シャドーイング練習: "재무 모델에도 부담" 저커버그가 털어놓은 메타 AI 올인의 대가 - 動画で英語スピーキングを学ぶ
練習モードを選ぶ
各モードは全画面の練習ルームで開きます。
96 文
文が短すぎたり長すぎる場合は、Editをタップして調整してください。
1
and i think that that it's both difficult for the people doing it
2
but i do think that
3
that also puts some level of strain on the rest of
4
the company on the financial model of the company actually quite puzzling to me at first the degree to
5
which you were like willing to make these massive bets is something actually nat
6
and i would often like talk about i was like wow we're really going big here like
7
and both of us would be like would you go
8
this big would you go this big you know
9
we're taking our compute we're not like peanut buttering across a lot of competing efforts we're basically
10
Basically, we're investing in the key people, like a lot of compute per researcher, and then they can do their work.
11
It was an MSL.
12
He told me to find the message to the king's first message.
13
He told me to talk about the time, and he told me to talk about it at 44 minutes.
14
That's the next story.
15
This was like the core of your strategy, right?
16
It's basically like you actually don't need a massive number of people.
17
But what you do need is for like each of the people to be as good as possible,
18
which kind of informed this kind of crazy recruiting push and like very intense filtering that we had to take,
19
you know, some of the best people from inside meta and a very intense process to get people from outside meta.
20
The people of the company are working as a start -up company as a big company.
21
The online learning team asked how big is the question.
22
The answer is how big is a pre -training team?
23
It doesn't need to be that big, but I think if you look at some of the other labs,
24
it's probably quite big in some other places.
25
So I think part of the reason why these models are doing well is because you have people who we're betting on,
26
and they have a lot of resources because we're...
27
That's what I'm going to say in front of you.
28
I'm going to compete with each other team.
29
I'm going to compete with each other.
30
I'm going to compete with each other.
31
I'm going to compete with each other.
32
I'm going to compete with each other.
33
I'm going to compete with each other.
34
I'm going to compete with each other.
35
I'm going to compete with each other.
36
I'm going to compete with each other.
37
I'm going to compete with each other.
38
I'm going to compete with each other.
39
I'm going to compete with each other.
40
In general, one of the, one of the bigger challenges of scaling bigger models is it's not just about how many flops
41
and how much compute you want to be able to scale data with the data.
42
I mean, this is like part of what when the original kind of scaling laws for pre -training were written.
43
I mean, this is sort of what came out of that is that.
44
So if the internet isn't growing that quickly and you want to scale bigger models that can be more intelligent,
45
I think one of the interesting problems is, all right, where's the data coming from?
46
Yes.
47
So, I don't know.
48
I think maybe when we got started, a lot of people were like, OK, like, how is this going to work?
49
It's like the kind of like the social media guy is working with the like data guy.
50
Like, is that like, are we going to build a frontier lab here or what?
51
But but I mean, but my but I mean, my like really strong conviction, because I talk to like all the leaders of all the labs,
52
like all the heads of research, like all this stuff.
53
And I was like, no, Alex is the guy.
54
Like, he's the fucking guy.
55
And we're going to basically be able to go pull together a team
56
and recruit the people that we need to go do this.
57
Jokerberg은 연구소를 이끌 사람에게 필요한 건 최신 학습기법의 전문성보다 환이 바뀔 때마다 생각을 빠르게 고치는 창업가의 기질이라고 말했습니다.
58
그런데 왕이 메타에 와서 처음 마주한 분위기는 달랐습니다.
59
이번엔 왕, 본인의 이야기입니다.
60
when i joined um i was in this it was in this um
61
you know meeting with some of the leads of the existing effort
62
and uh i was like okay i think the first thing
63
we need to do is train a really good reasoning model um by um
64
you know by some time frame i don't know when uh
65
but this is like clearly like the most important milestone
66
and um one of the the key leads in the group um who no
67
But it was like, oh, we don't need a reasoning model.
68
Does any, or any of our, do our consumers need a reasoning model?
69
Like, I feel like we don't need to do a reasoning model.
70
We can just do a normal model.
71
And I was, I had like this heart sinking moment.
72
Oh no, I've joined the wrong company.
