The future of AI is not something that will simply happen to us. It is the result of choices about who designs technology, who owns it, who controls it, and to whom they are accountable.
One of the most dangerous ways we talk about Artificial Intelligence today is as if it were a natural phenomenon. As if it were an approaching storm, a technological tide that cannot be stopped, and to which societies, workers, and democratic institutions must simply adapt.
This view is both mistaken and politically dangerous. AI does not come from nature. It is built. Its models are designed by people, trained on specific datasets, run on computing infrastructure financed by someone, owned by companies or public organizations, and deployed for purposes that someone has chosen.
Behind every AI system, therefore, are human decisions. And behind those decisions are economic interests, institutions, rules, and relations of power.
“Inevitable AI” Is a Political Narrative
Yuval Noah Harari’s observation that presenting the dominance of AI as “inevitable” can serve as a way of avoiding human responsibility goes directly to the heart of the problem.
When a development is described as inevitable, politics almost disappears from the discussion. If “this is simply the way things are going to be,” there seems to be little point in asking whether we want them to be that way, who benefits, who loses, or what alternatives we might choose.
Yet nothing inherently requires us to use AI to assess employees, select recipients of social benefits, conduct surveillance of citizens, or deliver personalized political persuasion. Nor is there any technological law requiring the most powerful models, datasets, and computing infrastructure to be concentrated in the hands of a small number of global corporations.
These are outcomes of choices.
That is why the right question is not “what will AI do to us?” It is “what kind of AI do we want to build, who will control it, and for what purposes will we allow it to be used?”
The Real Issue Is Power
Public debate often focuses on how “intelligent” AI systems will become. For democracy, however, the more important question is different: who gains power through them?
AI requires data, computing power, software, models, energy infrastructure, and access to digital platforms. When all of these are concentrated in the hands of the same small number of actors, technological power gradually becomes economic, informational, and political power.
A company that knows what we read, what we buy, and whom we communicate with, while at the same time operating systems capable of conducting personalized conversations with us, is not simply providing another digital service. It acquires an unprecedented ability to mediate information, knowledge, and ultimately the formation of preferences.
This is where the democratic dimension of AI lies.
Humans Must Retain the Final Say
The answer is not to reject technology. It is to define its boundaries democratically.
In public administration, AI can help a civil servant locate information, analyze data, or identify unusual patterns. But it should not become an impersonal final arbiter in decisions concerning a social benefit, a recruitment process, a fine, or any other matter affecting a person’s rights.
Citizens must know when AI has been used. They must be able to challenge the outcome and appeal to a human being who has the genuine authority to review and reverse it.
AI can be an adviser. Responsibility, however, must remain human and institutionally identifiable.
Democracy Also Needs Technological Alternatives
Democratic control cannot be secured through regulation alone. It also requires genuine freedom of choice.
If a state is completely dependent on a single supplier for its models, data, cloud infrastructure, and critical digital services, its capacity for political control is constrained. Digital sovereignty, therefore, should not mean that a country must build everything by itself. It should mean having the ability to control, inspect, migrate, modify, and, when necessary, replace the technologies on which it depends.
This is where open source software, open standards, data portability, and public digital infrastructure become especially important. Openness does not by itself guarantee democracy. But it creates the technical conditions for independent scrutiny, reuse, interoperability, and freedom from vendor lock-in.
From AI That Monitors Citizens to AI That Monitors Power
There is another possibility.
Instead of using AI primarily to enable governments or companies to know more about citizens, it can be used to enable citizens to know more about those who exercise power.
Open systems can analyze public procurement contracts, budgets, and government decisions. They can identify unusual spending patterns, explain complex administrative documents, and make public information genuinely accessible.
In this way, AI can become a technology of accountability, rather than merely a technology of administration.
Society Still Has a Choice
The fundamental political dilemma, therefore, is not whether we are “for” or “against” Artificial Intelligence.
The question is who controls it, for whose benefit, and under what rules.
AI is not a law of nature. It is a human creation and, precisely for that reason, it can be subject to democratic choice. Society has the right to decide where AI is used, what data it may process, who is accountable for it, which systems should be rejected, and which alternative infrastructures should be built.
The future of AI is not something that will simply happen to us. We are already building it. And responsibility for who will hold power in that future remains ours.
Sources:
The Economist, “An interview with Yuval Noah Harari | The Economist Insider”: Harari’s discussion with Zanny Minton Beddoes highlights the need to reject the idea that AI dominance is inevitable and places human responsibility back at the center of decisions concerning the development, deployment, and power of AI systems. The Economist Insider,
European Union, Artificial Intelligence Act, Regulation (EU) 2024/1689: The European regulatory framework provides for human oversight of high-risk AI systems and for humans to be able to disregard, override, or reverse the output of an AI system, institutionally reinforcing the principle that technology must not replace human responsibility. EUR-Lex, Regulation (EU) 2024/1689,
Council of Europe, Framework Convention on Artificial Intelligence and Human Rights, Democracy and the Rule of Law: The first international legally binding treaty on AI links the entire lifecycle of AI systems to human rights, democracy, the rule of law, transparency, and accountability. Council of Europe Framework Convention on AI,
UNESCO, Recommendation on the Ethics of Artificial Intelligence: UNESCO’s global Recommendation treats AI as a sociotechnical system whose development, design, and use must be governed by human rights, transparency, accountability, and human oversight, stressing that ultimate responsibility cannot be delegated to machines. UNESCO Recommendation on the Ethics of Artificial Intelligence.