1
00:00:00,345 --> 00:00:02,834
We might be the last generation of white-collar workers.

2
00:00:03,104 --> 00:00:05,793
Doomsday news is coming out of Silicon Valley.

3
00:00:06,384 --> 00:00:06,814
Last year,

4
00:00:07,264 --> 00:00:09,544
in the code of top Silicon Valley labs,

5
00:00:09,544 --> 00:00:12,353
70% to 80% was still written by humans.

6
00:00:12,625 --> 00:00:13,013
This year,

7
00:00:13,345 --> 00:00:14,784
that number has become

8
00:00:14,785 --> 00:00:16,274
less than 1%.

9
00:00:16,664 --> 00:00:18,513
Not a 1% drop,

10
00:00:18,585 --> 00:00:20,553
but only 1% remains.

11
00:00:21,024 --> 00:00:22,074
Within a single year,

12
00:00:22,105 --> 00:00:25,474
human programmers went from the main force to mere reviewers.

13
00:00:25,745 --> 00:00:26,634
And now,

14
00:00:26,745 --> 00:00:29,634
even the ability to review is struggling to keep up.

15
00:00:30,024 --> 00:00:33,753
These are first-hand data points from two podcasts by Xixiang Tech,

16
00:00:33,945 --> 00:00:37,073
shared directly by Silicon Valley founders.

17
00:00:38,064 --> 00:00:40,393
What sends chills down my spine is this next sentence:

18
00:00:40,664 --> 00:00:43,073
The world's top AI researchers

19
00:00:43,185 --> 00:00:44,793
are worried themselves

20
00:00:45,024 --> 00:00:47,473
that they'll be out of a job in a year or two.

21
00:00:47,625 --> 00:00:48,054
They say

22
00:00:48,465 --> 00:00:49,793
the next year or two

23
00:00:49,905 --> 00:00:53,594
might be the only remaining window of work for our generation.

24
00:00:53,905 --> 00:00:54,253
Hello,

25
00:00:54,505 --> 00:00:54,973
everyone,

26
00:00:55,225 --> 00:00:56,274
I'm Wang Lijie.

27
00:00:56,545 --> 00:00:57,713
In today's video,

28
00:00:57,905 --> 00:00:59,834
I'm not just breaking down a news story,

29
00:01:00,024 --> 00:01:03,194
but a timeline that is changing everyone's destiny.

30
00:01:03,585 --> 00:01:06,233
From this Silicon Valley field report,

31
00:01:06,265 --> 00:01:08,394
I've extracted six sets of data and facts.

32
00:01:08,665 --> 00:01:10,793
On the surface, they seem unrelated,

33
00:01:10,825 --> 00:01:13,993
but when connected with a specific logic,

34
00:01:14,025 --> 00:01:16,754
I see a conclusion that makes me very uneasy.

35
00:01:16,944 --> 00:01:18,073
Our generation

36
00:01:18,105 --> 00:01:21,194
may very well be the last generation of white-collar workers in history.

37
00:01:21,424 --> 00:01:22,134
This isn't a figure of speech,

38
00:01:22,385 --> 00:01:23,634
nor is it alarmist nonsense;

39
00:01:23,704 --> 00:01:25,353
it is a reality happening now.

40
00:01:25,704 --> 00:01:28,194
Let's quickly go through these six sets of data,

41
00:01:28,344 --> 00:01:29,993
so you can feel the gravity yourself.

42
00:01:30,629 --> 00:01:31,429
Group 1:

43
00:01:31,825 --> 00:01:34,343
Frontier labs and top programmers

44
00:01:34,344 --> 00:01:36,114
have basically stopped writing code.

45
00:01:36,784 --> 00:01:40,153
Last year, 70% to 80% of code in the system was human-written;

46
00:01:40,224 --> 00:01:42,194
this year, it's less than 1%.

47
00:01:42,625 --> 00:01:43,694
In many tasks, Claude Code and

48
00:01:43,694 --> 00:01:46,134
Codex have already reached

49
00:01:46,185 --> 00:01:47,254
Meta's internal L8

50
00:01:47,825 --> 00:01:48,993
or L9 levels—

51
00:01:49,105 --> 00:01:52,114
the level of Chief Architects and CTOs.

52
00:01:52,584 --> 00:01:54,634
Several founders privately admitted,

53
00:01:54,864 --> 00:01:57,034
"Claude is better than my own CTO."

54
00:01:57,424 --> 00:01:59,353
Consider the resumes of these CTOs;

55
00:01:59,465 --> 00:02:03,114
they fought their way up from Stanford and MIT.

56
00:02:03,784 --> 00:02:05,674
It's like when the Go champion

57
00:02:05,704 --> 00:02:08,879
first saw AlphaGo play that famous 37th move.

58
00:02:08,879 --> 00:02:09,593
At that moment,

59
00:02:09,745 --> 00:02:11,614
the feeling wasn't "I lost,"

60
00:02:11,905 --> 00:02:12,633
but rather,

61
00:02:12,664 --> 00:02:14,234
"the game of Go as I understood it

62
00:02:14,305 --> 00:02:16,473
might not be what Go actually is."

63
00:02:17,088 --> 00:02:17,889
Group 2:

64
00:02:18,305 --> 00:02:19,674
Token consumption levels.

65
00:02:20,064 --> 00:02:21,593
Top Silicon Valley engineers

66
00:02:21,625 --> 00:02:23,794
consume hundreds of dollars in tokens daily,

67
00:02:23,865 --> 00:02:25,433
thousands per week.

68
00:02:25,784 --> 00:02:27,153
Big tech firms like Meta

69
00:02:27,224 --> 00:02:30,033
give employees unlimited Claude access.

70
00:02:30,145 --> 00:02:31,274
Not only is it unlimited,

71
00:02:31,344 --> 00:02:34,834
but they track and rank each employee's token usage,

72
00:02:34,944 --> 00:02:37,873
encouraging others to mimic the high consumers.

73
00:02:38,185 --> 00:02:40,313
Some founders even use interview questions like,

74
00:02:40,465 --> 00:02:41,743
"Here's $1,000;

75
00:02:41,745 --> 00:02:43,794
let's see how fast you can burn through these tokens."

76
00:02:44,025 --> 00:02:45,074
If you can burn them quickly,

77
00:02:45,104 --> 00:02:47,674
it shows you know how to maximize AI productivity.

78
00:02:47,985 --> 00:02:48,873
Group 3:

79
00:02:49,224 --> 00:02:50,234
Research speed.

80
00:02:50,585 --> 00:02:54,074
It used to take two to three weeks to go from an idea to working code.

81
00:02:54,224 --> 00:02:56,033
Now, one or two days is enough.

