From the recently concluded US midterm elections and Israel’s parliamentary election to Taiwan’s upcoming nine-in-one local elections, artificial intelligence (AI) is rapidly making inroads into election campaigns. Some academics have even called this the first year of AI electioneering.

Unlike a year or two ago, when the technology was not yet mature enough, AI is now being used at various stages of campaigning. AI-generated political advertisements, deepfake videos and virtual spokespersons have all emerged. Campaign teams are using AI to analyse public sentiment and reach voters, while more voters are asking AI chatbots about policy positions.

It is hardly surprising that AI’s influence has spread from technology and business to politics. The shift in election campaigning from traditional media to digital channels, then social media and AI, seems merely a natural progression of technology. Meanwhile, incidents involving AI deepfakes impersonating leaders have become so common in Singapore and around the world that the many warnings to stay alert and avoid being deceived scarcely need repeating. So too has the familiar refrain that AI, as an efficient production tool, will amplify existing flaws in election campaigns, requiring governments and platforms to strengthen oversight and the public to remain vigilant.

Where, then, do the deeper changes that AI brings to election campaigns come forth? As more voters grow accustomed to using AI to obtain political information, the appeal is clear: it is fast, appears objective and neutral, and seems capable of tailoring policy analysis to individual needs. Yet this readily available convenience comes with three hidden costs.

First, even setting aside the risks of believing misinformation and AI hallucinations, many voters are likely still unfamiliar with AI prompts and ways of framing questions. With these rather “savvy” AI chatbots, a slight difference in a question can produce a vastly different answer. Responses tailored to you, when prompted by leading questions, also risk creating a personalised information bubble. How many voters hoping to use AI as a search engine are confident that they can question it critically and use it effectively?

Second, AI’s supposed neutrality may lie in its lack of overt affiliation with any political party. But the philosophy it subscribes to, the values it follows and the logic it applies all remain hidden in a black box. Tests of several chatbots by The Washington Post this year found differences in how models presented left- and right-wing views and supplied supporting arguments.

The AI industry touts “AI alignment” as evidence of its high ethical standards. In everyday life or business practices, there may be a set of universal principles to draw on. But in politics, which country’s or party’s values should AI be “aligned” with? Or is it merely being “aligned” with the goals of a particular AI company? If the price voters must pay is to hone their ability to think with AI, and platforms must examine and make transparent the underlying logic of AI alignment, political parties must confront the disruptive changes AI brings to the campaigning landscape.

As more voters turn to AI for information, content will no longer be written only for people, but also for AI. Parties will consider not just how to persuade voters, but also how to win favour with AI, making their positions easier to retrieve and cite as they compete to get their messages across through it. As party campaigning shifts from winning hearts and minds to catering to algorithms, using sheer volumes of information and authoritative sources to build public momentum, elections risk degenerating from a contest of ideas about the public good into a competition for informational dominance. As campaigns drift further from their original aim of persuasion through reason, voters will ultimately bear the cost.

A ballot paper weighs next to nothing, yet carries immense significance. When a democratic system gives people the right to elect their leaders, every ballot paper embodies trust in voters’ freedom of thought, independence of mind and concern for their country. Perhaps only a combination of individual critical thinking, platform oversight and political parties’ self-scrutiny can keep reason on course amid the AI wave. A ballot paper becomes all the more precious when it is not cast simply to follow the crowd.

(The author is a multimedia editor at Lianhe Zaobao.)