On the sidelines of the 78th United Nations General Assembly in New York last week, Taiwan’s Minister of Digital Affairs Audrey Tang delivered a speech at the Concordia Annual Summit on digital democracy and artificial intelligence. VOA spoke with Tang about how AI might help break through China’s censorship and the challenges and opportunities the technology brings to global democracy.

This interview has been edited for clarity and brevity.

VOA: You mentioned the concept of AI governance in your speech. In addition to being applied to democratic countries, can this concept also be applied to totalitarian countries?

Tang: From 2010 to 2012 and 2013, consultative democracy was also studied in some places in the CCP [Chinese Communist Party], but not at the central level. Later, including freedom of the press, assembly, association and expression, all aspects were restricted. If you look at the papers, there have been very few studies promoting consultative democracy through the internet in recent years. Conceptually and theoretically, if we look at the situation in 2000, it seemed possible. But if we look at the situation in recent years, there seems to be no research on this. Maybe the premise is freedom of the press, and people must fully understand what the truth is. Regardless of consultation or deliberation, this foundation is needed. If freedom of the press is deprived, it will not be easy to develop further.

VOA: What do you think of the potential and limitations of artificial intelligence in China?

Tang: Artificial intelligence technology can turn public information on the entire Internet into material. After the [open-source] model is successfully created, it can actually be saved on a USB flash drive. Anyone can run this model on a laptop or personal computer, which greatly challenges censorship. In the past, as long as a comprehensive totalitarian government guarded access to Google, Wikipedia or other websites, it would make it difficult for their people to get the truth. But now, as long as netizens can get the [open-source] language models on a flash drive … they will no longer need an internet connection, and they can always find out what happened and when exactly did it happen.

VOA: Former White House deputy national security adviser Matt Pottinger has proposed punching holes in China’s Great Firewall. Is that possible?

Tang: If the holes are temporary, they will be repaired once people find them. The language models I just mentioned don’t mean that there needs to be holes continuously, but that if there is a certain way to send information in, there is no need to connect to the outside anymore. The bad use is that if a Trojan horse is sent in, there is no need to control the computer anymore, and it can conduct network attacks by itself. This is a bad use. But there is another use, which is to send in the truth, and after that, it can even tell the users in the interactive Q&A format, including showing contemporaneous photos and videos from the time.

VOA: Does your upbringing and way of thinking help you with your current work?

Tang: I am nonbinary not only in terms of gender but also in terms of ideology. Many people have to make choices, such as the left and the right [politically]. For me, they all coexist. I don’t particularly feel I’m very close to or very far from half of the people because they are rightists or leftists. I don’t think that way. Everyone is at the same distance. If I can’t understand the ideas of a certain side, I will feel I’m not plural enough, and I should be in touch with them more.
…