82
00:02:56,224 --> 00:02:59,753
Semi-automated or even fully automated experimental workflows

83
00:02:59,944 --> 00:03:02,913
have already emerged within Anthropic and OpenAI.

84
00:03:03,305 --> 00:03:06,873
Mathematical derivations and research breakthroughs at OpenAI

85
00:03:06,944 --> 00:03:09,680
are already emerging directly from dialogues

86
00:03:09,680 --> 00:03:10,633
with Codex and Claude.

87
00:03:10,745 --> 00:03:12,753
They aren't being thought up by human engineers.

88
00:03:12,905 --> 00:03:15,674
The data pipeline of a leading multimodal team

89
00:03:15,705 --> 00:03:18,593
shrank from twelve months to just a few days or a week.

90
00:03:19,344 --> 00:03:22,093
In over fifty working days, Anthropic released

91
00:03:22,104 --> 00:03:23,834
more than seventy products and features.

92
00:03:23,865 --> 00:03:24,834
This pace

93
00:03:24,865 --> 00:03:27,193
was completely impossible in the internet era.

94
00:03:27,384 --> 00:03:28,234
Group 4.

95
00:03:28,544 --> 00:03:29,873
The most heart-wrenching set of data.

96
00:03:30,224 --> 00:03:31,993
The top AI researchers,

97
00:03:32,025 --> 00:03:33,993
the very people who built all of this,

98
00:03:34,104 --> 00:03:37,713
are starting to worry they'll be unemployed in a year or two.

99
00:03:37,944 --> 00:03:39,033
Because they know

100
00:03:39,145 --> 00:03:41,993
the next generation of automated AI researchers is coming.

101
00:03:42,104 --> 00:03:44,514
These are systems capable of replacing their own jobs.

102
00:03:45,104 --> 00:03:48,354
Vitalbridge wrote in their 2026 internal forecast:

103
00:03:48,504 --> 00:03:51,193
"Our generation might be the last of the white-collar workers."

104
00:03:51,504 --> 00:03:53,713
It sounded too radical at the start of the year.

105
00:03:53,824 --> 00:03:56,193
A quarter later, looking back,

106
00:03:56,224 --> 00:03:57,783
they say this is no longer a prediction;

107
00:03:57,784 --> 00:03:59,433
it is a reality in progress.

108
00:04:00,145 --> 00:04:03,713
U.S. college grad employment hit a record low this year.

109
00:04:04,104 --> 00:04:06,794
Potentially 30% of jobs could disappear.

110
00:04:07,305 --> 00:04:09,354
Meta is still conducting massive layoffs.

111
00:04:09,824 --> 00:04:12,463
The growth and training paths for talent

112
00:04:12,465 --> 00:04:14,633
have been cut short by AI.

113
00:04:15,025 --> 00:04:15,834
Group 5.

114
00:04:16,064 --> 00:04:17,394
The macroeconomic level.

115
00:04:17,904 --> 00:04:21,074
Knowledge and intelligence are being compressed into models,

116
00:04:21,225 --> 00:04:23,394
becoming billable computing resources.

117
00:04:23,985 --> 00:04:28,553
You used to trade reading, learning, and experience for a job.

118
00:04:28,785 --> 00:04:32,434
Now, that intelligence is compressed into strings of tokens.

119
00:04:33,064 --> 00:04:35,634
The U.S. is a nation of over 100 million middle-class people.

120
00:04:35,824 --> 00:04:39,673
Coders, lawyers, doctors, agents, bankers—

121
00:04:39,785 --> 00:04:42,634
these happen to be the jobs AI is best at replacing.

122
00:04:43,144 --> 00:04:46,314
India's IT outsourcing industry may have entered a decline.

123
00:04:46,824 --> 00:04:49,754
Demand for many SaaS companies is falling

124
00:04:50,064 --> 00:04:51,314
because users find

125
00:04:51,384 --> 00:04:53,113
that Claude alone is enough.

126
00:04:53,425 --> 00:04:54,233
Group 6.

127
00:04:54,745 --> 00:04:56,113
My advice to you:

128
00:04:56,704 --> 00:04:59,033
AI replaces people who don't embrace it.

129
00:04:59,264 --> 00:05:01,834
Those who do embrace it will be the beneficiaries.

130
00:05:02,064 --> 00:05:04,874
This is like the wave of layoffs in 90s China,

131
00:05:05,024 --> 00:05:07,233
forcing capable people out of their comfort zones.

132
00:05:07,504 --> 00:05:10,233
AI drastically reduces the cost of starting a business.

133
00:05:10,264 --> 00:05:12,394
One-person companies can now explode in scale.

134
00:05:12,625 --> 00:05:14,384
The core value of humans in the future

135
00:05:14,384 --> 00:05:17,353
will return to creativity and aesthetic taste.

136
00:05:17,944 --> 00:05:19,093
That's all six sets of data.

137
00:05:19,425 --> 00:05:20,394
It's very dense.

138
00:05:20,824 --> 00:05:22,793
But if I just told you,

139
00:05:22,904 --> 00:05:24,353
"So go learn AI quickly,"

140
00:05:24,425 --> 00:05:27,194
I'd be no different from every other AI influencer.

141
00:05:27,464 --> 00:05:29,153
I want to take you one level deeper

142
00:05:29,264 --> 00:05:31,144
to see what is actually happening

143
00:05:31,144 --> 00:05:32,593
beneath these six sets of data.

144
00:05:32,744 --> 00:05:33,464
Here,

145
00:05:33,464 --> 00:05:36,514
I want to bring up a few concepts we've discussed before.

146
00:05:36,824 --> 00:05:38,994
The first is the Cantillon Effect.

147
00:05:39,504 --> 00:05:42,314
I've explained the Cantillon Effect in previous episodes.

148
00:05:42,504 --> 00:05:45,274
270 years ago, this economist discovered

149
00:05:45,305 --> 00:05:47,793
that when new money is injected into an economy,

150
00:05:48,024 --> 00:05:50,593
those closest to the injection point benefit the most.

151
00:05:50,745 --> 00:05:52,423
Those furthest from the injection point

152
00:05:52,425 --> 00:05:53,334
not only fail to benefit,

153
00:05:53,665 --> 00:05:55,793
but also bear the cost of inflation.

154
00:05:56,225 --> 00:05:58,713
This principle used to explain currency,

155
00:05:59,064 --> 00:05:59,694
but today,

156
00:06:00,024 --> 00:06:01,793
I'm giving you a new version:

157
00:06:01,904 --> 00:06:03,673
The Intelligence Cantillon Effect.

