1
00:00:00,937 --> 00:00:03,602
Angela, welcome to the Road to Accountable AI.

2
00:00:04,440 --> 00:00:06,530
Hi, hello, thanks for having me.

3
00:00:07,637 --> 00:00:19,204
Before we get to a conversation about AI in China, can you talk a little bit about the
history of the tech sector there and maybe give people some framework for comparison to

4
00:00:19,204 --> 00:00:21,625
the US or other countries they may be familiar with?

5
00:00:22,446 --> 00:00:31,606
Well, China is really the only country other than the United States that have been able to
foster big tech companies so far.

6
00:00:32,106 --> 00:00:42,786
in early 2021, like Alibaba, Tencent, I mean, these companies were on the verge of
reaching the one trillion dollar market cap.

7
00:00:42,786 --> 00:00:51,486
And so soon would join this trillion dollar market cap club, although subsequently they
face a big crackdown.

8
00:00:52,450 --> 00:00:54,762
They suffer severe market cap loss.

9
00:00:54,762 --> 00:01:00,235
But seriously, we are talking about the world's largest big tech companies.

10
00:01:00,375 --> 00:01:15,736
And partly because in the consumer tech sector, Chinese government imposed this, erected
this great firewall that have made it very difficult for US social media businesses to

11
00:01:15,736 --> 00:01:17,067
operate in China.

12
00:01:17,067 --> 00:01:21,390
So in a way that that create entry barrier, very high entry barrier.

13
00:01:21,390 --> 00:01:23,050
for US companies.

14
00:01:23,230 --> 00:01:29,050
So far, only US firm that have been able to gain a little bit foothold is Microsoft.

15
00:01:29,070 --> 00:01:34,224
But still, Bing has only 1 % of the market share in China's search business.

16
00:01:36,393 --> 00:01:45,215
Many Americans that I talk to are under the impression that there's no difference between
the Chinese government or the Chinese Communist Party and private companies like Alibaba

17
00:01:45,215 --> 00:01:46,768
and ByteDance and so forth.

18
00:01:46,768 --> 00:01:48,369
Is that accurate?

19
00:01:49,176 --> 00:01:54,580
Well, I think that's definitely an oversimplification of the situation in China.

20
00:01:54,580 --> 00:02:00,724
Look, mean, Elon Musk also operates a hugely successful business in China.

21
00:02:00,724 --> 00:02:15,284
And China is the reason why he became the richest man in the world, because the Chinese
government really helped Tesla to create a very, to have access to a very convenient

22
00:02:15,284 --> 00:02:17,726
supply chain and also to

23
00:02:17,982 --> 00:02:24,084
open up a huge production assembly line in Shanghai.

24
00:02:24,084 --> 00:02:32,686
So the government was in many ways extending helping hands to its own tech firms and even
to Elon Musk's company as well.

25
00:02:32,686 --> 00:02:41,068
So in that way, that is something that you don't see in other jurisdictions.

26
00:02:41,268 --> 00:02:47,690
And it's also true that companies operating in China, they exhibit

27
00:02:47,690 --> 00:02:53,154
a much higher level of cooperation with the government.

28
00:02:54,195 --> 00:03:00,901
You don't see them fighting with the government in courts, in anti -trust or big data
cases.

29
00:03:00,901 --> 00:03:13,230
Even Elon Musk himself, which sent poor emoji to the US government, will be really, really
friendly to the Chinese government.

30
00:03:13,291 --> 00:03:17,206
He came and visited the Chinese government a few months ago.

31
00:03:17,206 --> 00:03:28,657
And that's why the Tesla car now can drive Tesla car in some of the government compounds,
which were restricted before.

32
00:03:28,958 --> 00:03:34,504
mean, so I think it's definitely different type of relationship.

33
00:03:34,504 --> 00:03:41,410
But I wouldn't say these companies are completely watchdog of the Chinese government.

34
00:03:42,485 --> 00:03:47,440
You mentioned earlier the crackdown on companies like Alibaba and so forth.

35
00:03:47,440 --> 00:03:51,463
So how did the relationship with the Chinese state and the tech sector change?

36
00:03:52,462 --> 00:03:56,012
For a really long time, they were in the honeymoon period.

37
00:03:56,012 --> 00:04:07,322
I would say from the early 2010s until late 2020, the Chinese government has played a very
important role propping up Chinese tech sector.

38
00:04:07,902 --> 00:04:15,682
And in this period, I would see regulation more as a helping hand rather than a grabbing
hand.

39
00:04:16,142 --> 00:04:22,786
However, since late 2020, we experienced a massive crackdown of

40
00:04:22,830 --> 00:04:31,250
China's own indigenous tech sector and big companies like Alibaba and Tansan, everybody
else were affected.

41
00:04:32,010 --> 00:04:36,540
And these companies really suffer a great deal, right?