73
I was like, oh, I've got to take like 10 steps back.
74
And I think...
75
The model model is a model model in the summer of the year's open -up.
76
The model model is a model model.
77
The model model is a model model model and the other leaders were able to change the AI.
78
The Instagram leader of Adam Moseri, made a small app in Instagram.
79
Collis told him that he could use a new idea that he could use a new idea.
80
The idea of this idea is that it's different.
81
Obviously, you can't...
82
A lot of what we had to do here was rotate a lot of the rest of the company and,
83
like, for things like the infrastructure team that now, you know, it's like building these huge clusters to support this,
84
I mean, that's obviously a huge thing.
85
So, huge parts of the company kind of had to pivot to do this,
86
and obviously you can't do too many of those at the same time, right?
87
Only one thing can be the top priority.
88
Yes, yes.
89
So, how do you handle that burden?
90
He ended up with a single product, commerce, and commercial sales.
91
It's been a big deal for years to come.
92
1, 2, 3, 4, 4, 5, 5, 6, 6, 7, 7, 8, 8, 9, 9, 9, 9, 9, 9, 9, 9, 10, 9, 10, 10,
93
11, 11, 11, 11, 12, 12, 13, 13, 14, 14, 14, 14, 14, 14, 14, 14, 14, 14, 14, 14, 14, 14, 14, 15, 14,
94
14, 15, 14, 15, 14, 15, 15, 15, 15, 15, 16, 15, 16, 16, 16, 17, 17, 17, 17, 17, 17, 17, 17, 18, 18,
95
18, 18, 19, 18, 19, 18, 19, 19, 19, 19, 19, 19, 19, 19, 19, 19, 19, 19, 19, 19, 19, 19, 19, 19, 19,
96
19, 19, 19, 19, 19, 19, 19, 19, 19, 19, 19, 19
アプリをダウンロード
流暢に話すために必要なすべて
AI発音採点文ごとに採点
IPA練習すべての音をマスター
単語学習自分の単語帳を作る
単語ゲーム遊びながら覚える
FOR YOU
おすすめ動画
C12
182 文
We Let $200 GPT-Pro Think for 7 Days on Unsolved Math
Daily ConversationUniversity StudentsWorking Adults
C11
623 文
The Only SaaS Moats Left in 2026
Daily ConversationUniversity StudentsWorking Adults
C11
389 文
They JUST Accidentally Revealed the Truth: The AI Bubble
Daily ConversationUniversity StudentsWorking Adults
C11
398 文
Sec Growth DataScience staff meeting Sep 14 2022
Daily ConversationUniversity StudentsWorking Adults
C11
207 文
Go’s Big Mistake
Daily ConversationUniversity StudentsWorking Adults
C11
280 文
How AI deprives young Singaporeans of jobs and critical thinking | Deep Dive
Daily ConversationUniversity StudentsWorking Adults
C11
270 文
I Looked Into How Devs Are Actually Using AI
Daily ConversationUniversity StudentsWorking Adults
C11
19 文
Mark Zuckerberg: “Alex, are you sure this is going to work?”
