Daron Acemoglu’s recent reflections on liberal democracy and artificial intelligence have reopened one of the most important debates of our time: what happens when a technology capable of transforming work, politics, information and even human behaviour develops faster than the institutions that are supposed to govern it? His concern is justified. Yet I would reverse the usual order of causality.
Artificial intelligence is not killing liberalism. Liberalism was already in retreat. AI has simply arrived at a moment when liberal democracy, the free-market economy and the institutional order supporting both have been weakening for years. It may accelerate that process dramatically, but it did not initiate it.
Acemoglu’s broader argument is not, in fact, that artificial intelligence suddenly destroyed liberal democracy.
His concern is that liberal democracy was already facing a profound crisis caused by deindustrialisation, inequality, declining worker power, economic dislocation, and a growing sense among citizens that they have less influence over the decisions shaping their lives.
Artificial intelligence now enters this already fragile environment with the capacity to amplify many of these weaknesses.
On that diagnosis, we are not far apart. Where I differ is in the historical sequence.
I see the present deterioration as part of a much longer systemic problem, one that predates both generative AI and the current technological debate by many years.
I first raised these concerns in the aftermath of the 2008 global financial crisis. In a study published in 2010 titled Capitalism, Financial Liberalisation and the Monetary Economy: A Critique of the System, I argued that the growing separation of finance from the real economy, the excessive concentration of capital and the increasingly intimate relationship between governments and financial institutions were undermining some of the very assumptions on which the free-market system was supposed to rest.
Since then, those contradictions have become even more visible. The crisis of liberalism, in other words, did not begin with artificial intelligence. It became structurally apparent long before AI acquired its current relevance.
A hybrid structure
For decades, capitalism and liberalism were often treated as interchangeable concepts. They are not.
Capitalism is an economic system centred on private property, capital accumulation and profit.
Liberalism is a broader political and philosophical order based on individual liberty, equality before the law, freedom of contract, property rights and, above all, limits on the exercise of power. Capitalism can therefore survive without liberalism.
Indeed, this may be precisely what we are now witnessing. Markets may continue to exist, companies may continue to make profits, and capital may continue to accumulate while the liberal principles that once legitimised and constrained the system gradually weaken.
The contradiction becomes particularly evident when corporations demand free markets but seek protection whenever competition becomes uncomfortable, when banks celebrate market discipline but expect public rescue when their own decisions go wrong, and when governments continue to speak the language of market economics while intervening in prices, credit, exchange rates, energy, trade and industrial allocation.
We still have markets, but what is increasingly difficult to establish is how free those markets actually are
After the global financial crisis, central banks flooded markets with unprecedented liquidity.
Governments rescued institutions considered too important to fail. During the pandemic, fiscal transfers, credit guarantees, and emergency support programmes expanded on a massive scale.
Energy shocks led to subsidies and price controls. Geopolitical tensions brought tariffs, export restrictions, industrial policy and increasingly explicit attempts to determine which industries, technologies and corporations should succeed.
Some of these interventions were understandable, and some were probably unavoidable, but their cumulative effect should not be ignored.
We still have markets, but what is increasingly difficult to establish is how free those markets actually are.
This is why I have argued over the years that the problem was not simply that liberalism had become distorted, but that capitalism itself had moved away from liberalism.
A system in which capital increasingly seeks political protection rather than genuine competition may still be capitalist, but it is becoming progressively less liberal.
When economic power seeks proximity to political power, and when political power begins to select winners, determine access to credit, influence prices and redesign market conditions in favour of certain actors, the result is no longer classical liberal capitalism.
It is a hybrid structure in which private ownership survives, but competition is increasingly mediated by political influence.
Trust precedes money
The political transformation follows a similar logic. Liberal democracy rests on two traditions that are related but not identical.
Democracy determines who governs; liberalism determines what those who govern are permitted to do.
This distinction is fundamental. Winning an election does not confer unlimited authority.
A parliamentary majority does not gain the right to abolish individual freedoms, weaken the independence of the judiciary, disregard minority rights or redesign institutions solely to preserve its own power.
Democracy is based on majority rule; liberalism places boundaries around majority rule. A democracy that does not respect individual free will cannot remain liberal for very long.