42
00:04:36,540 --> 00:04:44,130
mean, so since then, the government has really tightened its scrutiny over these farms.

43
00:04:44,130 --> 00:04:48,450
However, that campaign only lasted for 18 months.

44
00:04:48,450 --> 00:04:50,414
So the government relaxed.

45
00:04:50,414 --> 00:04:54,754
its regulatory control in early 2022.

46
00:04:54,754 --> 00:04:57,634
So now we are back to a lag stage.

47
00:04:57,634 --> 00:05:06,854
So there's a lot of volatility in Chinese regulation and there's something that is very
uncommonly seen in other jurisdictions.

48
00:05:06,854 --> 00:05:11,892
And that's something that I elaborated in detail in my book, High Wire.

49
00:05:14,005 --> 00:05:15,537
was the rationale for that crackdown?

50
00:05:15,537 --> 00:05:22,353
Was it similar to the antitrust motivation that we've seen in Europe, in the US, vis -a
-vis the big tech platforms?

51
00:05:24,002 --> 00:05:24,652
Yes and no.

52
00:05:24,652 --> 00:05:31,727
OK, yes, because these kind of actions are actually way overdue.

53
00:05:32,028 --> 00:05:39,472
And the Chinese tech sector is actually much more concentrated than the US and the
European tech sector.

54
00:05:39,653 --> 00:05:52,741
China is home to really two duopolies, Alibaba and Tencent, which hold a massive hundreds
of portfolio companies each in their own ecosystem.

55
00:05:53,038 --> 00:06:03,262
created this erected this wall gardens that prevent companies, rival companies to access
the ecosystem.

56
00:06:03,502 --> 00:06:06,844
And so create all sorts of competition issues.

57
00:06:07,504 --> 00:06:16,968
And however, because for a really long time, the government was trying to nurture this
industry, regulators took a very lack stand.

58
00:06:17,228 --> 00:06:21,422
So that kind of bureaucratic inertia or

59
00:06:21,422 --> 00:06:26,102
agency inaction also partly further exacerbate this problem.

60
00:06:26,182 --> 00:06:39,682
Now, in the late 2020, what really changed the game, however, is the government woke up
and realized, look, I these companies, the Alibaba, Tencent, they're just not, they're not

61
00:06:39,682 --> 00:06:44,442
just selling clothes or video games.

62
00:06:44,442 --> 00:06:50,694
They are now, you know, controlling our own financial system because

63
00:06:50,830 --> 00:07:05,830
At the end of 2020, Alibaba was, not Alibaba, Ant Group, which is a sister company of
Alibaba, and it's founded by Jack Ma, was on the verge of getting listed in Shanghai and

64
00:07:05,830 --> 00:07:06,350
Hong Kong.

65
00:07:06,350 --> 00:07:11,330
And this was supposed to be the biggest IPO of the year.

66
00:07:12,050 --> 00:07:19,394
And one of the big parts of Ant's business is the...

67
00:07:19,470 --> 00:07:32,470
the credit tag businesses, which Ant as an intermediary connecting ordinary users and the
small and medium sized state -owned banks.

68
00:07:32,470 --> 00:07:40,130
But Ant had very little skin in the game, so the Chinese government was quite concerned
about the moral hazard problem.

69
00:07:40,130 --> 00:07:49,358
But what worries them more is that Ant got an extremely high valuation of 320 billion US
dollars.

70
00:07:49,358 --> 00:07:58,138
In other words, that valuation at the time put in at a higher, now put in as a bigger bank
than JP Morgan at the time.

71
00:07:58,138 --> 00:08:12,698
I mean, so this was a 60 year old company that's been from Alibaba and it's grow to be
such a huge scale and size and with a big hidden problem of moral hazard.

72
00:08:12,698 --> 00:08:16,308
And that really worries the Chinese government.

73
00:08:16,308 --> 00:08:18,702
I mean, that explains why

74
00:08:18,702 --> 00:08:22,722
They suspended Ant's IPO at 11 hour.

75
00:08:23,202 --> 00:08:32,202
So a lot of people were looking at that whole saga as like, Jack Ma said something stupid
at the financial summit.

76
00:08:32,202 --> 00:08:35,521
And so Xi Jinping got very upset.

77
00:08:35,702 --> 00:08:39,772
And then they initiated this huge crackdown.

78
00:08:39,772 --> 00:08:41,822
But the story goes deeper than that.

79
00:08:41,822 --> 00:08:45,262
mean, actually the tension went way back.

80
00:08:45,262 --> 00:08:47,242
I mean, it's fundamentally.

81
00:08:47,254 --> 00:08:56,002
It was this huge concentration of these big tech companies and their pervasive influence
that really alarmed the Chinese government.

82
00:08:56,002 --> 00:09:05,780
And they realized, you know, these have threatened potentially financial stability,
economic stability, and ultimately their political control.

83
00:09:08,073 --> 00:09:10,074
Okay, let's pivot to AIs.

