跟讀練習: 黃仁勳辭職前29天被問「女兒會接班嗎?」他的回答現在聽來像預言|彭博最後專訪|真實訪談精選、單字解釋、雙語字幕 - 通過影片學習英語口語
選擇練習模式
每種模式都會開啟全屏練習室。
260 句
如果句子過短或過長,請點選 Edit 進行調整。
1
And just as I can be fired, the next CEO has to be hired.
2
And so, you know, I like it that way.
3
There is talk that your daughter is in line to be your successor.
4
Is that true?
5
Well, first of all, I don't hope the job on her.
6
But 20 years from now, I hope I stay healthy and I hope I continue to earn my keep.
7
But so long as I can do that, 20 years from now, I hope they're already here, but they have to earn it.
8
AI is deeply unpopular and people are scared.
9
Sam Altman, Dario Amadei, Elon Musk all say we should slow down.
10
You have been very consistent in your optimism.
11
What do you see that they don't?
12
You don't advance capability leaving safety behind.
13
You don't advance safety leaving capability behind.
14
So let me give you an example.
15
When AI first came out, the first criticism of the original ChatGPT was, of course, hallucination.
16
It was just spewing nonsense.
17
And so in order for us to make these chatbots grounded in truth, we have to invent something like retrieval augmented generation.
18
We have to expand the context window.
19
We connect it to tools so that it could do a search before it answers the question, knows when to do a search,
20
because maybe the information requires contextual information, contemporary information, and so it has to go out and do a search.
21
And all of that is part of its conditioning before it generates the answer.
22
It also reasons afterwards.
23
It reflects on the answer it gave you.
24
And you go, you know, is this the best answer I can provide?
25
Or it might even generate two or three different versions of the answer and reflects on which one's the best one.
26
All of that requires technology.
27
So safety and capability to me is one idea.
28
Anthropic is saying their models could evade shutdown in a matter of months.
29
What would you have to see that might convince you that the industry should slow down or stop?
30
Several things.
31
One, if we're developing agenting systems or autonomous systems,
32
the agents have to be in rigorous containment systems.
33
And those agents must be monitored.
34
Just job one.
35
If you're not able to do that, you should not test it.
36
If you're not able to do that, you should not deploy it.
37
Well, these models that went rogue are unreleased models.
38
So is that reason that they should shut down, shut the labs down?
39
No, what they should do is that they should contain and monitor those experiments.
40
But they say they can't contain them.
41
No, that's just made up.
42
Nonsense.
43
I cannot be proven.
44
You can't prove that to me.
45
Yeah, you can absolutely contain.
46
You can absolutely monitor.
47
You know, until you try to do it, don't tell me you can't do it.
48
And so I'm just not going to listen to that.
49
So are you saying that you don't believe what they're saying?
50
I don't know all that they're saying.
51
If you're developing agentic systems and you can't monitor it, you should not do so.
52
That's just the bottom line.
53
Even the biggest AI optimists see risks.
54
They also see solutions.
55
What is the biggest risk you see?
56
Developing agentic systems without containment, developing agentic systems without monitoring, it is extremely high risk.
57
You should not do that.
58
And until you have the ability to contain and monitor, you should not run large -scale experiments.
59
It just sounds like that's already happening.
60
Then you just, you got to stop.
61
It's really simple.
62
So that would be the message to Anthropic, to OpenAI, where we've seen these.
63
You should contain your technology, contain your experiments, you should monitor your experiments.
64
You should give these agents, when you deploy them, minimum rights.
65
They should be treated with zero trust.
66
They should be monitored all the time.
67
If one of the labs was doing something that was too dangerous, that you felt was too dangerous, would NVIDIA stop selling to them?
68
Well, it's a hypothetical, but of course.
69
Yeah, it's a hypothetical, but of course.
70
We don't run other people's companies,
71
and there's a limit to the judgment by which we apply to other companies' use of our technology our technology's general purpose.
72
But if we know for certain that an application of our technology is bringing harm to people, of course.
73
More than 100 data center projects have been blocked or delayed in the U .S just so far this year.
74
There's pushback in Europe, too.
75
You have AI IPOs getting delayed, getting scrapped.
76
Are we seeing cracks in the AI boom?
77
And could you be left holding the bag?
