Taiwan is very advanced, and mainland China is also progressing quickly in electronics. What are the differences between Taiwan’s and mainland China’s digital policies? China has many “surveillance” practices and claims to be the world’s safest country. Where will this “full application of digital technology to monitor citizens” lead?

What technologies are you following or particularly interested in?

In your theory, you talk about the importance of inclusivity in social development. How can we help those who cannot keep up with artificial intelligence and digital progress join in the common development of society?

Will AI not replace humans and begin controlling society? Won’t AI lead to humans being marginalized or classified and selected? In your book, you mentioned that as AI becomes more widespread, humans should focus on creating more public value. What role should humans play again?

Japan’s population is aging. Can continued technological advances, including AI, overcome health concerns? Also, how should people collaborate with AI and robots? For example, robots are widely used in factories moving toward automation. What kind of work should people value, and what role should they play?

The third question: Powerful technology companies amass more wealth than others and earlier. Investors who find these companies can donate a lot of money, and wealth continues to flow into the future. Will technological progress bring happiness to the majority of people? What do people need to be happy?

The second question: It’s said that by 2045 and 2050, AI will reach a technological turning point beyond human capabilities, which will have comprehensive impacts on humanity. Is it possible that the ethical and moral issues of AI we’re concerned about today won’t necessarily happen as predicted?

When you imagine the future, how far ahead do you typically think?

Sure, no worries. I’ll send you the link once the article goes live this week.

All right, that’s about all I wanted to ask today. Thank you so much for speaking with me, especially so late in the evening.

I see. One last question: over the past week, people have been talking a lot about DeepSeek. It’s only been a short time, but do you have any thoughts on its future?

I see, thank you. How do you, personally, feel about R1’s development, given you’ve used it for some time?

Understood. Has the censorship approach overshadowed its initial “wow factor”? Or would you say that if one simply downloads and retrains it locally, it’s not really affected?

Could the ideological layer also create a security issue by steering user answers in a certain direction?

I see. So the security concerns now raised in various places essentially mirror what you’ve described?

Are you saying the risk is analogous to other Chinese apps, such as Xiaohongshu (Little Red Book)?

Understood. In non-Chinese-speaking regions, the DeepSeek app soared to second place in app download charts last week. I’m curious: for people who download the app, where do you think the appeal lies, given the heavy content moderation?

Yes, precisely. If the DeepSeek app or website clearly imposes ideological filters whenever you ask about public affairs or society, might users just decide not to bother with DeepSeek?

Yes, I see. But for users who aren’t very tech-savvy and only use the app or website version, do you think its censorship approach might affect which AI product they choose? In other words, what are the hidden downsides for more “lay” users?

So using English might also yield different results in the online version?

Yes. But regarding the online version, do you think censorship has gone so far that, as some reports mention, it might even refuse to answer “Who is Xi Jinping?”? It appears many queries aren’t answered at all.

We touched on how different instructions or keywords might produce drastically different results. For instance, “Tiananmen Incident” might yield a starkly different answer from “Tiananmen Massacre.”

I see.

Right. So that underlying ideology is essentially because of the way it was trained—“fed” by data that inherently contains those biases?

Right. So at the time, once you realised that difference, you surmised that perhaps the offline version is less subject to the heavy moderation found on the online or app versions. Also, because you mentioned that it displays its “thinking” process, some users say that when they try to input ...

Understood. Earlier, we also noted that when you responded to people, you explained certain “unlocking” methods for it. Did you figure out how to bypass censorship because you knew that the offline version lacks some of the safeguards present in the online or app versions? How did you go about ...

I see. So, last week when many media outlets were comparing R1 to O1 Pro, how did you feel when you saw those comparisons?

When you first used R1, what were your impressions? Did you feel it had truly… perhaps surpassed OpenAI’s latest model?

Yes, the newest version.

Let’s begin by discussing your very first experience of using DeepSeek. When it launched last week, people were quite amazed by it. Could you tell us what that first encounter was like?

Audrey Tang, thank you very much indeed, Taiwan’s cyber ambassador at large.

Does that leave citizens of a certain country at risk of almost state action via DeepSeek, or are we talking more about individual, perhaps criminal enterprises?

You touched on some of the potential negative impacts of this breakthrough. Can you expand on your major concerns?

Does that mean its rivals are going to have to follow suit? That everything is going to become cheaper?

Okay, there’s plenty to unpack there. Let’s start perhaps with creating it, using a smaller model as it were. Is that part of why it’s simply been so much cheaper to produce DeepSeek than its competitors?

Yeah, I wondered what the strengths of DeepSeek were. What makes it so impressive at this early stage?

So let’s discuss this and some of the wider issues around AI with Audrey Tang, Taiwan’s cyber ambassador at large and former Minister of Technology. Welcome to the programme. What are the strengths of DeepSeek?

DeepSeek claims that its model was made at a fraction of the cost of its rivals. It’s also raised cybersecurity concerns in some countries, with the Australian Science Minister, for example, urging caution. Nevertheless, DeepSeek has become the most downloaded free app in the US just a week after it ...

On Tuesday, the US President Donald Trump called the rise of the Chinese company DeepSeek a wake-up call for the US tech industry after the emergence of its artificial intelligence (AI) model triggered shockwaves, not least on Wall Street. Shares in major tech firms such as NVIDIA fell sharply, with ...

All right, great. Well, thank you for speaking with me. It’s been a pleasure. And hopefully we’ll speak again.

Great, great. So, yeah, I think for now those are all of my questions and is there anything, you know, our audience is probably going to be mostly in Europe, but any other thoughts that you would just sort of last minute want them to keep in mind?

And broadband is a human right.

So what do you think would be a good way to sort of bridge that gap as, you know, input in society increasingly requires more technical knowledge? I mean, the obvious answer, I think, is education.

So, yeah. What would you say about ways we can think about it? Because to the. For the. For most people, most jobs are non, you know, not centered around technology. Our education systems may have, you know, the basics of computer science, but not necessarily, you know, enough for ...

Okay, great. So, yeah, what I would also want to talk about is, you know, people are scared at this point of technology. Like you mentioned this in your book, that in a lot of democracies around the world, you know, we mostly see it as. We haven’t been able ...

So how is this something you would generally advocate and if so, how to do it successfully?

Great. And so what would you say as far as you mentioned, the authentic authentication procedures? So we have seen that in Taiwan successfully because being used to hold platforms accountable. I think it’s also worth mentioning sort of the case of South Korea where, you know, they were attempting ...

Okay, great. Because yeah, I think with a lot of the discussion around these, you know, governments sometimes treat it as a military issue or, you know, foreign policy issue. But it sounds like, you know, that these measures that you’re talking about and more directly operating with the public ...

So, okay, so citizen input on how these information platforms are run. That seems like a very effective way to avoid the tricky issues of, you know, top down governments dictating to platforms.

What would be sort of the next step in the online sphere?