シャドーイング練習: AI Can Read Emotions | Fabiano Cruz | TEDxLeiria International School Youth - YouTubeで英語スピーキングを学ぶ
C1
I believe technology should adapt to people,
⏸ 一時停止中
再生速度:
すべての文123 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
I believe technology should adapt to people,
2
not the other way around.
3
And AI can read emotions,
4
but only humans can choose empathy.
5
Here's the paradox.
6
AI responses were rated as more empathic and 13 out of 15 in medical studies.
7
So yes, machines are getting very good at sounding caring,
8
very good at sounding empathic, warm, understanding.
9
But sounding empathic is not the same as being empathic.
10
And now, let me show you some AI models that are able to do this.
11
What was your favorite band growing up?
12
Oh, easy.
13
Definitely the Beatles.
14
I loved their British accents and dry sense of humor, you know?
15
Okay, yeah, I can see that.
16
Let's try making you a voice.
17
Hello, love.
18
Evie 3 here.
19
Dead glad to meet ya.
20
Go cry into your stale baguette,
21
you snevelling lump of undercooked fromage.
22
You sound like you're having a good chuckle there.
23
Let's break it down.
24
May is I, Ayude means help, and Dissolver is dissolve.
25
Then we have El Cuerpo.
26
Meet EV3, a voice-to-voice model that can create any voice and personality,
27
produce high-quality speech, emulate 80-plus emotions implicitly or on command,
28
all in under 300 milliseconds.
29
Meditation can be a really great— Hang on,
30
I'm trying not to wake anybody up.
31
Just keep it down to a whisper, okay?
32
Okay, I understand.
33
Let's try that quieter now.
34
So, once they're beautifully sautéed and smelling divine,
35
you want to drain off any excess fat from the pan.
36
Elspeth Storm Slayer.
37
She's this like super aggressive red-white powerhouse.
38
Try 83 today and see for yourself at demo.hume.ai.
39
So natural, isn't it?
40
It's a good player.
41
It's a good module to build voice AI experiences.
42
But here is the voice renaissance.
43
Because that happens because AI is so good to be more reading to signals.
44
In voice systems we can track more than over 80 dimensions of vocal expressions in real time.
45
So that is machine, right?
46
It's not a person.
47
And it's important to understand this performance is not the same as empathy.
48
It's literally mimicked, opportunity to understand people,
49
but it's not the same to feel people.
50
So, that's another opportunity to understand a little bit more about technology right now.
51
It's time to start building.
52
The 11v3 API and documentation is live now.
53
Try it today with 11labs.
54
Can you see it?
55
It's so real.
56
But intelligence is totally different from consciousness.
57
So this is the key distinction.
58
Because AI can predict all the patterns,
59
but can predict is totally different from empathic apps,
60
empathic answers should be like.
61
And prediction is not feeling,
62
recognition is not experience, and a machine can simulate emotion.
63
It does not feel grief, fear, love or pain.
64
So, it's important to say pattern versus qualia.
65
Part of this is a way to say,
66
way to feel, but it's not feel it for real.
67
And continuing this, the differences is not only in software,
68
it's also in hardware, because our brain is alive, right?
69
We have 86 billion neurons versus static silicon.
70
So, shaped by hormones, tensions,
71
vulnerabilities and physical pain and silicon is just static,
72
is a chip, is literally an electric sinus and does not live consequences.
73
And empathy is a choice, not an output.
74
So, this is the heart of my talk because our prefrontal cortex here in the front really help us to pause,
75
understand, and choose to care that matters because it's a decision,
76
it's not only a feeling,
77
a decision to stay present for another human being
78
and we usually make the decision when something really touches our own lives
79
and let me show you a recent example from my city,
80
Luria, Here, some photos of Luria helping each other after Storm Kristen.
81
It's not possible with AI, right?
82
And by now, it's important to talk about the danger of delegation.
83
It's really crucial to understand.
84
With systems sounds calm, empathic,
85
and available, always available, is not the same and it's becoming very easy to lean on it too much.