158
00:06:04,064 --> 00:06:04,374
You see,

159
00:06:04,865 --> 00:06:07,113
AI isn't something that appeared out of thin air.

160
00:06:07,425 --> 00:06:09,394
It is an "intelligence injection event."

161
00:06:09,665 --> 00:06:11,863
Every intellectual output in human history—

162
00:06:11,865 --> 00:06:15,874
every book, every line of code, every paper, every conversation—

163
00:06:15,904 --> 00:06:19,074
has been compressed into something called a Large Language Model.

164
00:06:19,185 --> 00:06:21,454
It has become a fluid, priceable resource,

165
00:06:21,785 --> 00:06:25,634
capable of being redistributed like water among different people.

166
00:06:26,064 --> 00:06:28,994
This is an unprecedented redistribution of intelligence.

167
00:06:29,504 --> 00:06:31,774
So, who is closest to this injection point?

168
00:06:32,745 --> 00:06:35,293
It's the researchers and engineers working inside

169
00:06:35,584 --> 00:06:38,194
Anthropic, OpenAI, and Meta.

170
00:06:38,305 --> 00:06:40,733
And the front-line Silicon Valley founders

171
00:06:40,745 --> 00:06:42,233
who burn hundreds of dollars in tokens every day.

172
00:06:42,504 --> 00:06:45,793
They aren't just getting 100% of average human intelligence;

173
00:06:45,944 --> 00:06:50,153
they're getting 1,000% or 10,000% in external 'plugin' intelligence.

174
00:06:50,504 --> 00:06:53,113
They finish three weeks of work in one day.

175
00:06:53,144 --> 00:06:54,983
It's not because they've become smarter;

176
00:06:54,985 --> 00:06:57,194
it's because they are standing right next to the injection point.

177
00:06:57,545 --> 00:06:59,334
Who is second closest to the injection point?

178
00:06:59,944 --> 00:07:02,394
Early adopters outside the San Francisco Bay Area

179
00:07:02,425 --> 00:07:06,113
who remain hyper-sensitive to AI.

180
00:07:06,345 --> 00:07:10,593
People like top-tier Chinese entrepreneurs, investors, and indie developers,

181
00:07:10,704 --> 00:07:12,274
including myself.

182
00:07:12,665 --> 00:07:14,793
We aren't at the injection point,

183
00:07:14,904 --> 00:07:16,553
but we are close enough

184
00:07:16,665 --> 00:07:19,634
to feel the direction of the flow immediately.

185
00:07:20,105 --> 00:07:21,853
Who is farthest from the injection point?

186
00:07:22,384 --> 00:07:25,024
The majority of white-collar workers who still work the way they did two years ago,

187
00:07:25,024 --> 00:07:28,894
unwilling or afraid to deeply embed AI into their workflows.

188
00:07:28,985 --> 00:07:30,153
It's not that they aren't smart;

189
00:07:30,425 --> 00:07:32,033
they just haven't realized yet.

190
00:07:32,064 --> 00:07:34,033
The current has already changed,

191
00:07:34,745 --> 00:07:36,194
but they are still standing in the old riverbed waiting for water.

192
00:07:36,225 --> 00:07:39,074
This is the first layer:

193
00:07:39,545 --> 00:07:40,454
the Intellectual Cantillon Effect.

194
00:07:40,865 --> 00:07:42,514
But let me pause here,

195
00:07:42,824 --> 00:07:44,514
and tear open something deeper for you to see.

196
00:07:44,545 --> 00:07:47,033
Because there is a second layer,

197
00:07:47,425 --> 00:07:48,533
a much more brutal one.

198
00:07:48,865 --> 00:07:50,673
In the traditional Cantillon Effect,

199
00:07:51,144 --> 00:07:52,834
inflation devalues the purchasing power of the cash in your hand.

200
00:07:52,904 --> 00:07:56,113
What you lose is the intermediary tool called money.

201
00:07:56,625 --> 00:07:59,673
But in the Intellectual Cantillon Effect,

202
00:08:00,024 --> 00:08:01,413
what is being devalued?

203
00:08:01,665 --> 00:08:02,574
It is devaluing

204
00:08:03,144 --> 00:08:03,894
you as a person.

205
00:08:04,225 --> 00:08:05,673
Think about it.

206
00:08:06,105 --> 00:08:07,314
How did society value a person in the past?

207
00:08:07,824 --> 00:08:09,853
It was based on the knowledge they mastered, the skills they possessed,

208
00:08:10,345 --> 00:08:13,254
and the cognitive tasks they could complete.

209
00:08:13,584 --> 00:08:15,473
A doctor's value

210
00:08:15,745 --> 00:08:17,103
came from studying for over a decade, rotating for tens of thousands of hours,

211
00:08:17,105 --> 00:08:20,334
and storing hundreds of thousands of medical decision trees in their head.

212
00:08:20,745 --> 00:08:23,473
A lawyer's value

213
00:08:23,745 --> 00:08:25,064
was remembering thousands of precedents

214
00:08:25,064 --> 00:08:26,574
and locating the right legal clause in seconds.

215
00:08:26,904 --> 00:08:29,593
A programmer's value

216
00:08:30,064 --> 00:08:31,473
was translating human business logic into machine language.

217
00:08:31,504 --> 00:08:34,793
What is the common foundation of these values?

218
00:08:35,144 --> 00:08:37,254
The scarcity of intelligence.

219
00:08:37,664 --> 00:08:39,234
You study for twenty years to become a doctor—

220
00:08:39,985 --> 00:08:42,553
those twenty years themselves are a massive barrier.

221
00:08:42,625 --> 00:08:45,234
This barrier keeps others out,

222
00:08:45,384 --> 00:08:47,793
which is why you, as a doctor, can command a high income.

223
00:08:47,904 --> 00:08:50,754
You can understand this as

224
00:08:50,904 --> 00:08:52,504
society setting a price for the scarce resource of intelligence.

225
00:08:52,504 --> 00:08:56,073
Whether that price is high

226
00:08:56,184 --> 00:08:57,293
depends on how scarce that intelligence is.

227
00:08:57,625 --> 00:08:59,673
But what is happening now?

228
00:09:00,225 --> 00:09:01,494
Intelligence is no longer scarce.

229
00:09:02,144 --> 00:09:03,874
For a few dozen dollars a month,

230
00:09:04,184 --> 00:09:05,673
you can access an intelligent system stronger than most professional lawyers,

231
00:09:05,705 --> 00:09:08,134
doctors, or programmers.

232
00:09:08,424 --> 00:09:11,714
For a few hundred dollars a month,

233
00:09:11,985 --> 00:09:13,354
you can access a system stronger than a Meta L8 Principal Architect.