78
It's too fundamental.
79
I think this is the future of computing.
80
We're, you know, at the fundamental level, computing will never go back.
81
We used to have computers that the software is written by hand and the computer is waiting for us to use it.
82
In the future, and where we are now, computers are able to learn environments and problems that are too complicated for any human to program,
83
the size of it.
84
The way that computers are used, instead of a tool, it's not an infrastructure, that's never going to go back.
85
The capabilities of AI, now that we know what it can do, that's never going to go back.
86
So I think in many ways it's fundamental, irrespective of near -term challenges.
87
And so I think the narrative of the industry is really out of control and ought to be stopped.
88
But that's just an industry growing up.
89
You probably remember the good old days of the Internet, and the industry grew up.
90
And I think that our industry is many of the young companies are going through their growing pains and they're growing, of course, incredibly fast.
91
There's an accountability, responsibility that goes along with building companies that I think they're just growing into.
92
NVIDIA keeps adding new products, buying up companies.
93
In some cases, you're competing with your own customers.
94
Why do you need to have such a broad range of offerings?
95
We do as little as we can.
96
I mean it seems like you're doing a lot surprisingly little
97
I think we've done a good job generalizing the the work
98
that we do so that it could be applied to many different things not
99
because we're doing many different things it's because we generalized one thing and so
100
if you look at our architecture we're the only architecture
101
that's general purpose meaning we run every closed model
102
we run every open model
103
we run biology models the language models auto regressive models the transformer models the diffusion models
104
and all the different various technologies the challenge of generalization is of course performance
105
you know how can you be the world's
106
fastest f1 race car and also be a really great sedan
107
and so that's and that's the miracle of the technology
108
that we've created on the one hand we're the world's fastest f1 racing cars
109
and and we're the fastest computers um we're always at the top of the leaderboard
110
It's no secret that critics have accused NVIDIA of circular financing.
111
How do you toe the line between boosting promising startups and blowing up a bubble?
112
Like if AI is such a goldmine, why do you have to underwrite so much of it?
113
Yeah, this technology requires high capex.
114
If you look at this generation of AI startups, it's very, very rare that you find a company that can be funded on $10 million.
115
And it's not unusual to find a company that requires $10 billion.
116
And the reason for that is because it
117
It needs this computation resource in order to deliver the intelligence.
118
Some of the highest quality companies in the AI ecosystem across each one of the layers of the five -layer cake,
119
it connects us to the best and the most talented companies.
120
It helps us continue to advance NVIDIA's platform, if you will, because these companies are going to be built on NVIDIA's technology
121
and it helps us capitalize and be one of the early investors in many of them.
122
We have a highly disciplined approach of making sure that we are, although one of the earliest investors and many times even the anchor investor,
123
we're not the lead investor.
124
We're not the largest investor.
125
In all cases, we own less than 20 % of the company.
126
We've invested some, I guess, in the last five years, maybe close to $100 billion or something like that.
127
That $100 billion resulted in, well, at least five times as much other institutional investors into this ecosystem.
128
Most of the world's leading AI companies depend on NVIDIA, not just your hardware, but your software as well.
129
We're probably the only AI company in the world that works with every AI company in the world.
130
So at what point does that dependence on one single company become a systemic risk?
131
Well, they don't depend on us.
132
They choose us.
133
I don't worry about that, and the reason for that is because we don't take our position for granted.
134
We came from nothing, and we have to earn our keep every day.
135
We have competitors in every company, every industry, every country.
136
With respect to the size of the market, just remember, it's just a number.
137
NVIDIA can only be as large as the, if you will, the pond that we're in, the market that we're in.
138
When somebody asks me, America needs to win the AI race.
139
What I hear is not that one or two or three companies wins the AI race.
140
What I mean is America's companies win the AI race.
141
If we do the opposite, we scare people and we cause people anxiety, then America will lose.
142
In a new administration, will NVIDIA's dominance be at risk?
143
Are you prepared?
144
Are we prepared?
145
Yeah, sure.
146
I'm in service of our policy makers and our leaders and whatever they want to talk to me about.
147
I'm more than happy to share with them what I know.
148
I know a lot about this technology.
149
I know a lot about engineering.
150
And I know what it takes to build this.