Daily ConversationUniversity StudentsWorking Adults
C1286 文
Rethinking Claude Code Best Practices — Lydia Hallie, Anthropic
Daily ConversationUniversity StudentsWorking Adults
C14
169 文
What Is Jev? The AI Model That Doesn't Generate Text
Daily ConversationUniversity StudentsWorking Adults
C18
55 文
When AIs act emotional
Daily ConversationUniversity StudentsWorking Adults
C11
315 文
How Disruptive Will Generative AI Be? A Micro-Level Analysis of Evidence and Expectations
Daily ConversationUniversity StudentsWorking Adults
このレッスンの語彙とスピーキングのポイント
このスピーキングレッスンは、動画「"재무 모델에도 부담" 저커버그가 털어놓은 메타 AI 올인의 대가」を教材にしています。 繰り返し出てくる語は次のとおりです：model, compete, reasoning。 この動画には、シャドーイング用の文が89文、単語が1033語あります。 音声の長さは8:30です。 話す速さは1分あたり約121語で安定しており、シャドーイングしやすいペースです。 単語の86%は英語の頻出3,000語に含まれます。残りは練習の前に確認しておきましょう。
この動画の重要語彙
動画に出てくる覚えておきたい単語15語を、発音と意味つきで紹介します。
| 単語 | 発音 | 意味 |
|---|---|---|
| compete 動詞 | /kəmˈpiːt/ | 争う, 競う |
| intense 形容詞 | /ɪnˈtɛns/ | 激しい |
| recruit 動詞 | /ɹɪˈkɹut/ | 募集する |
| infrastructure 名詞 | /ˈɪnfɹəˌstɹʌkt͡ʃə/ | インフラストラクチャ, インフラ |
| priority 名詞 | /pɹaɪˈɒɹɪti/ | 優先, 優先権 |
| butter 名詞 | /ˈbʌtɚ/ | バター, 牛酪 |
| resource 名詞 | /ɹɪˈzɔːs/ | 資源, 資産 |
| commerce 名詞 | /ˈkɑ.mɚs/ | 貿易, 交易 |
| invest 動詞 | /ɪnˈvɛst/ | 投入する |
| intelligent 形容詞 | /ɪnˈtɛlɪd͡ʒənt/ | 聡明な, 賢い |
| filter 名詞 | /ˈfɪltɚ/ | 濾過器, フィルター |
| burden 名詞 | /ˈbɝdn̩/ | 積み荷 |
| sink 動詞 | /ˈsɪŋk/ | 沈む |
| inform 動詞 | /ɪnˈfɔɹm/ | 告げる, 知らせる |
| conviction 名詞 | /kənˈvɪkʃən/ | 信念 |
繰り返し練習したい文
動画に出てくる短く完結した文です。日常会話でそのまま使えます。
- I'm going to compete with each other team.
- I'm going to compete with each other.
- I mean, that's obviously a huge thing.
注意したい発音
話し手はI'm, we're, don'tなど、短縮形や弱形を32回使っています。聞こえたとおりの短い形で発音しましょう。
- 長い単語（アクセントの位置に注意）: infrastructure /ˈɪnfɹəˌstɹʌkt͡ʃə/, priority /pɹaɪˈɒɹɪti/, intelligent /ɪnˈtɛlɪd͡ʒənt/
日本語話者が苦手な音：
- /v/ — /b/ にならないように、上の歯を下唇に当てる: invest /ɪnˈvɛst/, conviction /kənˈvɪkʃən/, pivot /ˈpɪv.ɪt/
- /f/ — 「フ」ではなく、上の歯と下唇で出す: infrastructure /ˈɪnfɹəˌstɹʌkt͡ʃə/, filter /ˈfɪltɚ/, inform /ɪnˈfɔɹm/, frontier /ˈfɹʌntɪə/, flop /flɑp/
- 子音の連続 — 間に母音を入れない: intense /ɪnˈtɛns/, infrastructure /ˈɪnfɹəˌstɹʌkt͡ʃə/, priority /pɹaɪˈɒɹɪti/, invest /ɪnˈvɛst/, intelligent /ɪnˈtɛlɪd͡ʒənt/
この動画での練習方法
- まず声を出さずに動画を最後まで聞き、知らない単語をメモします。
- 通常の速度で一文ずつシャドーイングし、話し手のリズムに合うまで繰り返します。
- 自分の声を録音して元の音声と比べます。compete, intense, recruitなどの単語に特に注意しましょう。
シャドーイングとは？英語上達に効果的な理由
シャドーイング（Shadowing）は、もともとプロの通訳者養成プログラムで開発された言語学習法で、多言語習得者として知られるDr. Alexander Arguelles によって広く普及されました。方法はシンプルですが非常に効果的：ネイティブスピーカーの英語を聞きながら、1〜2秒の遅延で声に出してすぐに繰り返す——まるで「影（shadow）」のように話者を追いかけます。文法ドリルや受動的なリスニングと異なり、シャドーイングは脳と口の筋肉が同時にリアルタイムで英語を処理・再現することを強制します。研究により、発音精度、抑揚、リズム、連音、リスニング力、そして会話の流暢さが大幅に向上することが確認されています。IELTSスピーキング対策や自然な英語コミュニケーションを目指す方に特におすすめです。