234
00:09:13,384 --> 00:09:16,533
The barrier you spent twenty years building—

235
00:09:16,784 --> 00:09:17,913
AI didn't just climb over it.

236
00:09:18,345 --> 00:09:21,114
AI leveled it.

237
00:09:21,184 --> 00:09:22,823
It turned a cliff into flat ground.

238
00:09:22,825 --> 00:09:24,053
So, what's being devalued isn't your money.

239
00:09:24,424 --> 00:09:26,433
[MISSING]

240
00:09:26,825 --> 00:09:29,153
[MISSING]

241
00:09:29,345 --> 00:09:33,714
What's devaluing is the very foundation of your pricing as a knowledge worker.

242
00:09:34,465 --> 00:09:37,553
Sixing says this might be the last generation of white-collar workers.

243
00:09:37,745 --> 00:09:39,433
It's not that white-collar workers will be unemployed;

244
00:09:39,745 --> 00:09:41,514
unemployment is just a surface-level symptom.

245
00:09:41,664 --> 00:09:42,813
The deepest layer is that

246
00:09:43,105 --> 00:09:44,793
the social role of 'white-collar'—

247
00:09:44,825 --> 00:09:48,313
this identity category established in the industrial age—

248
00:09:48,424 --> 00:09:51,394
has had its value foundation hollowed out from the bottom.

249
00:09:51,705 --> 00:09:54,274
This is the second concept we've discussed on this channel.

250
00:09:54,384 --> 00:09:56,433
In the trillion-parameter episode, I said

251
00:09:56,625 --> 00:09:58,553
parameters are not knowledge itself,

252
00:09:58,664 --> 00:10:01,193
but the geometric shape of knowledge relationships.

253
00:10:01,384 --> 00:10:03,234
'Compression is intelligence.'

254
00:10:03,544 --> 00:10:05,913
Now, this compression has taken shape.

255
00:10:06,144 --> 00:10:08,014
What happens once it takes shape?

256
00:10:08,625 --> 00:10:11,994
Formed intelligence is packaged into an API,

257
00:10:12,105 --> 00:10:14,913
becoming an infrastructure like electricity.

258
00:10:15,264 --> 00:10:18,433
Think about it—the same thing happened in history.

259
00:10:18,825 --> 00:10:20,144
150 years ago,

260
00:10:20,144 --> 00:10:21,974
what was humanity's most valuable skill?

261
00:10:22,465 --> 00:10:23,394
Physical strength.

262
00:10:23,585 --> 00:10:26,254
How much weight a strong man could carry,

263
00:10:26,504 --> 00:10:30,274
how heavy a millstone he could push, or how long he could farm continuously—

264
00:10:30,384 --> 00:10:32,673
these determined his social value.

265
00:10:32,865 --> 00:10:34,214
Then the steam engine arrived.

266
00:10:34,504 --> 00:10:36,024
Then the internal combustion engine,

267
00:10:36,024 --> 00:10:37,913
and later the electric motor.

268
00:10:38,184 --> 00:10:41,114
The resource of strength was compressed into machines,

269
00:10:41,184 --> 00:10:43,673
turning into electricity priced by the kilowatt-hour.

270
00:10:43,985 --> 00:10:45,333
Those who relied on physical labor for a living

271
00:10:45,664 --> 00:10:46,703
didn't disappear,

272
00:10:46,705 --> 00:10:48,114
they were forced to pivot.

273
00:10:48,424 --> 00:10:50,793
They either became workers operating the machines,

274
00:10:50,825 --> 00:10:54,193
or were squeezed into margin areas where machines couldn't replace them.

275
00:10:54,504 --> 00:10:57,754
What's happening today is structurally identical.

276
00:10:58,105 --> 00:11:00,423
Except this time, it's not physical strength being compressed;

277
00:11:00,424 --> 00:11:01,313
it's brainpower.

278
00:11:01,544 --> 00:11:03,464
Middle-class white-collar workers who live by their wits

279
00:11:03,465 --> 00:11:06,813
are in the exact same position as the physical laborers

280
00:11:06,904 --> 00:11:08,153
150 years ago.

281
00:11:08,825 --> 00:11:11,553
You either become a new worker operating AI,

282
00:11:11,625 --> 00:11:14,953
or get squeezed into the margins where AI can't replace you.

283
00:11:15,465 --> 00:11:15,813
And,

284
00:11:16,144 --> 00:11:17,504
the speed of this displacement

285
00:11:17,504 --> 00:11:19,234
is more than ten times faster than last time.

286
00:11:19,465 --> 00:11:21,394
The 19th-century Industrial Revolution,

287
00:11:21,465 --> 00:11:23,663
from steam engines to full replacement,

288
00:11:23,664 --> 00:11:25,153
took nearly a hundred years.

289
00:11:25,625 --> 00:11:27,874
That's long enough for generations to adjust slowly.

290
00:11:28,304 --> 00:11:30,024
In today's AI revolution,

291
00:11:30,024 --> 00:11:31,293
the data we see is that

292
00:11:31,585 --> 00:11:32,433
within a year,

293
00:11:32,465 --> 00:11:35,553
Silicon Valley engineers went from the driving force to bystanders.

294
00:11:35,985 --> 00:11:37,374
This isn't on a century scale;

295
00:11:37,664 --> 00:11:38,833
it's on a yearly scale.

296
00:11:39,184 --> 00:11:40,153
At this speed,

297
00:11:40,264 --> 00:11:43,313
social systems simply cannot keep up.

298
00:11:43,825 --> 00:11:45,234
The education system can't keep up.

299
00:11:45,504 --> 00:11:47,553
You study Computer Science for four years,

300
00:11:47,585 --> 00:11:51,073
only to graduate and find the market no longer needs what you learned.

301
00:11:51,745 --> 00:11:53,193
Social security can't keep up.

302
00:11:53,264 --> 00:11:55,354
Unemployment insurance and vocational training—

303
00:11:55,424 --> 00:11:58,274
these were designed for the rhythm of the industrial age.

304
00:11:58,345 --> 00:12:01,553
They can't catch an employment structure collapsing in a single year.

305
00:12:01,904 --> 00:12:02,254
Here,

306
00:12:02,585 --> 00:12:04,153
I want to tell a short story.

307
00:12:04,384 --> 00:12:06,874
I know a CTO I really admire.

308
00:12:07,065 --> 00:12:08,474
His resume is flawless:

309
00:12:08,544 --> 00:12:10,234
PhD in CS from Stanford,

310
00:12:10,345 --> 00:12:12,994
Head of Tech for a unicorn startup,

311
00:12:13,105 --> 00:12:14,874
high six-figure salary,

312
00:12:14,945 --> 00:12:17,033
managing an engineering team of dozens.