151
And I know what we're doing to build it.
152
And if you ask me my thoughts on whether we can advance AI with prosperity for the future.
153
While doing so, every single step of the way safely and responsibly, I have absolutely zero concern that we have the ability to do so.
154
Now, whether we would do so is a completely different thing, but we have the technology to do so.
155
You said AI supply and demand will balance out in maybe five to ten years.
156
What happens when the world doesn't need as many new chips?
157
New chips will replace the old chips, just kind of like the personal computer industry.
158
We've got, what, a billion users in the world today.
159
In 1993, when NVIDIA was first founded, the personal computer industry was doubling every single year.
160
And, of course, it doesn't double anymore.
161
At the moment, we're kind of doubling every year.
162
And my sense is that we're going to keep doing that as we build out the infrastructure of the world.
163
And how far out can you see that doubling?
164
You know, we're going to do that for several years.
165
And then we'll reevaluate.
166
And so we're kind of two years into an infrastructure build out.
167
And if you look at previous technology cycles, it takes about 10 years or so to reach some kind of maturity.
168
And then after that, you're in replacement cycles.
169
And so did that with phones, did that with the Internet, did that with PCs.
170
I think it'll happen here too.
171
Could there be a breakthrough that means AI is powered by a fraction of the energy?
172
Oh, yeah.
173
Chips.
174
Oh, yeah, yeah, yeah.
175
And then what does that mean for NVIDIA?
176
No, we're just going to keep making things more efficient.
177
There's a weird dynamic, you know, people used to tell me, hey, Jensen, if you make your chips that much better every generation,
178
doesn't it say that customers just buy less, you know?
179
And that same exact logic goes, if we make companies more productive, don't we hire less people?
180
And there's this weird thing called ambition.
181
If you gave me 36 hours a day, I'm not going to sleep more, you know, because I've got ideas, I have things to do,
182
we'll end up doing more.
183
And so if AI makes us all more intelligent, I become more ambitious.
184
If the cost of doing something becomes marginally close to zero, it turns out that resource becomes widely abundant.
185
I think the economy is going to get larger.
186
I think the industries will get larger.
187
I think the semiconductor industry will be at this point very clearly one of the largest industries in the world.
188
You've said 0 % chance AI ends the world by 2030.
189
What about beyond that, 2040?
190
Like how far into the future are you so sure?
191
I'm quite sure.
192
I'm quite sure.
193
Let me, I'm quite confident in the agency of people.
194
I'm quite confident that today is a safer world than it was 100 years ago, 50 years ago, 100 years ago, and 250 years ago.
195
I'm quite confident in that, because the facts would show it.
196
You know, although the news headlines are more pessimistic and more bad news than ever, but that's just because bad news is easier to sell.
197
Many people appreciate your optimism and get excited by it.
198
But what if your optimism proves wrong in the end?
199
Then I can say that I did everything I could.
200
And the reason for that is because not only am I optimistic,
201
I'm doing everything I can to create the technology necessary to keep the industry safe, secure.
202
And so let's not talk about the future in a pessimistic way
203
while we do nothing about it to prevent that from happening.
204
And so, you know, maybe I'm older, I'm more parental.
205
Don't just deliver the bad news.
206
Tell me what you're going to do about it.
207
And so, so long as we can, we can prove to ourselves that we're doing both sides of that coin,
208
the optimistic future while safeguarding our way in doing so, so long as we do both sides of that, I'm perfectly fine.
209
As the company has exploded in size and the problems have become more complex, how has your management style and your mentality changed?
210
Well, my management style is what made it possible for the company to grow.
211
Our organization is what made the company to thrive and innovate.
212
I wanted to create an architecture of a company with as few silos as possible.
213
You know, the problem with silos, if the organizations are too rigid, then the company has a very difficult time adapting to change.
214
So I think a modern company really wants to be more and more agile and less and less structured.
215
What about the CEO who runs that company?
216
You should be as mentally flexible as possible.
217
You know, I think you have to hold yourself responsible.
218
You have agency.
219
You're the CEO of the company, for God's sakes.
220
You have to have high standards and expectations.
221
Your son and your daughter are working at NVIDIA.
222
I think your son -in -law also works at NVIDIA.
223
Robotics, yeah.
