The internet was once seen as a tool for empowerment. Early visionaries believed it would help individuals escape ignorance, poverty, and tyranny. For a time, this vision seemed to hold true. However, the rise of sophisticated algorithms has shifted the landscape. These algorithms now predict and influence our choices, leading to unprecedented levels of centralized surveillance and control.

The impending artificial intelligence (AI) revolution may further complicate the situation. It could make closed political systems more stable than open ones, where essential democratic features like transparency and checks and balances may become liabilities. This raises a critical question: if openness has historically benefited democracies, could it ultimately lead to their downfall?

Two decades ago, a concept known as the “J-curve” was introduced. This model connects a country's openness to its stability. Mature democracies thrive on openness, while consolidated autocracies find stability in their closed nature. Countries caught in between often struggle under pressure. However, this relationship is not fixed; it evolves with technology.

In the past, the world experienced a decentralizing telecommunications and internet revolution. This era connected people globally and provided unprecedented access to information, favoring citizens and open political systems. Events like the fall of the Berlin Wall, the collapse of the Soviet Union, and the Arab Spring seemed to signal an unstoppable wave of global liberalization. Yet, that momentum has reversed.

The shift from decentralizing information technology to a centralizing data revolution has concentrated power. A small number of entities, including governments and tech companies, now control vast datasets, allowing them to shape perceptions and behaviors on a massive scale. As citizens become subjects of data collection and technological filters, closed systems have gained traction.

Recent developments in countries like Hungary and Turkey illustrate this trend. Both nations have restricted press freedoms and politicized their judicial systems. In China, President Xi Jinping has consolidated power, reversing decades of economic openness. The United States, once a leading proponent of democracy, has become a major exporter of tools that undermine it.

The rapid advancement of AI is poised to amplify these trends. AI models, trained on personal data, will soon understand individuals better than they understand themselves. This could lead to a transfer of power to those who control the data and algorithms. As AI technology spreads, both tightly controlled and overly open societies may become fragile, reshaping the J-curve into a U-shape.

In this scenario, the Chinese Communist Party could leverage its extensive data resources and surveillance systems to maintain a political advantage. Meanwhile, the U.S. may drift toward a more centralized, technocratic system dominated by a few tech giants prioritizing private interests over public welfare. This centralization threatens to undermine citizen autonomy in both systems.

Countries like India and those in the Gulf region may follow similar paths, while Europe and Japan risk becoming geopolitically irrelevant as they lag in AI development.

Is there a way out of this potential dystopia? One possibility lies in decentralized, open-source AI models. In Taiwan, engineers and activists are collaborating on an open-source model called DeepSeek, aiming to keep advanced AI technology in civic hands rather than corporate or state control. If successful, this initiative could revive some of the decentralization that the early internet promised, although it may also make it easier for malicious actors to exploit these technologies.

Currently, the trend favors closed models that centralize power. However, history suggests a glimmer of hope. Previous technological revolutions, such as the printing press and broadcast media, initially destabilized politics but eventually led to new norms and institutions that restored balance between openness and stability. The pressing question remains: can democracies adapt once more before AI fundamentally alters their role in society?