The rise of Algorithmic Power: a conversation with Hamilton Mann
As artificial intelligence plays an increasingly influential role in shaping decisions across society, Hamilton Mann (EDHEC EMBA 2010) argues that it is time to look beyond questions of performance, safety and transparency, and examine the power we are handing over to algorithms.
An AI researcher, Executive Chairman and President of the Artificial Integrity Institute, and Visiting Scholar at the Wharton Neuroscience Initiative at the University of Pennsylvania, Hamilton Mann has just published Trias Algorithmica: What Code Rules (Wiley, 2026). In his new book, he explores what makes algorithmic power legitimate and how it should be constrained as its influence grows.
To mark the book’s release, Hamilton Mann agreed to answer our questions about the rise of algorithmic power, the risks of its concentration, and the new forms of checks and balances that may be needed in the age of AI.
- Why did you feel it was important to speak out now, through this book, about the power we are increasingly entrusting to algorithms?
I think we have reached a point where some advanced algorithmic systems are no longer just tools that help us carry out our intentions. They are increasingly becoming part of the architecture through which life-altering decisions are made, opportunities are allocated, information is filtered, risks are assessed, and human choices are shaped.
Algorithmic power has become what I call the sixth power.
As we learn to coexist with increasingly intelligent forms of artificial agency, one of the major challenges of our age is to establish the foundations that will guide how algorithmic power should be governed, at a pace that allows our institutions to keep up with the expansion of that power.
What we see, for example, is a hiring system ranking candidates, a social network determining what becomes visible, a bank flagging a transaction. Each of these decisions can appear operational and even technical in isolation, but taken together they amount to something far more consequential. They mark the transfer of power to algorithmic systems.
Not only is this new form of power exercised across society often privately constituted, but these systems also shape rules that function like law, enforce them through the infrastructures and processes they control, and, to a large extent, evaluate their own outcomes as well as the consequences for those subject to those algorithmic rules, recombining the concentration of powers that constitutional democracies have historically sought to keep separate.
And this power is accumulating almost invisibly, because it arrives as a service.
The book addresses this challenge through a framework for the separation of algorithmic powers, asking how this power should be divided and constrained so that it does not become excessively concentrated and always meets counterpower.
- What would you like readers to take away, or perhaps question, after finishing Trias Algorithmica?
What I would first like readers to take away is the category change of the question we ask about AI. We have spent a great deal of time asking whether algorithmic systems are intelligent enough, safe enough, fair enough, transparent enough, or trustworthy enough. And those questions still matter. But as the capabilities of these systems continue to grow at an accelerating rate, driven in part by the so-called AI race, it is now time to ask a question one level above: what makes algorithmic power legitimate? Asking ourselves not just usage-related questions, but constitutional ones such as who should be entitled to exercise algorithmic power in society, under what conditions, what must prevent that power from becoming self-authorizing, self-executing, and self-judging, and what forms of counterpower must exist to constrain it.
I would like readers to become more attentive to the concentration of algorithmic power. Not to the increasingly visible concentration of resources, capabilities, and global reach in the hands of a small number of actors, which I think is not impossible to ignore, but above all to the very nature of the algorithmic power itself.
Concentrated power should never be presumed benign simply because the intentions behind it are presumed to be. It merely needs enough scope, enough confidence, and too few rivals to become dominating.
The modern technology industry has often proceeded from the opposite assumption: that concentration is acceptable so long as intentions are benevolent and innovation is rapid. But the more consequential a system becomes, the less persuasive speed is as a defense against scrutiny. A brake is useless only if one assumes that crashing is acceptable.
A powerful system should be trusted in direct proportion to how fully its power can be contested, divided, interpreted, and overruled. And power that cannot tolerate friction is not efficient in any ethically, morally and socially serious sense; it is simply unchecked.
If twentieth-century societies learned to build checks on concentrated industrial and bureaucratic power, the twenty-first must learn to build them around algorithmic systems that increasingly script the conditions of social life. Not because machines are evil. But because any architecture of power, once sufficiently integrated, begins to seek its own continuity.
- In your view, what is the one question about AI that leaders are not asking themselves enough today, but that will become impossible to ignore in the years ahead?
This question: What makes algorithmic power legitimate?
And this is the question that Trias Algorithmica seeks to answer.
Website : The Hamilton Mann Conversation
Discover the book : Trias Algorithmica
Who Is Hamilton Mann?
Hamilton Mann is an AI researcher and Executive Chairman and President of the Artificial Integrity Institute, a nonprofit organization dedicated to advancing AI systems that prioritize integrity over intelligence.
Introducing the concept of Artificial Integrity, Mann's bestselling book Artificial Integrity: The Paths to Leading AI Toward a Human-Centered Future (John Wiley & Sons, 2024) establishes a new AI paradigm grounded in what makes AI systems worthy of trust. It was selected as a Must-Read of 2024 and one of the Top 10 Essential AI Books for 2025 by the Next Big Idea Club, named one of the Top 10 New Management Books of 2025 by Thinkers50, and selected for Egon Zehnder's 2026 CEO Insights Annual Reading List.
In his latest book, Trias Algorithmica: What Code Rules (John Wiley & Sons, 2026), Mann extends this work from the integrity of AI systems to the legitimacy of algorithmic power, introducing Trias Algorithmica as a new governance paradigm for defining what makes algorithmic power legitimate. The book was selected as a Must-Read of 2026 by the Next Big Idea Club. It was described as “a fascinating assessment for the age of AI” by Simon Johnson, 2024 Nobel Laureate in Economic Sciences and former Chief Economist of the International Monetary Fund.
Mann is also a Visiting Scholar at the Wharton Neuroscience Initiative at the Wharton School, University of Pennsylvania. He lectures at INSEAD and HEC Paris, serves as a mentor at the MIT Priscilla King Gray (PKG) Center, and conducts doctoral research at École Nationale des Ponts et Chaussées – Institut Polytechnique de Paris.
In 2024, he was inducted into the Thinkers50 Radar as one of the 30 most prominent rising business thinkers globally, and in 2025 he received the Thinkers50 Distinguished Achievement Award for Digital Thinking. His work appears in Harvard Business Review, Stanford Social Innovation Review, California Management Review, Rotman Intelligence, and I by IMD, among others, and he regularly writes for Forbes US.
Comments0
Please log in to see or add a comment
Suggested Articles