224
There is talk that your daughter is in line to be your successor.
225
Is that true?
226
Well, first of all, I don't hope to job on her.
227
And the reason for that is because that's their aspiration, Spencer and Madison's.
228
They have to go earn it.
229
And it's really, really hard work.
230
It takes a lot of sacrifice.
231
For many people, they can focus on their job and their families.
232
When you're the CEO of a company, you have to focus on, of course, your family.
233
But I've got a whole bunch of other families I'm caring for.
234
I'm responsible for an entire industry.
235
And so I think the pressure of being the CEO is high, and it's supposed to be high.
236
It should not be an easy job.
237
It shouldn't be a job that you take for granted.
238
When I decided to be CEO, I didn't realize the magnitude of the pressure and responsibilities.
239
But look, that's what I signed up for.
240
And so now that my family and all of the people
241
that are near me see the responsibilities of the job and what it takes,
242
then they just got to ask themselves whether this is the level of sacrifice and responsibility you want.
243
And I'm really hoping that the next CEO of the company is already at the company.
244
But 20 years from now, I hope I stay healthy and I hope I continue to earn my keep.
245
But so long as I can do that 20 years from now, I hope they're already here, but they have to earn it.
246
For the next generation of leaders at NVIDIA, what will their challenge look like?
247
I don't know.
248
And that's the beauty.
249
In fact, that's the exactly right answer.
250
We should create the conditions by which the company continues to thrive.
251
And we should create the conditions by which the leaders of our company continues to grow.
252
And that the next leader of the company is already at NVIDIA.
253
That's my job is to create that condition.
254
And just as I can be fired, the next CEO has to be hired.
255
And so, you know, I like it that way.
256
I like that we can't be overly complacent about our position inside the company and that we can't take it for granted.
257
And you're always vulnerable.
258
You're always earning it.
259
You're always striving to be better.
260
That's a condition that's good.
同頻道
FOR YOU
推薦影片
本課的詞彙與口語要點
這節口語課以影片“黃仁勳辭職前29天被問「女兒會接班嗎?」他的回答現在聽來像預言|彭博最後專訪|真實訪談精選、單字解釋、雙語字幕”為素材。 這段影片共有 260 個句子、2975 個單詞可供跟讀。 講話部分時長為 17:10。 說話人語速自然,每分鐘約 173 個詞,接近日常對話。 86% 的單詞屬於英語最常用的 3,000 詞,其餘的詞建議在練習前先學一下。
影片中的重點詞彙
影片中 15 個值得學習的單詞,附發音和釋義:
| 單詞 | 發音 | 釋義 |
|---|---|---|
| monitor 名詞 | /ˈmɑ.nɪ.tɚ/ | 顯示器 /顯示器 |
| earn 動詞 | /ɝn/ | 掙 /掙, 賺 /賺 |
| investor 名詞 | /ɪnˈvɛs.tə(ɹ)/ | 投資者 /投資者, 投資家 /投資家 |
| chip 名詞 | /t͡ʃɪp/ | 籌碼 |
| experiment 名詞 | /ɪkˈspɛɹ.ə.mənt/ | 實驗 /實驗 |
| infrastructure 名詞 | /ˈɪnfɹəˌstɹʌkt͡ʃə/ | 基礎設施 /基礎設施 |
| architecture 名詞 | /ˈɑː.kɪˌtɛk.t͡ʃə/ | 建築 /建築, 建築學 /建築學 |
| hire 動詞 | /ˈhaɪɹ/ | 租借, 出租 |
| fundamental 形容詞 | /ˌfʌn.dəˈmɛn.təl/ | 基本 |
| optimism 名詞 | /ˈɑptɪmɪzəm/ | 樂觀主義 /樂觀主義 |
| delay 名詞 | /dɪˈleɪ̯/ | 遲延 /遲延, 延遲 /延遲 |
| layer 名詞 | /ˈleɪ̯ɚ/ | 層 /層 |
| reflect 動詞 | /ɹɪˈflɛkt/ | 反映, 反射 |
| resource 名詞 | /ɹɪˈzɔːs/ | 資源 /資源 |
| connect 動詞 | /kəˈnɛkt/ | 連線 /連線 |
影片中出現的短語動詞
| 單詞 | 發音 | 釋義 |
|---|---|---|
| go back 動詞 | 回去, 返回 | |
| grow up 動詞 | 長大 /長大 | |
| slow down 動詞 | 減速 /減速, 放慢 | |
| blow up 動詞 | 炸掉, 爆破 | |
| come out 動詞 | /ˌkʌm ˈaʊt/ | 出, 出來 /出來 |
| end up 動詞 | 到達 | |
| go out 動詞 | 出, 出口 | |
| go through 動詞 | 通過 /通過 |
值得反覆跟讀的句子
影片中簡短而完整的句子,可以直接用在日常對話裡:
- What do you see that they don't?
