by Tan Aik Keong (AK)
Winning an election today has stopped being about who puts up the most posters or shouts the loudest slogan. The real dividing line is who understands voters better, who reads sentiment faster, and who can get the right message to the right person at the right time. In other words, the point of AI entering a campaign isn't to "replace politicians" — it's to move campaigning from broad, generic messaging toward a new stage that's data-driven, sentiment-aware and able to respond in real time. That, at its core, is the whole idea: AI's goal isn't chasing reach — it's building voter confidence by delivering the right message, to the right audience, at the right time.
Many parties are still stuck in old thinking — that a campaign means posting constantly, holding frequent rallies, buying a few ads, and repeating the same track record a hundred times. The problem is that today's voter isn't standing in a public square waiting to be persuaded — they live inside an algorithm. When a campaign team has no first-party data, no audience segmentation, and no attribution on what's actually working, all they can do is run the same script for everyone — and the result is usually a lot of shouting that doesn't land. That's exactly the trap: flying blind on data, chasing vanity reach metrics, and having social media, ground teams and data teams all working in silos instead of one real voter intelligence platform.
Precision targeting
AI's first real value is precise, data-driven targeting. It can break "voters" down from a vague mass into distinct groups — young urban professionals, middle-class suburban families, first-time voters, undecided voters — and design different content for each, based on age, location, language preference, interests, engagement habits, even which issues get the strongest response. Using tools like Meta Custom Audiences alongside segmented data and behavioural signals to lift a targeted campaign score isn't about casting one wide net — it's knowing which group is sensitive to housing prices, which cares more about jobs, and which responds more to clean governance or local livelihood issues.
The US has already moved ahead on this front. The Associated Press has reported that the Biden campaign and Democratic candidates used AI to segment audiences, draft fundraising and mobilisation emails, analyse which messages performed better, and give grassroots volunteers more tailored talking points. What AI changes first isn't a candidate's ideas — it's how the whole campaign machine understands and reaches voters more efficiently.
Sentiment and narrative monitoring
AI's second value is sentiment analysis and real-time narrative monitoring. What does a campaign fear most? Not opposition — slowness. Plenty of parties don't lose because they lack a good argument; they lose because they only respond after a negative narrative has already spread online, turned into memes, clips and remixes, or even outright disinformation. By then, what people remember isn't your clarification — it's the other side's emotional framing. The key idea here is using large models and video/audio understanding tools to monitor clips, comments, sentiment scores and how fast something toxic is spreading, with the system then recommending the most appropriate next content and response strategy. That's not firefighting after the fact — it's early warning.
In other words, AI isn't just for generating images or drafting copy — it's for judgment: which issue is heating up? Which clip is starting to shift the narrative? Which emotion is most likely to ignite? A campaign team can then adjust tone, platform, audience and timing in real time. That "monitor → assess → respond" loop is the real core of campaigning going forward.
Reaching first-time and undecided voters
AI's third value is tailoring content for different voter groups — first-time voters especially deserve attention, because they often have no fixed party loyalty and rarely read long policy essays. What they do watch: short videos, what friends share, the emotional temperature of their feed. For this group, politics isn't a manifesto — it's a feeling of "does this actually affect me." Rightly, first-time and undecided voters belong at the centre of persona planning. What they need most isn't doctrine — it's to feel understood: student loans, wages, rent, transport, entrepreneurship, freedom, the future — that's the language that actually reaches them.
The battlefield itself has changed too. The old home turf was the ceramah, the newspaper column, the TV news segment. The real front line now is TikTok, Instagram, even Xiaohongshu (RedNote), and community WhatsApp chains. TikTok runs on algorithmic amplification, short content, instant feedback, influencer and grassroots collaboration, and meme-driven spread — and the prize is young voters' attention. Indonesian President Prabowo is a clear example, reshaping his public image through social media into a softer, more youth-friendly figure; in the US, both the Biden and Trump camps eventually moved onto TikTok too, because that's where young voters' attention already was.
Xiaohongshu isn't a traditional political platform, but it's becoming a hub for lifestyle content, identity and topic spread within Chinese-speaking communities. Reuters and AP both reported that when TikTok faced a possible ban in the US in 2025, a wave of users flooded into Xiaohongshu, briefly pushing it to the top of the US download charts. That shows how fast attention can migrate between platforms — and political narrative follows wherever attention goes. For a campaign team, that means you can't put all your resources into one platform like Facebook anymore; you need to understand each platform's own language: fast on TikTok, polished on Instagram, lifestyle-driven on Xiaohongshu, localised on WhatsApp.
A tool, not a licence to manipulate
That said, we need to be clear: AI can help a campaign, but that doesn't mean it should be used to manipulate voters. Recent experience in the US and across Asia is a reminder that beyond audience analysis, copywriting and sentiment monitoring, AI can also be used to produce deepfake videos, fake voices and manufactured emotional waves. South Korea has already restricted AI deepfake content during election periods, and India's elections have seen their own AI-fake-video controversies. AI, in other words, can be a sharp tool — or a sharp weapon.
At the end of the day, AI can't replace political judgment, but it can make a campaign feel closer to the people it's trying to reach. A genuinely skilled campaign doesn't treat a voter as just another ID in a database — it understands them through data, reads them through sentiment analysis, reaches them through the right platform, and responds with content that's sincere, accurate and timely. The outcome ahead may not depend on who shouts loudest — it'll depend on who best knows how to build voter trust and confidence in the algorithmic age. That's what actually matters about AI entering the campaign trail.
Part of the AK AI Corner column. Originally published in Oriental Daily (东方日报) on Mar 12, 2026.