TOKYO - Artificial intelligence may pose both opportunities and threats to democracy, says Yutaka Matsuo, a leading AI researcher at the University of Tokyo, who warns that voters must remain vigilant about the influence of AI-generated content in political campaigns.
In a wide-ranging interview, Matsuo emphasized that while AI can help deepen voters' understanding of political issues, it can also be misused to manipulate opinions, distort facts, and undermine fair elections.
Matsuo pointed out that with the spread of deepfakes and synthetic media on social networks, there is growing concern that AI is already harming democratic processes. "People need to be careful not to accept information at face value," he said. “Even though some platforms use AI for fact-checking, no system is perfect. It’s important to check sources and cross-reference information.”
Looking ahead, Matsuo said he believes that AI, if used correctly, could enhance democracy by helping both candidates and voters clarify their positions. For example, AI tools could summarize complex policies or engage in dialogue with users to help them think more deeply about their choices. However, he also warned that malicious uses—such as embedding language that only AI understands to bias analysis—could mislead voters who rely on AI-generated explanations.
He likened the potential for AI-driven manipulation to past cases such as Cambridge Analytica, where user data was weaponized to influence political behavior. “Marketing and opinion shaping are not that different,” he noted. “As techniques to persuade consumers evolve, they could be adapted to guide voter behavior as well.”
Matsuo stressed the importance of transparency in how AI is used in political communication. Voters, he said, have the right to know whether the information they are seeing has been tailored to them using AI and whether it is artificially generated. He added that this awareness is essential for making thoughtful voting decisions: “We should ask ourselves, why am I voting for this candidate? What influenced me? Was it a social media video I saw yesterday? Was it real?”
He also observed a divide between established political parties and newer ones when it comes to AI and social media. While major parties tend to be cautious, smaller parties often embrace these tools more aggressively, leveraging digital platforms to amplify their voices. “AI and SNS can empower smaller groups, but the impact depends on how they’re used,” he said. “There’s a wide range of outcomes, both positive and negative.”
Despite the potential for misuse, Matsuo said elections should ultimately be guided by accurate and reliable information. “AI can be a powerful tool for understanding,” he concluded, “but only if it is used properly and voters stay aware of how they’re being influenced.”
When asked whether Japan’s current election system is well-designed from an AI researcher's perspective, Matsuo described it as “a system that may not capture every nuanced preference of voters but one that avoids making major mistakes.” In other words, while imperfect, the democratic process is structured to reflect public will in a broadly stable way—a view that underscores the need for thoughtful integration of AI into future political systems.
Source: テレ東BIZ
