When Intelligence Stops Being the Problem

For most of modern history, we assumed that better intelligence would lead to better decisions.

If governments had more data, if experts had better models, if forecasts were more accurate, then policy would improve and outcomes would follow. Entire institutions were built around this belief. Expertise justified authority. Analysis conferred legitimacy.

That assumption is now under strain.

Anthropic’s CEO, Dario Amodei, just published a new essay and “The Adolescence of Technology” and it lands on an uncomfortable truth most AI debates are avoiding.

The real risk isn’t that AI becomes powerful. It will, that is a given. It’s that our institutions and societies aren’t mature enough to handle that power.

Intelligence is scaling faster than governance. Capability is outpacing responsibility.

We are entering a world in which intelligence is no longer scarce. Analysis, synthesis, translation and prediction are becoming cheap, fast and increasingly automated. Artificial intelligence systems can already generate convincing arguments for almost any position, often faster and more comprehensively than humans. As Dario Amodei eloquently outlines, this comes with significant risks for the future of humanity.

And yet the quality of our decisions is not improving in proportion. In some cases, it is deteriorating.

This points to a deeper problem.

The New Scarcity

When intelligence becomes abundant, it stops being the bottleneck. The new scarcity is judgment, and the ability and willingness to decide under extreme uncertainty.

Not the ability to generate options, but the willingness to choose between them. Not confidence, but responsibility. Not prediction, but commitment.

AI systems can now produce thousands of plausible futures, each supported by data and probability estimates. What they cannot do is decide which future deserves action or accept responsibility when that action leads to disappointment or harm.

That burden does not disappear as intelligence improves. It intensifies.

Power Outrunning Maturity

What we are witnessing is not simply technological acceleration, but a growing mismatch between capability and institutional maturity.

Our technologies are advancing faster than our political systems, governance frameworks and cultural norms can adapt. Intelligence is scaling rapidly; the structures meant to direct it are not.

Isaac Asimov already foresaw this in 1988, when he stated: “The saddest aspect of life right now is that science gathers knowledge faster than society gathers wisdom.” It is this quote that drove me to write my latest book Now What? How to Ride the Tsunami of Change.

This gap creates a dangerous illusion: that better analysis alone will resolve hard choices. In reality, it often does the opposite.

Why More Intelligence Can Paralyse

As analytical capacity grows, every option becomes defensible. Every course of action can be justified by data. Every delay can be rationalised. Every failure can be explained as reasonable given the information available at the time.

The result is a new kind of paralysis. Nothing is obviously wrong. Nothing is clearly right. Everything is arguable.

In this environment, institutions do not fail because they lack information. They fail because no one can clearly justify acting.

This helps explain why public trust is eroding even as access to information expands. The problem is not ignorance. It is the absence of accountable decision-making.

The Limits of Optimisation

Much of today’s debate assumes that better optimisation leads to better outcomes. But optimisation is not judgment.

Optimisation selects the best option given a defined objective. Judgment determines which objectives matter, which trade-offs are acceptable, and which risks society is willing to bear.

No model can resolve value conflicts. No system can encode responsibility. No algorithm can be praised, blamed or voted out of office.

When optimisation replaces judgment, decision-making may look rigorous, but accountability quietly disappears.

The Danger of Technical “Alignment”

Current discussions about AI safety often focus on aligning systems with human values. This assumes that values are stable, coherent and easily specified.

In reality, values conflict. Priorities shift. Trade-offs are unavoidable.

When alignment is treated as a technical solution, it risks becoming a way to avoid political and moral choice rather than confront it. The more “aligned” a system appears, the easier it becomes for humans to step back and say: the system recommended it.

That is not safety. It is moral outsourcing.

Authority After Intelligence

Historically, authority flowed from superior knowledge. In a world where high-quality analysis is widely available, that foundation weakens.

Authority must be earned differently.

It must come from the ability to prioritise amid abundance, to exclude plausible alternatives, to act without guarantees, and to accept consequences openly.

This is why confident answers increasingly feel hollow. Confidence is no longer anchored in scarcity.

What Remains Human

In a world saturated with intelligence, the most important role is not knowing more.

It is being willing to say: this matters more than that. We will act here and not there. We will accept these risks, but not those. We accept the risk of being wrong.

The future will not be shaped by the systems that predict best, but by the people and institutions willing to take responsibility for risks or decisions that cannot be proven real or correct in advance.

When intelligence is no longer the problem, responsibility becomes the work.

And responsibility cannot be delegated to machines.