84
00:09:10,074 --> 00:09:13,477
That's what I really want to focus on with you for most of our time.

85
00:09:13,498 --> 00:09:17,151
You talked about the high barriers to entry for the US tech companies in China.

86
00:09:17,151 --> 00:09:23,626
And similarly, the big Chinese companies with the possible exception of TikTok are not
prominent in the US.

87
00:09:23,646 --> 00:09:27,399
Is the same dynamic playing out in the AI sector?

88
00:09:28,406 --> 00:09:30,667
Yes, in a way, yes it is.

89
00:09:30,667 --> 00:09:34,449
Although AI, we are talking about many different services.

90
00:09:34,449 --> 00:09:37,120
There are hardware services, software services.

91
00:09:37,120 --> 00:09:41,151
Give an example of facial recognition technology.

92
00:09:41,151 --> 00:09:45,113
This is the most widely applied technology in China.

93
00:09:45,113 --> 00:09:50,595
And China really has become the biggest exporter of facial recognition technology.

94
00:09:50,595 --> 00:09:56,810
But no Western countries want to purchase the Chinese tech product.

95
00:09:56,810 --> 00:09:57,160
right?

96
00:09:57,160 --> 00:10:00,543
Even though they are the most accurate, they are really efficient.

97
00:10:00,543 --> 00:10:02,484
If you come to China, you will experience that.

98
00:10:02,484 --> 00:10:03,305
It's everywhere.

99
00:10:03,305 --> 00:10:04,525
The camera is everywhere.

100
00:10:04,525 --> 00:10:11,821
And you basically go through all the security check seamlessly and highly efficient.

101
00:10:11,821 --> 00:10:15,583
So I was quite frustrated when I live in US or in Europe.

102
00:10:15,583 --> 00:10:19,226
You still need a lot of human control of the system.

103
00:10:19,226 --> 00:10:25,740
But in China, you really experience extreme level of efficiency with this technology.

104
00:10:26,494 --> 00:10:36,780
And similarly, in China, you can't access a track GPT, although it's not the government's
choice to block it.

105
00:10:36,940 --> 00:10:42,903
It's more of a track GPT doesn't open AI doesn't want to operate in China, even block its
API in China.

106
00:10:42,903 --> 00:10:53,769
But Microsoft operate in China and Microsoft continue to offer similar services through
its Azure to Chinese companies.

107
00:10:55,638 --> 00:11:04,120
a little bit integration, but to the extent this dynamic, this kind of splinter net
dynamics continues in the age of AI.

108
00:11:06,017 --> 00:11:09,959
And what has been the Chinese state's approach to regulating AI?

109
00:11:10,722 --> 00:11:15,724
Yeah, I mean, the Chinese government has a very ambivalent approach in regulating AI.

110
00:11:15,724 --> 00:11:18,625
Well, I mean, it depends on which technology we're talking about.

111
00:11:18,625 --> 00:11:22,407
mean, go back to facial recognition technology.

112
00:11:23,247 --> 00:11:30,570
This technology has been so widely used in China in public and commercial settings.

113
00:11:30,951 --> 00:11:34,192
And this has generated a lot of controversy.

114
00:11:34,192 --> 00:11:39,724
And in fact, in some cases, even some lawsuits from some activist lawyers.

115
00:11:40,675 --> 00:11:49,841
And I think there are growing discontent with some of the Chinese citizens against this
kind of public surveillance.

116
00:11:50,022 --> 00:11:59,068
However, to this day, China has not adopted a comprehensive legislation to regulate this
technology.

117
00:12:00,609 --> 00:12:03,552
For sure, if China wants to have one, it can have it tomorrow.

118
00:12:03,552 --> 00:12:07,064
And it has been drafting it for a long time.

119
00:12:07,064 --> 00:12:12,697
So the reason for a delay is not that China does not have the capacity to write this law
or to adopt it.

120
00:12:12,697 --> 00:12:22,933
It's more of other institutional and bureaucratic interests holding it back because they
don't want to stifle the commercial growth of this technology.

121
00:12:22,933 --> 00:12:36,450
I mean, if an AI technology does not affect the fundamentals, the bottom lines of the
Chinese Communist Party, I don't see China is in any way in a rush.

122
00:12:36,460 --> 00:12:46,778
to adopt a piece of legislation that might potentially hold back innovation that might
deter investment and entrepreneurship.

123
00:12:46,778 --> 00:12:53,281
Now, on the other hand, when CheckGPT emerged, China was the first one to rush to the
scene.

124
00:12:53,342 --> 00:13:02,068
And pulling back actually suspended a lot of the CheckGPT -like services in Apple's App
Store.

125
00:13:04,430 --> 00:13:07,232
just three months after its launch.

126
00:13:07,272 --> 00:13:19,701
And China was also the first one to promulgate quite comprehensive rules called interim
measures to regulate genitive AI in just six months after CHKGPT's release.