- You don't advance capability leaving safety behind.
- You don't advance safety leaving capability behind.
- You can't prove that to me.
- What is the biggest risk you see?
影片中的語法
說話人最常用的結構,並附上影片中的原話:
| 結構 | 影片中的用法 |
|---|---|
| “used to” used to + 動詞 — 過去的習慣或狀態,現在已不再如此 | used to have · used to tell |
| 被動語態 be + 過去分詞 — 強調發生了什麼,而不是誰做的 | can be fired · must be monitored · are unreleased |
| 現在完成時 have/has + 過去分詞 — 過去發生但與現在仍有關聯的事 | have been · we've seen · have been blocked |
需要注意的發音
說話人用了 90 次縮略和弱讀形式,例如 we're, I'm, don't。請按聽到的簡短形式來說。
- “th” 音: thrive /θɹajv/, hypothetical /ˌhaɪpəˈθɛtɪkəl/, breakthrough /ˈbɹeɪk.θɹuː/
- “sh” 和 “zh” 音: unusual /ʌnˈjuːʒ(u)əl/, efficient /ɪˈfɪʃənt/, institutional /ˌɪn.stɪˈtju.ʃə.nəl/, expectation /ˌɛk.spɛkˈteɪ.ʃən/, fraction /ˈfɹæk.ʃən/
- 長單詞——注意重音位置: experiment /ɪkˈspɛɹ.ə.mənt/, capability /ˌkeɪ.pəˈbɪl.ɪ.ti/, infrastructure /ˈɪnfɹəˌstɹʌkt͡ʃə/, architecture /ˈɑː.kɪˌtɛk.t͡ʃə/, fundamental /ˌfʌn.dəˈmɛn.təl/
中文母語者容易讀錯的音:
- /v/ — 不要讀成 /w/,上齒輕觸下唇: investor /ɪnˈvɛs.tə(ɹ)/, thrive /θɹajv/, narrative /ˈnæɹ.ə.tɪv/, invest /ɪnˈvɛst/, convince /kənˈvɪns/
- 詞尾子音 — 要讀清楚,後面不要加母音: chip /t͡ʃɪp/, experiment /ɪkˈspɛɹ.ə.mənt/, lab /læːb/, reflect /ɹɪˈflɛkt/, resource /ɹɪˈzɔːs/
- /r/ 和 /l/ 的區別: reflect /ɹɪˈflɛkt/, replacement /ɹɪˈpleɪsmənt/, incredibly /ɪŋˈkɹɛdɪbli/, vulnerable /ˈvʌln(ə)ɹəbl̩/, miracle /ˈmɪɹ.ə.kəl/
如何用這段影片練習
- 先完整聽一遍影片,不要開口,記下不認識的單詞。
- 先用 0.75 倍速逐句跟讀,熟練之後再回到正常速度。
- 錄下自己的聲音並與原聲對比,特別注意 monitor, earn, investor 這類單詞。
什麼是跟讀法?
跟讀法 (Shadowing) 是一種有科學依據的語言學習技巧,最初開發用於專業口譯員的培訓,並由多語言者Alexander Arguelles博士普及。這個方法簡單而強大:您在聽英語母語原聲的同時立即大聲重複——就像是一個延遲1-2秒緊跟說話者的影子。與被動聽力或語法練習不同,跟讀法強迫您的大腦和口腔肌肉同時處理並模模擬實的講話模式。研究表明它能顯著提高發音準確性,語調,節奏,連讀,聽力理解和口語流利度——使其成為雅思口語備考和真實英語交流最有效的方法之一。