This is why reducing democracy to elections is so dangerous. Elections are necessary, but they are not sufficient.
A political system can hold elections and still weaken the rule of law. It can count votes and simultaneously erode the separation of powers. It can invoke popular sovereignty while narrowing individual liberty.
The distinction between electoral democracy and electoral autocracy is therefore no longer merely academic.
The economic consequences cannot be separated from this political deterioration
It has become one of the defining political questions of our age. Liberal democracy cannot be reduced to the presence of a ballot box, because liberalism begins precisely where the will of the majority is constrained by the rights of the individual.
The economic consequences cannot be separated from this political deterioration.
Markets depend on trust long before they depend on money. Investors must trust that property rights will be respected.
Companies must trust that contracts will be enforced. Entrepreneurs must believe that competitors will not receive preferential treatment because of political connections.
Citizens must have some confidence that regulatory decisions will not change arbitrarily overnight.
Without these guarantees, markets may continue to function mechanically, but they cease to be genuinely liberal markets.
Economic liberalism cannot indefinitely survive the disappearance of political liberalism.
A form of governmentality
This brings us back to Michel Foucault. It would be intellectually careless to claim that Foucault simply predicted “the end of liberalism” in the 1970s.
He did not. Yet his 1978–79 lectures at the Collège de France, later published as The Birth of Biopolitics, remain remarkably useful for understanding what followed.
Foucault treated liberalism not merely as a doctrine demanding less state intervention, but as a form of governmentality: a way of organising power, shaping behaviour, and defining the relationship between the individual, the market, and the state.
One of the most important implications of his analysis was that neoliberalism did not necessarily entail the disappearance of the state.
Political power can concentrate economic power, while economic power can increasingly shape political decisions
The state could instead become actively involved in creating, maintaining, and extending market mechanisms.
This point looks increasingly relevant today because the modern state and the modern market are not simply adversaries.
They often shape one another. Governments regulate markets, while markets influence governments. Political power can concentrate economic power, while economic power can increasingly shape political decisions.
Surviving the AI age
Artificial intelligence is emerging in precisely this environment. This is why Acemoglu’s warnings deserve to be taken seriously.
AI is likely to reshape employment, productivity, education, information and decision-making on a scale that few technologies have achieved before.
Yet billions of people whose lives will be transformed by these systems have almost no meaningful say in how they are designed, who controls them, or what objectives they should serve.
This is a democratic problem before it is a technological one. If the principal direction of AI development is determined by a small number of technology companies, investors and executives, then the technology could reinforce existing concentrations of economic and political power.
If automation becomes primarily a method of replacing labour rather than enhancing human capability, inequality may deepen.
If algorithms increasingly determine which information citizens see, political autonomy itself becomes harder to define.
But none of this began with AI. Capital concentration began earlier. The decline of organised labour began earlier. The weakening of the middle class began earlier. The growing distance between finance and the real economy began earlier. The influence of large corporations over public policy began earlier.
The habit of treating citizens as democratic participants only once every few years, rather than as individuals whose freedom must be protected every day, began long ago.
Artificial intelligence did not create these tendencies. It gives them a remarkably powerful accelerator.
The central question, therefore, is not whether AI is inherently liberal or authoritarian, egalitarian or oligarchic.
Technologies do not operate in an institutional vacuum. In a society with strong competition, independent courts, accountable institutions, transparent regulation, and genuine protection of individual rights, artificial intelligence could increase productivity, expand human capabilities, and create considerable prosperity.
In an economy dominated by oligopolies, weak institutions, political polarisation and declining social mobility, exactly the same technology could concentrate power even further.
The difference will not be determined by the intelligence of the machines, but by the quality of the institutions governing them.
For this reason, I find the question “Will AI destroy liberalism?” slightly misleading. The more urgent question is whether the institutions capable of sustaining liberalism still possess the strength to survive the AI age.
Liberalism is not protected by technology. It is protected by the rule of law, institutional restraint, competition, accountability, and respect for individual free will. Remove those foundations and elections will still continue.
Stock exchanges can remain open. Companies can keep making profits. Governments can continue to speak the language of democracy and markets. Yet democracy may no longer be liberal, and markets may no longer be free.
That process did not begin with artificial intelligence; it started long before. AI may simply make it much harder to pretend it is not happening.