127
00:13:19,701 --> 00:13:26,426
And then a year ago, China already adopted this law way ahead of the EU's AI Act.

128
00:13:26,426 --> 00:13:29,607
Now, what explained that proactivity there?

129
00:13:30,008 --> 00:13:32,930
This is also quite intuitive because

130
00:13:32,940 --> 00:13:43,858
This again goes back to the bottom line of the Chinese Communist Party because they are
concerned that AI, genitive output, particularly from this genitive AI technology may not

131
00:13:43,858 --> 00:13:48,641
be aligned with the political value of the Chinese Communist Party.

132
00:13:48,881 --> 00:13:52,824
So they call it the socialist core values, right?

133
00:13:52,824 --> 00:13:54,965
mean, which is a very broad concept.

134
00:13:55,005 --> 00:13:56,110
And...

135
00:13:56,110 --> 00:13:59,550
And this is something that the government was most concerned about.

136
00:13:59,550 --> 00:14:00,900
So you can imagine, right?

137
00:14:00,900 --> 00:14:05,390
mean, so this type of regulation will be really targeted at content moderation.

138
00:14:05,390 --> 00:14:09,530
But funny enough, you see the exact contrast in the US.

139
00:14:09,530 --> 00:14:20,660
This is something that the US really lacks because of Section 230 and a lot of
constitutional rights protection in the US, right?

140
00:14:20,660 --> 00:14:25,268
But in China, content moderation is the single area.

141
00:14:25,268 --> 00:14:28,643
which the government has imposed the most control.

142
00:14:30,635 --> 00:14:31,846
So that's interesting.

143
00:14:31,846 --> 00:14:39,740
what you're saying is some things like facial recognition and emotion detection and so
forth are the things that are most heavily regulated in the US and Europe.

144
00:14:39,740 --> 00:14:46,033
And some things like content moderation, which are the least regulated here or the most
regulated in China and place versa.

145
00:14:47,212 --> 00:14:54,646
Yeah, I mean, I'm sure that autonomous driving and facial recognition, they will be still
subject to some sort of sector regulation.

146
00:14:54,646 --> 00:15:03,491
I once talked to one of the GC at general council at a leading Chinese automobile company.

147
00:15:03,491 --> 00:15:13,516
Is that my day job, I spent half of my day job just meeting different government officials
and different government regulators and from various sectors.

148
00:15:13,516 --> 00:15:16,664
I mean, they're not just dealing with one agency.

149
00:15:16,664 --> 00:15:19,615
They're dealing with the whole bureaucracy here.

150
00:15:19,736 --> 00:15:22,557
However, they're taking a consensual approach.

151
00:15:22,557 --> 00:15:26,990
It's not like the government make a big announcement like how they deal with Alibaba.

152
00:15:26,990 --> 00:15:34,704
I mean, suddenly, an announcement out of the blue shocked the whole huge investor, the
huge equity market.

153
00:15:34,704 --> 00:15:37,286
It's more a consensual approach.

154
00:15:37,286 --> 00:15:40,828
here are the things that you need to do and you need to tweak.

155
00:15:40,828 --> 00:15:44,710
And here's what you can do to do compliance.

156
00:15:45,196 --> 00:15:48,493
So it's more friendly approach, right?

157
00:15:48,493 --> 00:15:55,446
mean, and so a more softer way of regulating these businesses.

158
00:15:56,459 --> 00:16:09,663
Given the nature of generative AI though, where it's generative, it's you can't be
absolutely certain what it's going to create, is this extreme strictness about content and

159
00:16:09,663 --> 00:16:18,195
presumably a great deal of fear among the companies if they get it wrong in terms of what
the consequences might be, you much more so than if open AI has some bad output.

160
00:16:18,195 --> 00:16:26,077
Is that going to be, you think, a fundamental limit on the success and the quality of
Chinese large language models?

161
00:16:27,830 --> 00:16:44,934
It definitely will hinder the development of the B2C products because in China, if you
want to introduce public saving, generative AI product and services, you need to obtain a

162
00:16:44,934 --> 00:16:48,167
license from the cyberspace administration of China.

163
00:16:48,167 --> 00:16:49,838
They don't officially call it license.

164
00:16:49,838 --> 00:16:52,910
They said that you need to conduct a security assessment.

165
00:16:52,910 --> 00:16:57,644
Basically, you need to go to CAC and then file registration and then

166
00:16:57,644 --> 00:17:02,838
get some sort of green light before you can introduce your services to the public.

167
00:17:02,838 --> 00:17:15,449
So in this way, company already voluntarily complied to conduct a self -censorship and
impose a lot of the filtering at the last stage of the development of the AI model to make

168
00:17:15,449 --> 00:17:26,337
sure the content generated will not generate problematic, harmful, or inappropriate
content.

169
00:17:27,014 --> 00:17:38,769
And that will definitely affect the user experience because very often time you put it in
a prom and then it tells you that they don't have the answer, this is not available.

