Illustration by Haley Hsu

AI revolution in politics

Impacts and ethics of AI in elections

Artificial intelligence (AI) is making waves across various sectors and politics is no exception. As we navigate an era of increasingly sophisticated AI-generated content and deepfake technologies, it’s important to understand how these advancements reshape political landscapes worldwide. 


From influencing voter outreach to posing ethical dilemmas, integrating AI into politics brings both opportunities and challenges which need thorough examination.


Deepfake technology—which comes from the terms deep learning and fake—leverages neural networks to create highly realistic audio and video content which can deceive even the keenest eyes and ears. This synthetic media is created by machine-learning algorithms named for the deep-learning methods used in the creation process and the fake events they depict.


“Deepfakes rely on models that are trained on a large amount of data using neural networks that mimic how our brains work,” stated Dr. Wu-chang Feng, a professor of computer science at Portland State. These models are trained on various content types—such as images, videos, audio and text—to generate plausible yet false content.


This technology has profound implications for political communication. “The ability to generate plausible, but untruthful content and to instantly spread it is unprecedented,” Feng stated. Informed participation is a cornerstone of democracy, but when the lines between truth and fiction blur, the very foundation of democratic engagement is at risk.


Dr. Bart Massey—an associate professor of computer science at Portland State—added that while deepfakes are not a novel concept, AI accelerates their creation and dissemination. “A skilled editor could simulate realistic content with traditional means over weeks,” Massey said. “Now, AI can do it in seconds.” This rapid production of convincing deepfakes means that misinformation can spread more quickly and widely than ever before.


The ethical implications of AI-generated deepfakes in political campaigns are manifold. Feng highlighted new issues in copyright, liability and privacy which arise using generative AI. 


Massey questioned the ethical legitimacy of using deepfakes for voter outreach, suggesting it’s hard to imagine a scenario where such use could be ethically justified. “It’s not that complicated ethically, but what is complicated is the larger impact of AI on steering people’s opinions and ideas,” he said.


Some regulatory bodies are beginning to take action in response to these ethical concerns. For instance, the United States Federal Communications Commission (FCC) has made AI-generated voices in robocalls illegal. FCC Chairwoman Jessica Rosenworcel has also proposed new rules requiring political ads on TV and radio to include disclaimers about AI-generated content, emphasizing that “consumers have a right to know when AI tools are being used.”


As deepfakes become more sophisticated, so do the methods for detecting them. Feng mentioned how automation can effectively identify specific kinds of deepfakes, particularly fake-news articles. 


However, Massey warned that this technological arms race is ongoing, and while defenders are winning now, the future is uncertain. “The age in which we can trust experts to detect deepfakes may be coming to an end,” he said, highlighting the potential for fake experts to muddy the waters further.


The use of AI in politics is not just theoretical, it’s already happening on a large scale. A recent article by WIRED illustrated how Indian politicians use deepfake technology for voter outreach. In one instance, a politician created a deepfake of himself to address 300,000 potential voters, delivering personalized messages in their regional dialects. This practice is effective and cost-efficient, making it an attractive tool for political campaigns.


However, this raises significant concerns about transparency and misinformation. The personalized nature of these messages means that voters might believe they are receiving genuine communication from a candidate when, in fact, they are interacting with a digital clone.


Beyond deepfakes, AI’s influence on politics extends to manipulating public opinion and propagating biased information. 


Massey pointed out how the ability to generate large quantities of text and other content means that AI can “sway the entire content of the internet around a particular politician or political cause.” 


The vast amounts of data collected on individuals facilitate this manipulation, allowing for highly targeted and personalized messaging.


This phenomenon is reminiscent of what Massey described as memetic infections, where AI-driven content acts like a disease spreading through a population. The rapid and widespread dissemination of tailored messages can significantly alter public perception and debate, undermining the democratic process.


Looking ahead, the role of AI in politics is poised to grow even more significant. Feng suggested that regulation may be necessary to handle malicious uses of AI. Policymakers such as Rosenworcel and Senator Amy Klobuchar echoed this sentiment by advocating for clearer guidelines and transparency in the use of AI in political advertising, according to CNN.


Moreover, tech companies are also taking steps to address the issue. According to the BBC, Meta—aka Facebook—requires political campaigns to disclose the use of deepfakes and ban its in-house generative AI tools for political advertising.


The integration of AI into the political sphere is a double-edged sword. On one hand, it offers unprecedented opportunities for personalized voter engagement and efficient communication. On the other hand, it poses significant ethical, legal and practical challenges that need to be safeguarded against to protect the integrity of democratic processes.


Understanding the implications of AI in politics is crucial for students and future leaders. It’s not just about being informed voters, it’s about being critical thinkers who can navigate and scrutinize the digital content we consume. The broader context of AI in politics involves “taking a lot of the agency for free choice away from people,” Massey said. And it’s up to us to reclaim that agency.