articulate interdependence. As the cliché says: it’s challenging to maintain a successful firm in a failing society.

The second question is: What brought you joy in the last three to five years? Joy is a much harder thing to fake than humility. Accomplishment causes pride; only contribution causes joy. Leaders who focused on their jobs, their performance goals, will have no broader references. They will not have the stories or language to speak about progress.

The third question is: What do you see as the performance goals and the progress goals of this organization? Both are no doubt important and yet the distinction matters. Leaders with competence will of course be able to articulate clearly on the former. Only leaders with character will also be able to articulate meaningfully on the latter.

I want to be honest about a structural challenge: boards seem to disappoint both in terms of composition and process. Board composition still reflects networks of social similarity. As a process, board governance reflects formality and ceremony more than candor and challenge (until it’s too late). It’s not obvious how to address this. But there must be better ways. We need more research on this.

I have been sitting with an observation for the last two years, and I want to test it with you.

AI, as a cultural phenomenon, has all the structural properties of a major religion. It has doctrine — the belief that intelligence is substrate-independent and that greater intelligence is always better. It has apostles — Altman, Pichai, Zuckerberg — who travel the world delivering a gospel of transformation. It has heretics, like Gary Marcus, who question whether there is a mathematical basis behind some of the most powerful claims on reasoning models, scaling self-driving cars, and conscious machines. And it has prophets of doom, like Eliezer Yudkowsky.

My question is not whether this characterization is fair, but whether any time soon we will consider the AI moment through the lens of normality. AI is an important consequential technology, with clear benefits and risks, still at the very beginning of its evolution.

Rangan: Religions can be an important platform for creating and reproducing social order. Faith can help when we encounter phenomena that exceed our comprehension— something that reorganizes the boundary between what is understandable and what we may take as divine or pre-ordained. Machine intelligence may have some of that magical quality.

Religions provide helpful ways for integrating faith with sentiments of lack of control and accompanying anxiety.

At least at this moment, it seems that AI is developing more rapidly than the moral infrastructure that such power may require to govern its use.

Traditional assets of capitalist systems operate with frameworks of natural, human, physical, financial, and intellectual capital. What we are creating with AI is what I call cyber capital — an unprecedented and self-reinforcing cognitive leverage.

Cyber capital is not good or bad in itself. Fire is not good or bad. Fire used by a mature civilization is beneficial. Fire used without any understanding of consequence results in destruction. The AI companies are currently building modern fire. The moral capital — the individual and institutional capacity to use it wisely — seems, worryingly, to be lagging.

This is why I insist on the distinction between AI and what some refer to as BI: beneficial intelligence. The question is not whether we can build systems more capable than humans at specific tasks. The question is whether their deployment produces outcomes that are genuinely better for human beings in a substantive sense — not just more efficient, but better in a way that would survive moral scrutiny.

The J-curve is my frame for the return to normality you are asking about.

Recovery comes not when the technology matures, but when the society – with its institutions, norms and educational frameworks - catches up. We are in the early trough. The path through is moral and institutional, not only technological.

I can offer an analogy to intuit what may happen with AI.

The obesity crisis was not caused by food. It was caused by the industrialization of food supply without a corresponding education of how individuals and communities relate to nutrition. We have created extraordinary abundance and made it extremely easy to consume in ways that are harmful.

AI could do something similar to intelligence. Democratization of intelligence is something beautiful. Immature and immoderate consumption of intelligence could be an entirely different thing.

We are witnessing first attempts to regulate AI. Regulation, whatever its limitations, operates at scale. It is blunt, but it lands. Moral education of leaders — which is essentially what you are advocating — is slow and difficult to assess.

When it comes to AI, can we educate our way out of a structural incentive problem? The people deploying AI at scale are responding to financial incentives and competitive pressure, not to a seminar on humanities.

Rangan: Regulation can limit the scope of action. What regulation cannot do is shape interests. And the outcome formula makes clear that interest is not a residual variable.

If you have power and only self-interest in the consequences, you will find ways around any regulatory constraint, because you are applying your legal imagination, and legal imagination is often more creative than regulation.

Education ought to be the formation of broad interests. Philosophy draws a distinction that I think is fundamental: between prudential reason — the capacity to identify and deploy the most efficient means to a given end — and moral reason, which is the capacity to interrogate the ends and intended beneficiaries. Almost all leadership education is in prudential reason. We almost never teach leaders to ask whether the end they are pursuing is the right one, and if so, for whom, and by what standard.

There are perhaps fifteen thousand people in the world who sit on the boards of institutions determining the direction of AI deployment. If you could shift the moral orientations of that population — not their stated values, which are already excellent, but their actual interests, the things they genuinely care about when no one is watching — you would change the trajectory of the consequences of the technology.

The deeper shift I am interested in is the move from consumers to creators. The printing press eventually created not just readers but writers, editors, publishers — people who had a creative relationship with the medium. The cure for unhealthy consumption is not prohibition. It is the cultivation of a different relationship with the substance — one that involves understanding, craft, and genuine choice.