Listen to an AI audio recording of this newsletter

I recently tried to boost a clip on Instagram from my ​Outthinker podcast with Eric Ries​.

Meta rejected it as political content.

The apparent problem? Eric was questioning existing securities laws.

Think about what happened there. An algorithm interpreted an idea, classified it according to rules established by itself (well, by the private company that designed it), and decided whether I could amplify it.

No discussion. No consideration of the broader argument. Just a decision.

The timing was striking because ​Hamilton Mann​, another recent guest I recently interviewed for an upcoming podcast, had warned me about exactly this.

When AI becomes the authority

Hamilton argues that we are letting AI take too much power, allowing algorithms to increasingly become judge, jury, and executioner.

We learned centuries ago that functioning systems require a separation of powers, with checks and balances that prevent any one authority from becoming judge, jury, and executioner. That friction is intentional: it forces debate, gives different parties a voice, and creates the conditions for better decisions.

My experience with Meta was a relatively small example, but it made the risk tangible.

The algorithm judged the content. It interpreted the rules. Then it executed the decision.

As Hamilton put it during our conversation: “No need to inform [you] about what is going on. No need to have [your] permission, no need to get into a conversation about it. This is automatic.”

And that matters because more and more of what we see, hear, share, buy, approve, and reject is being filtered through automated systems. These systems are not simply organizing information anymore. They are increasingly making judgments about what is acceptable, relevant, risky, or permissible. And they are executing those judgements without our input.

The question is no longer simply whether AI can make decisions.

It is whether we are comfortable with the authority we are ceding to it, particularly when the rules it is enforcing may themselves deserve to be questioned.

Playing the game

This also helped me see a deeper connection between Eric and Hamilton.

Most business experts teach you how to play the game.

Learn the rules. Understand what drives success. Execute better than everyone else.

That is level-one thinking. It is essential. If you do not understand the game you are playing, you are unlikely to win it.

Level two is discovering the game. You test the boundaries, question assumptions, and uncover possibilities others have missed.

The ​Fosbury Flop​ is a classic example. For decades, high jumpers largely competed by improving variations of existing techniques. ​Dick Fosbury​ approached the bar backward, something that initially looked awkward and even absurd. But he had recognized something others had missed: the rules determined how high the bar was, not how an athlete had to cross it.

He did not perfect the technique of high jumping. He discovered more possibility inside the rules.

That is an important distinction.

Level-two thinkers look at the same playing field everyone else sees and find room to maneuver that others have overlooked. Much of my own work lives here too: borrowing a framework from another domain, reframing a problem, or finding a path competitors have not considered.

But Eric and Hamilton are pushing into different territory.

Changing the game

Level three is changing the game itself.

Instead of asking, “What else is possible within these rules?” you ask, “Should these be the rules at all?”

Eric is questioning structures within capitalism that we often treat as fixed. In our conversation, we talked about shareholder primacy, the now-dominant idea that a company exists primarily to enrich its shareholders. Eric’s point is that this “rule” is far more recent and fragile than we tend to assume. It became embedded in how companies operate without a single vote ever being cast to establish it.

Hamilton is asking a parallel question about algorithms. As he told me, “Without knowing it, you are delegating part of the legislative, the executive and the judicial power to the algorithm that you use.” He argues we need to change the rules by which AI is evolving.

Both are asking us to examine systems that have become so familiar that we stop seeing them as choices.

That is level-three thinking.

Those are fundamentally different questions from how to compete more effectively.

And they are much more threatening to established systems.

Organizations are designed to operate within known boundaries. Compliance teams determine whether something follows the rules. Algorithms are trained to identify whether something fits inside predefined categories. Institutions become very good at preserving consistency.

That is usually valuable.

But it also creates a natural tension when someone challenges the assumptions behind the system itself.

Compliance asks: Are we following the rules?

Innovation often asks: How can we win differently within them?

But level-three thinking asks: What if the rules themselves are the problem?

That is what makes voices like Eric’s and Hamilton’s so important. They are not simply offering a better way to operate inside the current system. They are questioning the assumptions underneath it.

Which brings me back to Meta.

The strange irony is that Eric was raising a level-three question about whether existing rules should change. Before that argument could travel further, it encountered a system designed to enforce the existing rules.

We are entering an era in which algorithms will increasingly interpret and enforce the boundaries around us.

That makes it even more important that humans retain the ability to question those boundaries in the first place.

Join ​Outthinker​ today to connect with leaders who are questioning the rules, challenging assumptions, and helping shape what comes next.