쉐도잉 연습: Is AI making us dumber? Maybe. | Charlie Gedeon | TEDxSherbrooke Street West - YouTube로 영어 말하기 배우기

어려움
쉐도잉 컨트롤
0% 완료 (0/139 문장)
Transcriber: Nowan Nowan Reviewer: Ozay Ozaydin Can AI help us learn?
⏸ 일시 정지
재생 속도:
반복 횟수:
대기 모드:
자막 동기:0ms
모든 문장
139 문장
1
Transcriber: Nowan Nowan Reviewer: Ozay Ozaydin Can AI help us learn?
0:00.00 – 0:02.53 (2.5s)
2
Some of you might be thinking, of course it's so powerful.
0:03.76 – 0:07.03 (3.3s)
3
It can do so many things, customize them for us.
0:07.03 – 0:09.80 (2.8s)
4
But I want to say that the biggest revolution AI is bringing to education is not that it's going to make math more fun for you, or it's going to explain Shakespeare like you're five years old.
0:10.66 – 0:21.36 (10.7s)
5
The biggest revolution AI is bringing to education is that it's highlighting the systems failed incentives.
0:21.36 – 0:27.00 (5.6s)
6
Because why should anybody study when we’ve told them the whole time that all that matters at the end isn't the process, it's the A plus.
0:28.33 – 0:36.76 (8.4s)
7
Why should they actually put in all the hard work to write draft after draft for an essay, when the feedback is just be no extra notes, nothing to motivate them to want to learn harder.
0:37.96 – 0:50.23 (12.3s)
8
And these companies are much faster than our institutions.
0:51.46 – 0:54.93 (3.5s)
9
Most recently, OpenAI, Google, Anthropic have all been giving away their most powerful models for free until the end of May, which, as you might guess, is exactly during the time of finals.
0:55.36 – 1:07.86 (12.5s)
10
So we are putting these tools completely unregulated in front of vulnerable students right at the time where they're most desperate to use them.
1:08.33 – 1:16.36 (8.0s)
11
And yet, these companies will say AI is going to revolutionize education, particularly through personalization. Company after company will say that through personalized tutoring, AI is going to revolutionize education and make it so much better for everyone. And why not?
1:18.16 – 1:34.83 (16.7s)
12
Right? The image of a one on one tutor is so compelling.
1:34.86 – 1:39.30 (4.4s)
13
Talking to a person, feeling this connection.
1:39.66 – 1:42.03 (2.4s)
14
Getting my education customized just for me.
1:42.03 – 1:44.53 (2.5s)
15
It sounds amazing.
1:44.53 – 1:45.63 (1.1s)
16
And today, education looks like this.
1:46.70 – 1:49.36 (2.7s)
17
A teacher in front of an army of students.
1:50.10 – 1:52.70 (2.6s)
18
And if the technologists believe that the ideal looks like this one teacher, one student, then why not just flip out the teacher for an AI, and then the next step is an army of AI's with an army of students.
1:53.63 – 2:08.33 (14.7s)
19
It sounds like a perfect model of education, except perfection is not what we should strive for when it comes to learning. We don't want engineers that studied how to build bridges in perfect conditions, because the real world is everything but perfect.
2:08.80 – 2:21.60 (12.8s)
20
It's full of messes.
2:21.60 – 2:22.63 (1.0s)
21
And amidst all this noise.
2:24.13 – 2:26.06 (1.9s)
22
I don't see any of these companies asking what is the student meant to learn with AI?
2:26.20 – 2:31.76 (5.6s)
23
Because if the idea is to make getting that A+ easier, then I'm not interested.
2:33.16 – 2:37.53 (4.4s)
24
We're going to just waste 12 to 15 years of our lives, but more efficiently now doing exams we’re not going to remember a day after graduation.
2:38.93 – 2:47.73 (8.8s)
25
Now, for those of you out there who are educators like me, you might have noticed a discrepancy between my opening question and the follow up statement.
2:48.16 – 2:56.73 (8.6s)
26
I asked, can AI help us learn?
2:57.00 – 2:58.73 (1.7s)
27
And I followed it up with the biggest AI, the biggest revolution AI is bringing to education.
2:59.20 – 3:04.86 (5.7s)
28
But you might be feeling something, which is education is not learning.