170
00:17:38,769 --> 00:17:42,320
So definitely affect user experience.

171
00:17:42,320 --> 00:17:50,523
But let's also bear in mind that the government does not regulate enterprise use AI
application.

172
00:17:50,564 --> 00:17:56,962
In fact, interim measures of regulating generative AI only apply to public phasing
services.

173
00:17:56,962 --> 00:18:12,075
So if you are selling to a company like Huawei, ZTE, or TCL, or whatever, you're not
subject to this kind of censorship control.

174
00:18:12,075 --> 00:18:18,420
So to that extent, I don't think it necessarily hold back the development of the
technology itself.

175
00:18:18,420 --> 00:18:26,304
However, it definitely reduce user experience in a way that also make this product a bit
less

176
00:18:26,304 --> 00:18:28,395
interest attractive.

177
00:18:28,395 --> 00:18:38,708
But to be honest with you, all these farms in China are struggling to capital to profit
from this new genitive AI tools, right?

178
00:18:38,708 --> 00:18:45,970
Company like Baidu, like Huashou, they, everybody is offering this kind of tools, but
they're burning every money every day.

179
00:18:45,970 --> 00:18:54,262
To this day, none of them have come up with a sustainable business strategy planned.

180
00:18:54,262 --> 00:18:56,813
as to how I can make money on this thing.

181
00:18:56,813 --> 00:19:00,904
I mean, yes, OpenAI, can charge money fee.

182
00:19:00,984 --> 00:19:03,584
But in China, people get used to free stuff.

183
00:19:03,625 --> 00:19:06,045
And so there's a big problem.

184
00:19:07,566 --> 00:19:14,027
Although this generated a huge hype, but it's not really bringing in a lot of profits yet
for this company.

185
00:19:14,027 --> 00:19:16,628
It's still at the investment stage.

186
00:19:17,548 --> 00:19:23,030
And it's very expensive because before, for this consumer tech business,

187
00:19:24,078 --> 00:19:25,569
It's kind of like a fixed cost.

188
00:19:25,569 --> 00:19:27,259
You develop this and you can use it.

189
00:19:27,259 --> 00:19:35,021
But here, every time somebody put in an inquiry, you have to deploy your computing
resource, which is hugely expensive.

190
00:19:35,021 --> 00:19:42,744
And Chinese companies are really short of CPU at the moment to train the models, not to
say for deployment.

191
00:19:42,744 --> 00:19:50,018
So that, again, that limit on China's GPU resources, again, also

192
00:19:50,018 --> 00:19:52,981
kind of undermine user experience, right?

193
00:19:53,302 --> 00:19:59,109
mean, so because companies don't want you to access the most powerful and most advanced
model.

194
00:19:59,109 --> 00:20:00,480
And so, yeah.

195
00:20:01,323 --> 00:20:09,419
So you're talking about the US chips acts and other limits on Chinese access to the most
powerful GPUs.

196
00:20:09,419 --> 00:20:13,001
How much of an impact is that really having on the Chinese AIS?

197
00:20:13,964 --> 00:20:15,755
It definitely has a very big impact.

198
00:20:15,755 --> 00:20:24,177
mean, because this is really the biggest bottleneck on the development of Chinese AI
farms, especially the big tech companies.

199
00:20:24,177 --> 00:20:36,540
You probably have read in recent weeks some of the reports in Financial Times and also
Journal saying that, mean, this kind of expert control doesn't seem, the US expert control

200
00:20:36,540 --> 00:20:38,391
program doesn't seem to be very effective.

201
00:20:38,391 --> 00:20:42,962
Now it's even cheaper to buy CPU in China than in the United States because

202
00:20:43,086 --> 00:20:47,670
There are these kind of export restriction laws are full of loopholes.

203
00:20:47,670 --> 00:20:54,515
And by the way, it's not that difficult to smuggle Nivea's H100.

204
00:20:54,515 --> 00:20:56,156
It's not that big.

205
00:20:56,156 --> 00:21:05,143
There's a story that a lot of the Chinese students, everyone bring back H100.

206
00:21:08,686 --> 00:21:10,256
But this is just one of the ways.

207
00:21:10,256 --> 00:21:14,936
mean, I'm sure there are 100 million other ways that Chinese companies can get around it.

208
00:21:14,936 --> 00:21:22,516
But it will be hard for the Chinese big tech giants like Alibaba and Tencent because these
companies, these are big companies, right?

209
00:21:22,516 --> 00:21:26,206
mean, so they will put compliance.

210
00:21:26,466 --> 00:21:28,286
They will highly value compliance.

211
00:21:28,286 --> 00:21:30,326
They have business operation everywhere.

212
00:21:30,326 --> 00:21:33,726
So this company will definitely...

213
00:21:34,026 --> 00:21:35,217
suffer a great deal.

