At this year’s World Economic Forum in Davos, Yuval Noah Harari cautioned that Artificial Intelligence is no longer a tool. It is emerging as an agent capable of generating language, shaping narratives, and potentially out-competing humans in every domain built on words. Law, religion, finance and politics, he argued, are thus vulnerable.
The observation was structural. If AI can produce persuasive language at scale, what happens to human agency in societies organised through speech, debate and interpretation?
Human civilisation rests on shared narratives. Religion depends on scripture and commentary. Law depends on interpretation. Democracy depends on debate. Markets depend on contracts and persuasion. Nationalism depends on story. Parliamentary arguments, judicial verdicts, campaign speeches, policy documents and news reportage rely on linguistic construction. If AI systems increasingly generate these constructions, then control over AI systems becomes control over discourse itself.
Thus the concern is that a small cluster of corporations and States controlling computational infrastructure can produce persuasive language at industrial scale. When AI drafts campaign messaging, filters social media feeds, generates news summaries or tailors political narratives to micro-demographics, democratic participation is subtly reconfigured. Citizens may believe they are responding freely. Yet the architecture shaping their perception may already be optimised elsewhere.
The danger is not machine rebellion. It is linguistic centralisation. Large language models are trained on vast datasets that remain opaque to public scrutiny. Their design, optimisation and moderation are governed by corporate priorities and geopolitical considerations.
India depends significantly on foreign AI infrastructure. This dependency is not neutral. In a country marked by intense electoral competition, linguistic diversity and rapid digital penetration, AI-driven language tools can expand access to knowledge and bridge linguistic divides. But they can also automate propaganda, amplify misinformation, and manufacture emotional resonance at unprecedented speed.
The Digital Personal Data Protection Act addresses privacy concerns but does not meaningfully regulate algorithmic influence over political speech. Electoral safeguards have yet to fully account for synthetic text, automated persuasion and AI-generated public commentary. Institutions such as the Election Commission of India and media regulators are only beginning to confront the scale of transformation. Without transparency mandates, audit frameworks and public oversight, AI-generated language risks becoming an invisible layer beneath democratic choice.
What distinguishes this technological shift from earlier industrial revolutions is its intimacy. Earlier, transformations mechanised labour. This one erodes cognition. The inner voice, once considered sovereign territory, now operates within machine-mediated environments. For centuries, we equated language with selfhood. To speak was to think. To write was to reason. But if language can be generated externally and reinserted into our cognitive space, the boundary between persuasion and manipulation grows thin. When that boundary thins, democratic citizenship weakens.
More than two thousand years ago, Laozi’s book, Tao Te Ching, observed that the truth that can be expressed in words is not the ultimate truth. In the age of generative AI, that insight acquires civic urgency. The old world in which humans held monopoly over language is ending. The new world of machine-mediated discourse has arrived without full democratic negotiation.
Whether this transformation strengthens democratic participation or hollows it out depends on institutional courage, regulatory clarity, public literacy and the willingness to treat AI as political infrastructure.
Ashutosh Kumar Thakur writes on society, literature, arts and environment