313
00:12:17,384 --> 00:12:20,913
I used to see him as the template for the intellectual class—

314
00:12:21,184 --> 00:12:25,313
the upper limit of what humans can achieve through education and hard work.

315
00:12:25,705 --> 00:12:28,193
Recently, he spoke privately with an investor

316
00:12:28,264 --> 00:12:29,193
and said one thing.

317
00:12:29,424 --> 00:12:29,774
He said,

318
00:12:30,184 --> 00:12:32,144
'The core of my daily work now

319
00:12:32,144 --> 00:12:33,793
is no longer writing code;

320
00:12:33,865 --> 00:12:35,663
it's breaking down tasks for Claude.'

321
00:12:35,664 --> 00:12:38,313
Then I review the code it writes.

322
00:12:38,985 --> 00:12:40,264
But the speed of reviewing

323
00:12:40,264 --> 00:12:42,913
can hardly keep up with the speed at which Claude generates code.

324
00:12:42,985 --> 00:12:44,073
Think about that.

325
00:12:44,225 --> 00:12:46,823
He's not saying AI is assisting his work;

326
00:12:46,825 --> 00:12:47,703
he's saying

327
00:12:47,705 --> 00:12:52,114
his job has shifted from creating to chasing the speed of AI's creation.

328
00:12:52,264 --> 00:12:54,663
His value no longer comes from what he knows,

329
00:12:54,664 --> 00:12:58,313
but from how fast he can verify AI's output.

330
00:12:58,865 --> 00:12:59,453
And now,

331
00:12:59,664 --> 00:13:01,874
even verification is falling behind.

332
00:13:02,465 --> 00:13:04,474
What he said next was even more heart-wrenching.

333
00:13:04,945 --> 00:13:05,293
He said,

334
00:13:05,705 --> 00:13:06,984
"I'm starting to wonder:

335
00:13:06,985 --> 00:13:11,783
My PhD, all the coding experience I've built over 15 years—

336
00:13:11,784 --> 00:13:13,173
how much is it all still worth?

337
00:13:14,345 --> 00:13:16,514
If a junior engineer reporting to me,

338
00:13:16,585 --> 00:13:17,714
equipped with Claude,

339
00:13:17,784 --> 00:13:21,114
can finish in two days what used to take me two weeks,

340
00:13:21,384 --> 00:13:24,653
what's my value as a CTO?

341
00:13:25,105 --> 00:13:25,974
My management experience?

342
00:13:26,625 --> 00:13:26,974
But

343
00:13:27,345 --> 00:13:29,754
AI is also learning how to manage AI."

344
00:13:30,225 --> 00:13:32,553
This is where that 'existential dread' comes from.

345
00:13:32,904 --> 00:13:34,673
It's not just 'will I be replaced?'

346
00:13:34,825 --> 00:13:36,793
It's 'is everything I've accumulated in the past

347
00:13:36,904 --> 00:13:39,153
being systematically reset to zero?'

348
00:13:39,465 --> 00:13:40,214
You might say,

349
00:13:40,504 --> 00:13:42,594
"Leo, you're being too pessimistic.

350
00:13:42,705 --> 00:13:45,533
Doesn't AI bring growth and new jobs?"

351
00:13:45,945 --> 00:13:48,354
Let me break down this common logic for you.

352
00:13:48,664 --> 00:13:50,673
The argument usually goes like this:

353
00:13:50,705 --> 00:13:53,833
Every technological revolution in history has created new jobs.

354
00:13:53,865 --> 00:13:55,224
Old ones vanish,

355
00:13:55,225 --> 00:13:56,374
but new ones emerge.

356
00:13:56,664 --> 00:13:58,193
Overall, it's progress.

357
00:13:58,225 --> 00:13:59,453
So this time will be the same;

358
00:13:59,664 --> 00:14:00,673
no need to panic.

359
00:14:00,865 --> 00:14:02,494
This logic sounds very reasonable,

360
00:14:02,625 --> 00:14:02,974
right?

361
00:14:03,345 --> 00:14:05,474
It's based on historical induction.

362
00:14:05,504 --> 00:14:06,943
It happened every time before,

363
00:14:06,945 --> 00:14:08,714
so it will happen again.

364
00:14:09,504 --> 00:14:11,833
But there's an equivocation here,

365
00:14:11,985 --> 00:14:13,833
and I want to point it out precisely.

366
00:14:14,184 --> 00:14:15,783
In past revolutions,

367
00:14:15,784 --> 00:14:18,063
machines replaced manual labor.

368
00:14:18,065 --> 00:14:19,504
The new jobs created

369
00:14:19,504 --> 00:14:21,514
were higher-level cognitive roles.

370
00:14:21,985 --> 00:14:23,744
Steam engines replaced porters

371
00:14:23,745 --> 00:14:25,793
but created engineers.

372
00:14:26,024 --> 00:14:27,864
Automation replaced assembly line workers

373
00:14:27,865 --> 00:14:29,313
but created programmers.

374
00:14:29,945 --> 00:14:31,073
Notice the pattern:

375
00:14:31,424 --> 00:14:33,673
machines replace the lower levels,

376
00:14:33,705 --> 00:14:35,634
and humans migrate to higher levels.

377
00:14:35,904 --> 00:14:37,144
Humanity as a whole

378
00:14:37,144 --> 00:14:40,193
has been migrating up a climbing ladder.

379
00:14:40,504 --> 00:14:41,754
But this time is different.

380
00:14:42,024 --> 00:14:43,384
This time, AI is not replacing

381
00:14:43,384 --> 00:14:45,153
a certain level of labor;

382
00:14:45,225 --> 00:14:46,994
it's replacing the ladder itself.

383
00:14:47,945 --> 00:14:50,553
This ladder from lower to higher levels—

384
00:14:50,585 --> 00:14:51,714
on the human side—

385
00:14:51,745 --> 00:14:56,714
the top tiers are abstract thinking, creative problem-solving, and complex judgment.

386
00:14:56,784 --> 00:14:57,673
And these

387
00:14:57,784 --> 00:15:00,634
are exactly what LLMs are best at.

388
00:15:00,865 --> 00:15:01,734
It's not their weakness;

389
00:15:01,985 --> 00:15:03,234
it's their strength.

390
00:15:03,625 --> 00:15:06,114
As we discussed in the Moravec's Paradox episode,

391
00:15:06,384 --> 00:15:09,234
AI's capabilities are the inverse of humans'.

392
00:15:09,504 --> 00:15:10,874
Things humans find hard—

393
00:15:11,105 --> 00:15:13,553
playing Go, solving math Olympiads, writing code—

394
00:15:13,585 --> 00:15:14,793
AI finds easy.