214
00:21:35,217 --> 00:21:46,733
And in fact, Alibaba, which operates China's leading cloud business now, want to cut back
some of the cloud access from small and medium sized companies because even they are

215
00:21:46,733 --> 00:21:48,794
running out of this capacity.

216
00:21:48,794 --> 00:21:52,436
And that definitely has an impact.

217
00:21:52,436 --> 00:21:58,599
Now, on the other hand, so in the short term, it seems the US program has worked.

218
00:21:58,599 --> 00:22:03,938
But in the medium and long run, it's also creating a big problem for the US.

219
00:22:03,938 --> 00:22:18,896
first because this kind of restriction, it's also very costly for the US in a way it's
counterproductive in a way because you are restricting your own companies to sell to a

220
00:22:18,896 --> 00:22:25,130
very lucrative Chinese market while your allies may not be restricting it.

221
00:22:25,130 --> 00:22:33,524
so in fact, a lot of the US toolmakers suffer a lot of, suffer tremendous loss because of
that.

222
00:22:34,118 --> 00:22:44,241
because companies in Japan and Korea, or Netherlands, they were not restricted in selling
to Chinese companies, while US companies are.

223
00:22:44,241 --> 00:22:51,703
And by the way, I also need to mention that I think the US got really lucky with the AI
boom.

224
00:22:51,783 --> 00:22:59,845
If the US did not experience this huge AI boom, this kind of extra restriction would
really hurt.

225
00:23:01,250 --> 00:23:03,451
the business operation of companies like Nvidia.

226
00:23:03,451 --> 00:23:10,955
If you may recall a few years ago, Nvidia even felt bankruptcy because of all these
restrictions.

227
00:23:10,955 --> 00:23:22,582
that's why, know, so a lot of the adverse impact of the US export restriction were not
really felt, to be honest, thanks to the AI boom.

228
00:23:22,582 --> 00:23:24,663
So if we consider that counterfactual.

229
00:23:24,663 --> 00:23:30,702
So first is hurting US business and in a way indirectly will harm US innovation because
you need

230
00:23:30,702 --> 00:23:34,322
the Chinese market, you need that profit in order to supply and fry.

231
00:23:34,322 --> 00:23:41,682
Just look at Intel, which recently cut, you know, suffered a huge loss in the equity
market.

232
00:23:41,682 --> 00:23:56,422
Because China accounted 35 % of its revenue and the US government want Intel to succeed in
US to build more factories in the US and develop more in the US, in a way cutting back the

233
00:23:56,422 --> 00:23:57,662
business in China.

234
00:23:57,662 --> 00:23:59,798
And the company really suffered.

235
00:23:59,798 --> 00:24:00,118
Right.

236
00:24:00,118 --> 00:24:11,566
mean, and I'm not saying that that was the single China was the single most important
contributing factor, but it definitely had played a role in hurting Intel's business.

237
00:24:11,566 --> 00:24:12,006
Right.

238
00:24:12,006 --> 00:24:18,911
But more importantly, above all, all these export restrictions is incentivizing China to
become more self -sufficient.

239
00:24:18,911 --> 00:24:28,618
And you are seeing Huawei, for instance, you know, at the beginning did suffer a loss, but
in the past couple of years, it's doing phenomenally well.

240
00:24:28,618 --> 00:24:29,898
And it's it's

241
00:24:30,054 --> 00:24:31,896
It's iPhone businesses back, right?

242
00:24:31,896 --> 00:24:43,690
I mean, and people were teasing while Huawei was releasing its new tripod, try screens, a
phone, vis -a -vis the Apple iPhone, right?

243
00:24:43,690 --> 00:24:51,808
mean, and it also closely aligned the public and the private incentives.

244
00:24:52,058 --> 00:24:57,820
because now companies like Alibaba and Tencent, these are the big tech companies, they
were not able to access Nivea chips.

245
00:24:57,820 --> 00:24:59,701
They have to buy it from Huawei, right?

246
00:24:59,701 --> 00:25:09,523
I mean, so in a way they are contributing to the profits of Huawei and that becomes like a
virtual cycle that continue to nurture China's development.

247
00:25:09,804 --> 00:25:17,240
Now, nobody knows at what point China will become really become self -sufficient and
because the go -go post

248
00:25:17,240 --> 00:25:21,302
keep moving because the US is also moving ahead very rapidly.

249
00:25:21,302 --> 00:25:23,735
This is something that we don't know.

250
00:25:23,735 --> 00:25:31,681
But from an academic standpoint, we are evaluating the trade -off between the short -term
benefits to the US and the long -term loss.

251
00:25:31,681 --> 00:25:34,081
So there's a lot of variables.

252
00:25:34,683 --> 00:25:44,610
But in a way, what we see the US is doing right now, we see very often a lot of these
export control and restrictions.

253
00:25:44,718 --> 00:25:51,471
They don't seem to be very well thought out and there's not a very clear feedback
mechanism to evaluate success.