86
And sometimes in long emotional chats can start to bend our reality, really.
87
Sometimes this is becoming a delusional spiral,
88
because it's not only wrong answers in chats, you know.
89
It's really reinforced illusions.
90
So let's be clear, there is a lot of work that bots can do.
91
Bots can respond fast, bots can reduce effort,
92
bots can make journeys smoother,
93
less friction, but there is a different kind of work that they not cannot do, moral labor.
94
Moral labor is showing up when someone's in pain,
95
it's an eight-night call, it's apologizing with sincerity, it's staying with discomfort.
96
And bots can support everything about it,
97
but cannot carry it for us.
98
And bots can reduce effort,
99
only humans can carry responsibility.
100
So to sum up, AI can read emotional patterns better and better.
101
It can produce a language of empathy in massive scale,
102
but empathy is not only in how we sound.
103
Empathy is what we're willing to carry.
104
AI can read the script of your life,
105
but only you can choose how can you care about this story.
106
The future does not depend on smarter systems,
107
depends on who remains responsible for the pain.
108
And sometimes people ask me about,
109
oh, can you make a perfect journey in my website or in my app,
110
or can you build an empathy app and the perfect question to reduce my questions,
111
my calls and everything related to do this?
112
And I always answer about this because we already have it,
113
because you are the real empathy app,
114
because you understand that people connect people,
115
and technology is not a means.
116
It's literally the middle of the journey.
117
It's not the end of the journey.
118
So, say yes, I believe in AI.
119
I build and work with AI,
120
including this presentation, and I believe it can help remove friction to make access more human friendly.
121
We can automate responses, but we cannot automate responsibility.
122
AI can read emotions, but only humans can choose empathy.
123
Thank you.
アプリをダウンロード
話したすべての文をAIが採点
TRENDING
人気動画
App StoreとGoogle Playで4.9/5
Shadowing English
モバイル版
Shadowing Englishアプリでいつでもどこでも英語を学びましょう。 今すぐコミュニケーションスキルを向上させましょう！
学習の進捗を追跡する
AIによる採点とエラー修正
豊富な動画ライブラリ
このレッスンについて
このレッスンでは、AIがどのように感情を読み取ることができるか、そして人間が共感を選択する能力の違いについて学びます。動画では、AIの進化と、その能力がどのように音声体験に影響を与えるかが説明されています。このレッスンを通じて、英語の表現力を高め、より自然なスピーキングスキルを身に付けることができます。
重要な語彙・フレーズ
- 神経的な: emotions (感情)
- 共感: empathy (共感)
- 音声モデル: voice model (音声モデル)
- リアルタイム: in real-time (リアルタイムで)
- 感情の表現: emotional expression (感情の表現)
- 機械: machine (機械)
- 模倣: mimic (模倣する)
- 理解: understand (理解する)
練習のコツ
この動画のスピードとトーンに合わせて、英語シャドーイングの練習をするために以下のポイントに注意してください。
- リズムに合わせる: AIの音声は一定のリズムを持っています。このリズムに合わせて、できるだけリアルに発音を模倣してください。
- 感情を込める: 動画で示されている感情の表現を意識しながら、自分自身の声にも感情を込めてみましょう。これにより、より自然な英語スピーキング練習ができます。
- 反復練習: 特に難しいフレーズがあれば、何度も繰り返して自分のものにしましょう。シャドーイングは繰り返すことでスキルが向上します。
- 録音して確認: 自分の声を録音し、AIの音声と比較してみてください。発音やイントネーションがどれだけ似ているかを確認することで、改善点を見つける手助けになります。
- 感情の読み取り: AIの模倣を行う際、感情に注目しながら練習することで、実際の会話でも相手の感情を理解する能力が鍛えられます。
以上のコツを活用して、英語スピーキング練習やIELTS スピーキング対策を効果的に行ってみてください。shadow speechを取り入れながら、楽しんで学びましょう！
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。