395
00:15:15,105 --> 00:15:16,514
Things humans find easy—

396
00:15:16,705 --> 00:15:19,014
picking up a pencil, walking on uneven ground,

397
00:15:19,424 --> 00:15:20,953
detecting micro-expressions—

398
00:15:21,105 --> 00:15:22,634
AI finds incredibly difficult.

399
00:15:23,065 --> 00:15:26,433
So when AI starts striking from the top of the ladder,

400
00:15:26,585 --> 00:15:27,693
where are you going to migrate?

401
00:15:28,168 --> 00:15:28,614
Upward?

402
00:15:29,264 --> 00:15:31,354
AI has already occupied the top positions.

403
00:15:31,548 --> 00:15:32,014
Downward?

404
00:15:33,184 --> 00:15:34,953
Lower positions are manual labor,

405
00:15:35,024 --> 00:15:37,953
but AI robotics is also advancing,

406
00:15:38,144 --> 00:15:39,953
blocking that path too.

407
00:15:40,184 --> 00:15:42,793
This is why historical induction fails.

408
00:15:42,985 --> 00:15:45,394
It's not just a cliché that "this time is different."

409
00:15:45,745 --> 00:15:49,714
It's because the displacement is happening in the opposite direction.

410
00:15:49,985 --> 00:15:51,693
In the past, jobs were replaced from bottom to top;

411
00:15:52,065 --> 00:15:53,514
people could move upward.

412
00:15:53,945 --> 00:15:55,734
This time, it's being replaced from top to bottom.

413
00:15:56,065 --> 00:15:57,594
There is nowhere to run.

414
00:15:58,024 --> 00:16:01,360
That's why I told ValueVerse we might be the last generation

415
00:16:01,360 --> 00:16:02,673
of white-collar workers.

416
00:16:02,705 --> 00:16:04,313
This isn't an emotional panic;

417
00:16:04,424 --> 00:16:06,913
it's based on structural analysis.

418
00:16:07,184 --> 00:16:10,474
The value foundation of white-collar roles is being hollowed out.

419
00:16:10,745 --> 00:16:11,663
And this time,

420
00:16:11,664 --> 00:16:14,634
no higher-level roles are waiting to absorb us.

421
00:16:14,945 --> 00:16:15,653
You might ask,

422
00:16:15,985 --> 00:16:17,374
"What about creativity and aesthetics?"

423
00:16:17,825 --> 00:16:20,014
Aren't those the final bastions of humanity?

424
00:16:20,544 --> 00:16:22,594
Let me pour cold water on that too.

425
00:16:23,024 --> 00:16:24,693
Creativity and aesthetics

426
00:16:25,065 --> 00:16:26,433
sound wonderful,

427
00:16:26,504 --> 00:16:28,394
but they have one fatal flaw:

428
00:16:28,745 --> 00:16:29,693
Demand for them

429
00:16:29,985 --> 00:16:31,553
is very limited.

430
00:16:31,985 --> 00:16:33,114
Think about it.

431
00:16:33,384 --> 00:16:37,153
Modern society supports over 100 million middle-class people

432
00:16:37,264 --> 00:16:41,274
because there's massive demand for medium-complexity intellectual labor.

433
00:16:41,544 --> 00:16:46,754
Contracts, coding, accounting, diagnosis, design, analysis—

434
00:16:46,865 --> 00:16:48,614
these tasks occur in every company,

435
00:16:48,865 --> 00:16:51,673
in every industry, every day.

436
00:16:51,865 --> 00:16:54,354
The demand is measured in the hundreds of millions.

437
00:16:55,065 --> 00:16:57,183
But at the level of creativity and aesthetics,

438
00:16:57,184 --> 00:16:58,734
what is the scale of demand?

439
00:16:59,424 --> 00:17:00,903
For a hit movie,

440
00:17:00,904 --> 00:17:03,714
tens of millions of people consume a single creative output.

441
00:17:03,985 --> 00:17:05,513
For a viral song,

442
00:17:05,545 --> 00:17:08,273
hundreds of millions consume one creative act.

443
00:17:08,744 --> 00:17:10,624
Labor at the creative level

444
00:17:10,625 --> 00:17:12,634
is a case of "winner-takes-all."

445
00:17:13,025 --> 00:17:14,753
The top 1%

446
00:17:14,785 --> 00:17:17,074
will capture 90% of the income.

447
00:17:17,545 --> 00:17:19,354
The middle 20%

448
00:17:19,424 --> 00:17:21,114
might barely survive,

449
00:17:21,625 --> 00:17:23,463
while the remaining 80%

450
00:17:23,464 --> 00:17:25,193
can't even get into the game.

451
00:17:25,704 --> 00:17:29,993
So, saying human value lies in creativity and aesthetics

452
00:17:30,184 --> 00:17:32,153
sounds like it's opening a door,

453
00:17:32,424 --> 00:17:34,953
but it's actually opening a very narrow gate.

454
00:17:35,305 --> 00:17:36,824
The number of people it can sustain

455
00:17:36,825 --> 00:17:40,473
is orders of magnitude smaller than the current middle class.

456
00:17:40,825 --> 00:17:42,223
This isn't pessimism;

457
00:17:42,224 --> 00:17:44,273
it's an economic reality.

458
00:17:44,704 --> 00:17:45,013
So,

459
00:17:45,305 --> 00:17:47,263
in this structural dilemma,

460
00:17:47,265 --> 00:17:48,253
as an ordinary person,

461
00:17:48,464 --> 00:17:49,614
what should we actually do?

462
00:17:50,305 --> 00:17:52,513
I think we should look at it in three layers.

463
00:17:52,944 --> 00:17:53,653
The first layer,

464
00:17:53,904 --> 00:17:55,233
and the most urgent:

465
00:17:55,424 --> 00:17:56,374
Actively embrace AI.

466
00:17:56,664 --> 00:17:58,634
Move yourself toward the injection points.

467
00:17:58,984 --> 00:18:00,594
This is ValueVerse's advice,

468
00:18:00,625 --> 00:18:03,913
and it's the core action of the Cantillon Effect I've been discussing.

469
00:18:04,144 --> 00:18:05,874
AI doesn't replace white-collar workers;

470
00:18:05,944 --> 00:18:08,594
it replaces those who don't embrace AI.

471
00:18:08,744 --> 00:18:12,473
By embracing AI and embedding it in your workflow,

472
00:18:12,545 --> 00:18:14,344
you move from being the one replaced

473
00:18:14,345 --> 00:18:16,273
to the one replacing others.