254
00:25:51,471 --> 00:25:54,663
And I think this is something that we should query.

255
00:25:54,663 --> 00:26:01,185
Has the US government done enough to evaluate what it is doing and whether it's effective
or not?

256
00:26:03,233 --> 00:26:13,313
The other thing that is happening in the background is on the Chinese side, the shift in
the Chinese government's approach more generally to the economy since the pandemic.

257
00:26:13,313 --> 00:26:22,493
And there was a fairly shocking statistic in the Financial Times recently that there was
something like in 2018, there were 50 ,000 venture -backed startups created in China and

258
00:26:22,493 --> 00:26:25,273
2023 it was 1 ,200 and still going down.

259
00:26:25,273 --> 00:26:29,713
I want to ask you, first of all, is that accurate in terms of what you see in the market
there?

260
00:26:29,713 --> 00:26:32,729
And if so, what does that mean for the Chinese AI sector?

261
00:26:33,550 --> 00:26:33,930
Right.

262
00:26:33,930 --> 00:26:38,490
mean, that is definitely compatible with my research findings in my book.

263
00:26:38,490 --> 00:26:44,850
actually have one chapter specifically devoted to assessing the impact of China's tech
crackdown.

264
00:26:44,850 --> 00:26:57,050
Is that because of this kind of really heavy handed intervention during that 18 month law
enforcement campaign, it had really undermined investor confidence and stifled

265
00:26:57,050 --> 00:26:58,520
entrepreneurial spirit.

266
00:26:58,520 --> 00:27:03,270
And as a result, both the domestic and international businesses

267
00:27:03,270 --> 00:27:07,453
investors have retreated from China's consumer tech businesses.

268
00:27:07,453 --> 00:27:17,960
And that explains this dramatic drop of funding into China's tech sector.

269
00:27:17,960 --> 00:27:31,610
But on the other hand, a lot of funding were channeled into the high tech sector, the so
-called hardcore tech businesses like semiconductors, like EV, high manufacturing, et

270
00:27:31,610 --> 00:27:32,366
cetera.

271
00:27:32,366 --> 00:27:32,716
Right?

272
00:27:32,716 --> 00:27:40,906
I mean, but still overall, the whole macroeconomic environment in China in the past couple
of years was not very good.

273
00:27:40,906 --> 00:27:41,186
Right?

274
00:27:41,186 --> 00:27:46,946
I mean, the Chinese economy was doing phenomenally well in early 2021.

275
00:27:46,946 --> 00:27:57,246
If we recall, China was the first one to recover from COVID and the first one to resume a
production line where everybody else was suffering.

276
00:27:57,246 --> 00:28:00,730
But since then, everything has gone downhill.

277
00:28:00,734 --> 00:28:16,373
I mean, a series of policy missteps like the COVID pandemic control and this massive tech
crackdown, as well as the property crackdowns, right?

278
00:28:16,373 --> 00:28:20,985
I mean, has directly contributed to China's economic slowdown.

279
00:28:20,985 --> 00:28:29,590
And that definitely reduced the appetize of the foreign investors in Chinese tech sector.

280
00:28:30,923 --> 00:28:36,454
So let me ask you just one or two more questions and shift back to AI and regulation.

281
00:28:36,655 --> 00:28:45,297
You talked about that China could, if it wanted to adopt a comprehensive AI law, but
hasn't, but has some of these more narrow laws.

282
00:28:46,237 --> 00:29:00,101
How do you see the development of either regulatory coordination or at least regulatory
discussions across the world around AI, given what's happening in Europe and the US and

283
00:29:00,101 --> 00:29:01,061
other places?

284
00:29:02,670 --> 00:29:19,852
There's a lot of cooperation going on, but so far we see that the dialogue has remained at
quite a high level on agreeing on very basic principles like accountability, transparency,

285
00:29:19,852 --> 00:29:21,122
this kind of thing.

286
00:29:21,263 --> 00:29:32,170
We haven't yet seen very meaningful and significant and concrete proposals as to how to
ensure AI safety.

287
00:29:32,994 --> 00:29:45,618
And part of the reason is because, I think most important reason is because the two
biggest AI superpower, China and the US, are engaged in locking this intense tech rivalry.

288
00:29:46,418 --> 00:29:58,872
And how could you persuade China to really cooperate when at the same time the US is
doubling down on export restrictions of AI chips to China, right?

289
00:29:58,872 --> 00:29:59,842
mean, so.

290
00:30:00,382 --> 00:30:12,819
Even though you have seen the two sides at the highest level actually have engaged in a
couple of dialogues in San Francisco and then in Geneva, right?

291
00:30:12,819 --> 00:30:16,390
But they seems to be talking without really doing anything.

292
00:30:16,891 --> 00:30:27,477
So I would say lower your expectation of anything that will really generate from this AI
international cooperation.

