Indonesia puts Generative AI to the test in world’s third largest democracy’s elections

Fika Juliana Putri, a 19-year-old shopkeeper in East Jakarta, intends to cast her vote in Indonesia’s upcoming presidential election for a once-feared former special forces commander. She expresses her fondness for him, citing his endearing demeanor.

Across Indonesia, billboards proudly display a cartoon depiction of General Prabowo Subianto, created using generative AI technology. This animated likeness adorns sweatshirts, stickers, and garners attention on social media platforms like TikTok under the #Prabowo tag, accumulating approximately 19 billion views.

Although Prabowo serves as Indonesia’s defence minister, his AI-generated persona, characterised by a cherubic appearance, engages in playful gestures like Korean-style finger hearts and affectionately holding his cat, Bobby. This portrayal resonates particularly well with Gen Z voters, who constitute about half of Indonesia’s 205 million eligible voters.

With the general elections scheduled for February 14 in Indonesia, experts suggest that this unprecedented use of generative AI could reshape political campaigning on a large scale.

The AI-generated cartoon plays a pivotal role in Prabowo’s electoral rebranding, as he now emphasizes a “gemoy” persona, which translates to cute and cuddly in Indonesian slang. This shift marks a departure from his previous image as a fiery nationalist during two unsuccessful presidential campaigns.

“I’ll vote for him because he’s gemoy,” explains Putri, who is voting for the first time. “That’s the main reason.”

Prabowo, along with numerous other candidates, harnesses generative AI tools to produce campaign materials, monitor social media sentiment, develop interactive chatbots, and target potential voters. However, both the Prabowo campaign and Midjourney Inc., the U.S. firm behind the AI technology, declined to comment on their collaboration.

Experts observe that this election in Indonesia, the world’s third-largest democracy, serves as a testing ground for the potential impact of generative AI on political campaigns.

Reuters conducted interviews with 26 individuals, including political advisors, lobbyists, tech executives, experts, and artists involved in creating generative AI images for politicians, to assess the use of AI technology in Indonesia’s campaign. Several interviewees requested anonymity due to lack of authorisation to speak to the media.

Vendors and campaigns are pushing the boundaries of guidelines set by providers like Midjourney and OpenAI, the leading market player in generative AI. Despite the absence of binding regulations from the Indonesian government on AI tool usage, concerns arise regarding the ethical implications of such technology in politics.

As countries with one-third of the world’s population gear up for elections this year, including the United States and India, the Indonesian elections serve as a litmus test for OpenAI’s policies. Many AI tools employed in Indonesia’s election are powered by OpenAI, as confirmed by senior campaign staffers.

However, OpenAI recently updated its rules to prohibit the use of its technology for political campaigning, including the creation of personalized campaign materials. Despite this, political consultant Yose Rizal acknowledges using OpenAI’s GPT-4 and 3.5 software for his Pemilu.AI app, aimed at crafting hyper-local campaign strategies for legislative candidates.

Rizal plans to introduce Pemilu.AI to India ahead of its general election, citing the Indonesian election as a precursor. He emphasises compliance with regulations, having collaborated closely with Microsoft for hosting on its Azure cloud service.

Rizal mentions trialing a version of Pemilu.AI on Google’s AI, indicating a broader adoption of AI tools in political communication.

The elections in Indonesia also underscore the potential of generative AI in shaping political discourse. Anies Baswedan’s campaign, for instance, launched a WhatsApp chatbot powered by OpenAI to address policy-related queries.

Despite controversies surrounding AI-generated content, such as the release of “deepfake” videos featuring the late leader Suharto, there’s limited evidence of widespread misinformation dissemination compared to previous elections. However, concerns persist regarding AI’s potential for voter manipulation and the need for robust regulations to safeguard democratic processes.