474
00:18:16,984 --> 00:18:18,334
This is the basic fundamental step.

475
00:18:18,625 --> 00:18:19,384
Do it today,

476
00:18:19,384 --> 00:18:20,314
or it'll be too late tomorrow.

477
00:18:20,704 --> 00:18:22,913
But in all honesty,

478
00:18:23,384 --> 00:18:24,814
doing this only buys you time.

479
00:18:25,144 --> 00:18:26,233
It doesn't solve the problem,

480
00:18:26,744 --> 00:18:28,794
because AI is also embracing AI.

481
00:18:28,984 --> 00:18:32,074
The next generation of AI researchers are writing the next generation of AI.

482
00:18:32,464 --> 00:18:34,304
Everything you do, AI is doing too.

483
00:18:34,305 --> 00:18:35,634
And it's doing it faster than you.

484
00:18:36,065 --> 00:18:37,784
So, embracing AI...

485
00:18:37,785 --> 00:18:39,834
at most buys you a few years before being replaced.

486
00:18:39,984 --> 00:18:42,473
It won't keep you on the safe side forever.

487
00:18:42,984 --> 00:18:43,733
Now, the second level:

488
00:18:44,105 --> 00:18:45,344
something more fundamental.

489
00:18:45,345 --> 00:18:47,673
Rethink your relationship with work.

490
00:18:47,944 --> 00:18:51,074
Xiang said this may be the last generation of white-collar workers.

491
00:18:51,305 --> 00:18:52,463
Looking at it from another perspective,

492
00:18:52,464 --> 00:18:54,753
this judgment can also be a liberation.

493
00:18:55,025 --> 00:18:57,733
Our generation was indoctrinated with a life script:

494
00:18:58,065 --> 00:19:02,473
study hard, find a good job, stay stable, get promoted, and retire.

495
00:19:02,664 --> 00:19:04,584
The underlying assumption of this script is that...

496
00:19:04,585 --> 00:19:05,534
your intellectual labor...

497
00:19:05,865 --> 00:19:08,104
will be consistently needed by society,

498
00:19:08,105 --> 00:19:09,953
and consistently priced.

499
00:19:10,184 --> 00:19:10,534
Now,

500
00:19:10,825 --> 00:19:12,834
this assumption is collapsing.

501
00:19:13,144 --> 00:19:14,874
Collapse isn't necessarily a bad thing.

502
00:19:15,065 --> 00:19:15,614
Think about it:

503
00:19:15,865 --> 00:19:17,534
is this script itself even reasonable?

504
00:19:18,384 --> 00:19:20,554
A person spends 20 years studying,

505
00:19:20,585 --> 00:19:24,733
then 40 years in a cubicle repeating tasks AI can finish...

506
00:19:24,744 --> 00:19:26,064
in a single second,

507
00:19:26,065 --> 00:19:27,034
and finally retires.

508
00:19:27,545 --> 00:19:30,463
This script is a product of the Industrial Age.

509
00:19:30,464 --> 00:19:34,394
It's a compromise our species made for survival in that era;

510
00:19:34,505 --> 00:19:36,433
it's not some eternal truth.

511
00:19:36,904 --> 00:19:39,913
AI is forcing us to face a more fundamental question ahead of time:

512
00:19:40,265 --> 00:19:42,314
In an era where intelligence is not scarce,

513
00:19:42,384 --> 00:19:44,173
what do we actually live for?

514
00:19:44,664 --> 00:19:45,894
What is the meaning...

515
00:19:46,184 --> 00:19:47,013
of a person's existence?

516
00:19:47,545 --> 00:19:50,114
These questions were masked by work in the past.

517
00:19:50,184 --> 00:19:51,384
Because you had a job,

518
00:19:51,384 --> 00:19:52,874
you didn't have to think about them.

519
00:19:53,105 --> 00:19:55,993
Now AI has stripped away the fig leaf of "work."

520
00:19:56,009 --> 00:19:56,808
These questions...

521
00:19:56,984 --> 00:19:58,433
are being slammed right in your face.

522
00:19:58,825 --> 00:20:00,354
I don't think this is bad.

523
00:20:00,664 --> 00:20:02,743
I see it as a forced,

524
00:20:02,744 --> 00:20:04,544
but potentially beneficial,

525
00:20:04,545 --> 00:20:05,713
re-examination.

526
00:20:06,184 --> 00:20:07,384
Third level:

527
00:20:07,384 --> 00:20:09,263
something I've been thinking about lately.

528
00:20:09,265 --> 00:20:10,114
Going inward.

529
00:20:10,424 --> 00:20:12,193
We’ve talked many times on this channel...

530
00:20:12,305 --> 00:20:14,074
about the relationship between 'Yuanshen' and 'Shishen'.

531
00:20:14,384 --> 00:20:18,354
Shishen is the chatterbox analytical machine in your cerebral cortex.

532
00:20:18,625 --> 00:20:21,834
Yuanshen is the quiet observer behind that machine.

533
00:20:22,184 --> 00:20:23,453
Civilization over the past centuries...

534
00:20:23,825 --> 00:20:25,153
has been a civilization of 'Shishen'.

535
00:20:25,384 --> 00:20:28,874
We reward those good at analysis, calculation, and reasoning.

536
00:20:28,984 --> 00:20:31,273
We crowned 'Shishen' as the king of intelligence.

537
00:20:31,545 --> 00:20:32,913
But this 'Shishen'...

538
00:20:33,065 --> 00:20:35,594
is exactly what AI is best at simulating.

539
00:20:35,984 --> 00:20:38,104
The underlying logic of Large Language Models is:

540
00:20:38,105 --> 00:20:41,193
receive input, call parameters, generate output.

541
00:20:41,984 --> 00:20:44,594
This is exactly...

542
00:20:44,664 --> 00:20:45,794
what your default mode network does.

543
00:20:46,184 --> 00:20:46,973
The emergence of AI...

544
00:20:47,345 --> 00:20:49,513
is telling us, in a way,

545
00:20:49,545 --> 00:20:52,554
that the intelligence we've revered for centuries...

546
00:20:52,585 --> 00:20:54,753
might never have been the full picture of intelligence.

547
00:20:55,025 --> 00:20:56,913
It is just a subset;

548
00:20:56,984 --> 00:20:59,913
and the subset most easily replicated by machines.

549
00:21:00,384 --> 00:21:03,534
So where are the parts that are not easily replicated?

550
00:21:04,144 --> 00:21:04,973
They are in 'Yuanshen'.

551
00:21:05,305 --> 00:21:06,173
In awareness.

552
00:21:06,505 --> 00:21:08,034
In direct experience.

