By Kai Curry
NORTHWEST ASIAN WEEKLY
Audrey Tang, the first nonbinary, transgender member of Taiwan’s cabinet, is on the cutting edge of artificial intelligence (AI) technologies.
Tang, who serves as Taiwan’s digital minister, has been instrumental in devising digital means to improve the transparency and approval rating of Taiwan’s government. Tang spoke on this and other concerns related to AI, in a Feb. 12 Seattle Arts & Lectures talk moderated by author Ted Chiang.
Audrey Tang speaks, during their Seattle Arts & Lectures talk on Feb. 12, 2025. Courtesy of Kai Curry.
Tang, a software developer, first gained the public eye as part of the student-led Sunflower Movement in 2014. This movement—which Tang was insistent to describe as a “demonstration” and not a “protest”—aimed to interrupt the Taiwanese government’s plan for a trade deal with China. This deal, said Tang, would have allowed companies such as Huawei access to Taiwan’s digital network.
The movement was a demonstration, said Tang, because the students offered “a different possibility.” For one, they wished to bring “transparency to the previously opaque political process” in a government that, according to Tang, had only a 9% approval rating at that time.
The demonstration worked.
“The (Taiwanese) government’s approach became, ‘Ok we can’t beat them, so invite them in,’” Tang recalled. “A key insight from the Sunflower experience was that it is not inevitable for social media to polarize people.”
Two years later, in 2016, Tang was invited to join the cabinet as the digital minister. Tang believes that polarization among people is “a consequence of the design of the platform,” and such polarization can be seen in various social media websites. In an effort to combat this, Tang set about to “create bridges” between people, ideas, and digital systems using “pro-social media tools.”
A now-famous example was Tang’s use of vTaiwan, an online discussion platform, to discuss the controversial introduction of Uber into Taiwan.
“People feared for their lives from Uber,” Tang recalled. Tang termed Uber’s digital tools, like its app, a “misaligned” use of AI.
The vTaiwan forum demonstrates Tang’s approach to the majority of problem-solving for divisive issues. The forum invited people to share their thoughts on the problem, and soon, Tang said, a “narrative became apparent” through what Tang called “bridging statements” or sentiments where people on opposed sides actually agreed.. Tang calls this “uncommon ground.”
From there, the government could create regulations, based upon the input of the community.
“The trust that grew from the process made new rules legitimate,” Tang said. “Not forced, not top-down; instead, crowd sourced.”
Following this, trust in Taiwan’s government grew.
And Tang’s work doesn’t stop there. In a recent election, Tang and volunteers employed what Tang calls “pre-bunking.” They don’t wait until misinformation has already spread, Tang explained. Instead, they have volunteer fact-checkers and people from the community share disinformation tactics online. Oftentimes, Tang said, this stops disinformation, such as deepfaking—creating a fake, AI-generated video or image of something that looks and sounds convincingly real—entirely.
In this way, Tang said, she is bringing people “from the demand side to the supply side.” Instead of top down systems that are designed to create maximum engagement (think doom scrolling), that “sell to the highest bidder” and “promote isolation and polarization,” Tang envisions AI and digital systems as copilots that allow people to “mix and match, rework things they don’t like, and contribute to the whole.”
Tang is a strong proponent of open sourcing and encouraged everyone in the tech industry in the audience to promote and use it.
“You get to keep the things that you have developed,” Tang said. Open sourcing aligns with his preference to have systems that overlap and synchronize, so that no one needs to feel the fear of missing out (“FOMO”) by leaving one platform for another.
“There’s nothing to lose if you open source things. The only thing you lose a little,” Tang said half-jokingly, is face, because “people may see the shortcomings of your work.”
Tang and Chiang also discussed the origins of AI. Chiang proposed that AI was originally focused on “understanding human cognition” and is now focused on “how can we make things that will do the jobs that people do?” Assistive technology, Tang agreed, should “extend our cognition,” such as in the case of a telescope or a microscope. It should give humans agency.
“As soon as they start replacing human-to-human relationships then, to me,” Tang said, “it switches from the assistive realm into this automating realm, which is probably almost always antisocial.”
Chiang said he disliked the amount of money spent on new AI only to develop ideas to replace humans, as well as the way that AI seems to “create distance” between humans and the world.
Many in the audience were concerned about the voting and election process in the United States, as well as how to implement Tang’s ideas in a nation much larger than Taiwan.
Tang said that methods such as pre-bunking don’t work well in the “heat” of an election that is already underway. Instead, Tang recommends starting at least half a year in advance.
Tang also pushed back against the idea that because Taiwan is small, it is therefore an easy location to implement his techniques. They reminded the audience that Taiwan has 20 languages and four official religions. In Taiwan, Tang said that they see diversity “not as a fire” to be put out but “energy to be channeled into co-creation.”
Tang said it’s possible to start using the strategies introduced in the talk at a city or state level. AI should not be “locked in an ivory tower,” Tang said, but used by everyone, and democracy should evolve at the same pace as technology.
“When we all have open, robust safety tools and a mindset of humility in AI design,” Tang said, “we can create a future where technology thrives in harmony with democratic values.”
Kai can be reached at newstips@nwasianweekly.com.