293
00:30:27,477 --> 00:30:29,910
But I do think that it's extremely important.

294
00:30:29,910 --> 00:30:36,975
Most importantly, I think the two sides should exert more efforts in understanding the
risk of AI.

295
00:30:37,436 --> 00:30:45,741
Because so far, there is a huge controversy and everybody think that this is quite
speculative.

296
00:30:46,142 --> 00:30:57,410
So if the Chinese government come to realize that there are serious threat, because if
something is a serious threat to...

297
00:30:58,266 --> 00:31:02,147
its citizens and the society and the government.

298
00:31:02,147 --> 00:31:10,990
I'm pretty sure the Chinese government will deal with it very decisively, just like how it
has been able to manage COVID -19.

299
00:31:10,990 --> 00:31:15,351
Xi Jinping was even touting its zero COVID strategy.

300
00:31:15,831 --> 00:31:21,692
The big problem is nobody knows where these things will become an existential threat and
how likely it is.

301
00:31:21,813 --> 00:31:27,374
So think at least on that point, I think the US and China can at least cooperate.

302
00:31:27,465 --> 00:31:29,577
and engage more.

303
00:31:30,621 --> 00:31:35,834
And finally, what do you expect for the future development of AI in China in the next few
years?

304
00:31:35,834 --> 00:31:44,678
I know it's a broad question, as you've accurately pointed out, there's different things
that we talk about when we talk about AI, but what should people look to in the next few

305
00:31:44,678 --> 00:31:46,789
years in terms of the Chinese AI sector?

306
00:31:48,430 --> 00:31:53,733
China is actually really traveling very closely behind the US.

307
00:31:54,353 --> 00:31:59,176
Mark Zuckerberg was saying that Chinese companies are just four or five months behind us.

308
00:31:59,336 --> 00:32:05,009
And others say, look, I with the large language models, China is maybe a few years behind.

309
00:32:05,009 --> 00:32:10,042
But if we are talking about text to image, text to video, China is really on par.

310
00:32:10,042 --> 00:32:15,865
You saw recently, Huashou released this text to video app.

311
00:32:15,865 --> 00:32:17,265
It's really good.

312
00:32:18,018 --> 00:32:25,431
they recently, but they have to kind of, they can't really introduce it to the market
because it's too expensive to deploy this thing, right?

313
00:32:25,431 --> 00:32:27,572
Because they're running out of CPU, right?

314
00:32:27,572 --> 00:32:30,693
I mean, so let's watch this space.

315
00:32:30,693 --> 00:32:41,618
This is not like semiconductor where China needs decades of experience and support in
order to develop that sector.

316
00:32:41,618 --> 00:32:46,550
China already is hugely successful, particularly in the consumer.

317
00:32:46,550 --> 00:32:48,431
application side of AI.

318
00:32:48,431 --> 00:32:54,601
And that's why, you know, TikTok can take over the US market so quickly, right?

319
00:32:54,601 --> 00:32:56,993
I mean, this is a Chinese technology.

320
00:32:58,393 --> 00:33:05,155
But with regard to the hardware application, like robotics, for instance, China is also
very good, right?

321
00:33:05,155 --> 00:33:10,377
Facial recognition, voice recognition, but everything else China is still lagging behind.

322
00:33:10,377 --> 00:33:15,938
And query how much the semi -concept AI chips might be also holding

323
00:33:15,938 --> 00:33:21,331
back China's development, although at the same time, people are saying that these days
there's new technology that emerge.

324
00:33:21,331 --> 00:33:27,124
You don't really need that much chips computing power in order to train very powerful
large language models.

325
00:33:27,345 --> 00:33:33,198
So I would definitely see Chinese as a very serious rival to the United States.

326
00:33:33,198 --> 00:33:39,692
But at the same time, I would also highlight that there is also a lot of swing states that
are really catching up.

327
00:33:39,692 --> 00:33:43,820
mean, even UAE is producing really

328
00:33:43,820 --> 00:33:46,311
really good in cutting edge large language model.

329
00:33:46,311 --> 00:33:52,812
And again, this is a foreign -taran country that are able to mobilize tremendous resources
to develop this technology.

330
00:33:52,812 --> 00:33:55,533
And they have money, they have energy, right?

331
00:33:55,533 --> 00:34:00,675
mean, and then they don't suffer from export restrictions like China, right?

332
00:34:00,675 --> 00:34:07,557
And in fact, US companies want to cooperate with G42 from the UAE, unlike in China, right?

333
00:34:07,557 --> 00:34:13,108
I so there's other big players that should be under a close radar.

334
00:34:13,244 --> 00:34:15,304
and closely watch as well.

335
00:34:16,065 --> 00:34:18,096
Great, Angela, thank you so much for your time.

336
00:34:18,380 --> 00:34:18,963
My pleasure.

337
00:34:18,963 --> 00:34:20,287
Thanks for having me.