Nadira, one of tvOne’s newest anchors, delivers the latest news on Thailand’s election in perfect, measured English. Two days earlier, she had reported on the South East Asian Games sports spectacle in flawless Bahasa Indonesia. Her knowing expression never changes, and her flowing hijab is always perfectly set. She’s a virtual television anchor, one of three introduced in April by tvOne, one of Indonesia’s most-watched broadcast channels.
Nadira is based on the real-life news anchor Fahada Indi, who has worked at tvOne for four years. Nadira’s voice and movements were modeled on Indi’s own, and when Rest of World visited the channel’s studios, the similarities in their appearance and demeanor were evident, right down to their signature hijabs.
Despite global angst about the potential for advanced automation to replace human labor, Indi told Rest of World that working on the cutting edge of technology was an “honor.” More than that, it was practical: “It’s a huge help that I can do it without being physically present at the offices,” she said.
Indi and Nadira work in tandem. Indi narrates real-time news developments as she reports from anywhere across the archipelago, while back in the studio, the polished Nadira avatar gestures and moves to Indi’s voice. The channel’s translation system uses artificial intelligence to convert those vocals into British and American-accented English, Chinese, as well as local dialects like Javanese and Sundanese, for different audiences across the country’s more than 300 distinct ethnic groups.
The channel also uses two other avatars, both female, to appeal to some of the country’s leading demographic groups. Sasya is an imposing-looking Chinese-Indonesian avatar, and Bhoomi is a curly-haired Eastern Indonesian.
Digital avatars have crept into the mainstream in Indonesia. While some virtual humans present the news, others have gained a following as influencers. These “metahumans,” as they’re called, are used not just for lifestyle advertising, but also in more authoritative positions: The immigration office of Indonesia’s primary international airport now has four virtual spokespeople.
Many of these developments are being promoted as AI operations, even if artificial intelligence plays only a supporting role. Excitement — even hype — in Indonesia around the AI boom is widespread. “I am confident that [AI] could potentially become the most impactful technological revolution ever created by mankind,” declared OpenAI CEO Sam Altman to a crowd in Jakarta last week, standing alongside some of Indonesia’s most powerful officials.
In May, Soekarno Hatta Airport’s immigration department became the first government office to use metahumans. That month, the department launched its four-person virtual immigration support team — some of its avatars were inspired by Indonesian celebrities including Ahmad Al Ghazali, Anjasmara, and Ario Bayu. The figures serve as public service communication officers: friendly, government-aligned faces armed with the latest information for travelers.
As with tvOne, each avatar is crafted to represent an Indonesian demographic. Despite being promoted by the Indonesian government as AI avatars, the figures are not yet interactive, although they have been generated using AI software. Indonesians have not been wholly impressed — in the rollout video, the avatars’ speech tracks were visibly out of sync with the animation, and the figures looked awkwardly cut out against a flat background.
Justito Adiprasetio, a lecturer in media studies and popular culture at Padjadjaran University, told Rest of World that though several Indonesian media organizations have made a “leap” into AI, some have done so without fully understanding how to use the technology. “Many [media] utilize it for branding, but they don’t have the capacity for it,” said Adiprasetio. “Because [AI] cannot suddenly be implemented without involving full awareness or understanding the effects of technology.”
There is a difference between digital assets supported by AI, and true AI-generated products. Most metahumans, for now, fall into the former category. They include Vira, a metahuman personality layered over an AI chatbot for BCA — Indonesia’s largest privately owned bank — by local startup Genexyz. The company specializes in creating metahuman influencers such as Laverda Salsabila, a virtual Indonesian folk singer with real-life endorsement deals.
Salsabila follows in the footsteps of virtual influencers like the forever 19-year-old Lil Miquela. The latter made waves in the Western world in the late 2010s after brand tie-ups with the likes of Prada and Calvin Klein, but failed to hold mainstream attention.
Metahumans still need a human touch to gain the trust of their audiences, tvOne’s CEO, Taufan Eko Nugroho, told Rest of World. The team experimented with news fully generated by AI, but found that using human presenters had better results when it came to aspects such as intonations.
“We pay great attention to details, such as the [avatars’] eye contact and camera position,” Nugroho said. As long as the news has human decision-making behind it, he said, the media can avoid problems related to AI-generated products, such as false information and stereotyping.
The metahuman news anchors are designed and maintained by an in-house team. None of the existing staff have been laid off, Nugroho said. The channel plans to develop more of these avatars in the future.
Humans aren’t yet at the point of being entirely replaced, according to Norma Dani Risdiandita, Widya Wicara’s vice president of technology.
“Because human creativity is still needed to fill out a virtual persona like a newscaster,” Risdiandita told Rest of World. “AI has a long way to go before it can match human inventiveness.”