553
00:21:08,065 --> 00:21:09,834
In creative intuition.

554
00:21:09,904 --> 00:21:11,794
In the deeper dimensions of consciousness.

555
00:21:11,984 --> 00:21:13,913
These things cannot be tokenized.

556
00:21:13,944 --> 00:21:15,703
They can't be compressed into a model.

557
00:21:15,704 --> 00:21:17,354
They can't be called via API.

558
00:21:17,785 --> 00:21:18,653
They are alive,

559
00:21:18,944 --> 00:21:20,433
not just a combination of rules.

560
00:21:20,984 --> 00:21:21,973
I feel more and more...

561
00:21:22,224 --> 00:21:23,584
The impact of AI

562
00:21:23,585 --> 00:21:26,554
may have an underlying purpose we haven't yet discerned.

563
00:21:26,785 --> 00:21:29,834
It takes over everything the 'Conscious Mind' can do

564
00:21:29,903 --> 00:21:30,664
in order to

565
00:21:30,664 --> 00:21:31,693
force you to discover

566
00:21:31,984 --> 00:21:33,634
that you are not just an ego-bound mind.

567
00:21:33,865 --> 00:21:36,334
It forces you to recognize that which lies behind the ego,

568
00:21:36,625 --> 00:21:39,034
the more fundamental 'you' that has always existed.

569
00:21:39,345 --> 00:21:42,594
This direction isn't suitable as advice for everyone,

570
00:21:42,625 --> 00:21:45,874
because not everyone is ready to go this way.

571
00:21:46,224 --> 00:21:48,354
But I want to propose this possibility,

572
00:21:49,105 --> 00:21:51,433
because I am on this path myself.

573
00:21:51,545 --> 00:21:55,074
I believe this might be the most worthwhile path in the AI era.

574
00:21:55,124 --> 00:21:55,865
At the very least,

575
00:21:55,865 --> 00:21:58,874
it's the path that can bring you a lasting sense of stability.

576
00:21:59,025 --> 00:21:59,614
On that note,

577
00:21:59,704 --> 00:22:00,824
I'd like to pause here

578
00:22:00,825 --> 00:22:02,634
and invite you to leave a comment.

579
00:22:02,904 --> 00:22:05,624
From this firsthand report from Silicon Valley,

580
00:22:05,625 --> 00:22:06,733
what do you feel?

581
00:22:07,265 --> 00:22:08,013
Is it a sense of urgency?

582
00:22:08,384 --> 00:22:09,013
Anxiety?

583
00:22:09,424 --> 00:22:10,173
Or liberation?

584
00:22:10,984 --> 00:22:12,473
In the work you do now,

585
00:22:12,505 --> 00:22:15,213
do you think you'll be replaced by AI within five years?

586
00:22:15,744 --> 00:22:16,253
If so,

587
00:22:16,505 --> 00:22:17,334
what do you plan to do?

588
00:22:17,825 --> 00:22:18,453
If not,

589
00:22:18,744 --> 00:22:20,294
where does your confidence come from?

590
00:22:20,825 --> 00:22:21,173
Also,

591
00:22:21,464 --> 00:22:22,753
for the next video,

592
00:22:22,904 --> 00:22:24,213
what would you like me to talk about?

593
00:22:25,265 --> 00:22:28,114
Dig deeper into AI's impact on specific industries?

594
00:22:28,184 --> 00:22:32,034
Or how to build AI-proof skills in this era?

595
00:22:32,184 --> 00:22:35,104
Or perhaps how the 'Divine Spirit' awakens

596
00:22:35,105 --> 00:22:36,453
once the 'Conscious Mind' is taken over by AI?

597
00:22:36,984 --> 00:22:38,753
Tell me what you'd like to see.

598
00:22:39,084 --> 00:22:39,825
Finally,

599
00:22:39,825 --> 00:22:42,753
I want to end today's video with a certain image.

600
00:22:43,265 --> 00:22:44,473
150 years ago,

601
00:22:44,585 --> 00:22:47,023
how did the last generation that valued physical strength

602
00:22:47,025 --> 00:22:48,614
view the steam engine?

603
00:22:49,144 --> 00:22:51,953
They must have felt just as we do regarding AI today:

604
00:22:52,105 --> 00:22:53,153
curiosity at first,

605
00:22:53,265 --> 00:22:54,433
then shock,

606
00:22:54,505 --> 00:22:55,453
then fear,

607
00:22:55,744 --> 00:22:56,953
then anger.

608
00:22:57,384 --> 00:22:58,493
Some smashed the machines,

609
00:22:58,825 --> 00:23:00,513
some denied them,

610
00:23:00,585 --> 00:23:02,554
some pretended they didn't exist,

611
00:23:02,704 --> 00:23:05,713
others fought to become the ones operating them.

612
00:23:06,144 --> 00:23:06,534
In the end,

613
00:23:06,825 --> 00:23:08,183
the machines came anyway,

614
00:23:08,184 --> 00:23:11,473
completely rewriting the world based on manual labor.

615
00:23:11,785 --> 00:23:14,034
Today, we stand at a similar crossroads.

616
00:23:14,664 --> 00:23:15,213
The difference is,

617
00:23:15,585 --> 00:23:17,074
we are not bystanders.

618
00:23:17,184 --> 00:23:17,653
We ourselves

619
00:23:18,184 --> 00:23:20,233
are the ones relying on 'strength.'

620
00:23:20,384 --> 00:23:20,814
Except,

621
00:23:21,144 --> 00:23:22,953
this time it's mental strength,

622
00:23:22,984 --> 00:23:25,634
and the steam engine is now a Large Language Model.

623
00:23:25,944 --> 00:23:28,074
I don't know how this generation will get through it,

624
00:23:28,384 --> 00:23:29,713
but I know one thing:

625
00:23:29,984 --> 00:23:32,713
history doesn't give you the same chance twice.

626
00:23:33,265 --> 00:23:35,153
Among those people 150 years ago,

627
00:23:35,305 --> 00:23:37,273
those who first learned to operate the machines

628
00:23:37,384 --> 00:23:39,273
became the first generation of elite workers.

629
00:23:39,744 --> 00:23:42,394
Those who insisted they'd never use such things

630
00:23:42,424 --> 00:23:43,753
were left behind by time.

631
00:23:44,424 --> 00:23:45,334
Today's watershed moment

632
00:23:45,625 --> 00:23:47,074
is already right in front of you.

633
00:23:47,505 --> 00:23:47,894
Time

634
00:23:48,305 --> 00:23:49,273
will not wait.

635
00:23:49,825 --> 00:23:50,493
I'm Leo Wang.

636
00:23:50,785 --> 00:23:51,773
See you next time.