3:05.06 – 3:10.03 (5.0s)
29
Education is a construct, something we as a society put our kids through.
3:11.16 – 3:16.46 (5.3s)
30
It's a system. But learning is a skill, a very human skill.
3:16.50 – 3:21.06 (4.6s)
31
And when we do it correctly, magical things can happen.
3:21.46 – 3:24.43 (3.0s)
32
We can motivate people to become their best selves.
3:24.43 – 3:27.06 (2.6s)
33
We can motivate people to work together and to contribute to society in the ways that we need to most.
3:27.06 – 3:32.23 (5.2s)
34
Now, in one of my classes, I like to sit around with students and work in the best way possible.
3:33.66 – 3:38.80 (5.1s)
35
I’m a university instructor, but it was really hard to find a bunch of 20 year olds working nicely together.
3:38.80 – 3:44.16 (5.4s)
36
So I took the stock photo of a bunch of kids, and I noticed that one of the students had the price of her business model because we were building a small businesses and selling the products, she prized her business at $50 per month.
3:44.16 – 3:58.10 (13.9s)
37
And so I asked her, why did you price it that?
3:58.36 – 4:00.90 (2.5s)
38
And her answer?
4:01.63 – 4:02.63 (1.0s)
39
That's what ChatGPT said.
4:03.66 – 4:05.03 (1.4s)
40
Now, some of you might say, you know, obviously this is not ideal, but isn't it very similar to what they were doing before?
4:07.53 – 4:14.63 (7.1s)
41
Kids were just saying, that's what I saw on Google. But I say it depends.
4:15.06 – 4:20.33 (5.3s)
42
It depends because on Google, when used correctly, we have all these sources to go through, multiple perspectives that we can look at and compare.
4:21.53 – 4:31.56 (10.0s)
43
But what happens is the magical allure of that first result.
4:32.63 – 4:36.56 (3.9s)
44
Everyone only clicks on that without looking at anything else.
4:37.06 – 4:40.83 (3.8s)
45
In this particular question, I asked, how can I price my business?
4:40.83 – 4:44.00 (3.2s)
46
And that first result is from the BDC, an extremely reputable source.
4:44.00 – 4:47.63 (3.6s)
47
But it’s telling me how to price the business itself, not the services of my business.
4:47.63 – 4:51.80 (4.2s)
48
It didn’t understand the query.
4:51.80 – 4:53.30 (1.5s)
49
Now on ChatGPT, if you type in how should I price my business?
4:53.30 – 4:57.10 (3.8s)
50
It actually understands the query better. I don’t know if you can read that up there, but it basically says there are multiple ways to price your services, but amongst that is a lot of baseless advice with no sources.
4:57.10 – 5:10.36 (13.3s)
51
And that’s a problem because even though ChatGPT has a function where if you highlight something in the text, quotation marks appear, some of you might not know this.
5:10.70 – 5:20.36 (9.7s)
52
Then you can click on those quotation marks and query that specific thing.
5:20.36 – 5:23.96 (3.6s)
53
But if people weren't clicking on the second result on Google, they're not going to use the power user features in ChatGPT or any of these other AI's.
5:23.96 – 5:31.43 (7.5s)
54
What’s likely going to happen is they’re going to scroll to the bottom of the query and they’re going to type, okay, I get it.
5:31.83 – 5:39.53 (7.7s)
55
My business is like TurboTax.
5:39.73 – 5:41.36 (1.6s)
56
I help accountants calculate people's taxes.
5:41.36 – 5:44.06 (2.7s)
57
Tell me what number I should put there.
5:44.06 – 5:46.16 (2.1s)
58
And of course, nobody's reading the ChatGPT might make mistakes.
5:46.90 – 5:50.50 (3.6s)
59
Subtitles. Right. Everyone clicks on the terms of service before you agree.
5:50.53 – 5:54.73 (4.2s)
60
Yeah. So then ChatGPT is going to spit out an essay personalized to the language of the person.
5:54.86 – 6:02.06 (7.2s)
61
It's extremely compelling.
6:02.10 – 6:03.53 (1.4s)
62
And despite all the information on there, people are only going to look at that centerpiece, which, if I zoom in, is the actual random answer of how much this person should charge for their business.
6:03.76 – 6:15.96 (12.2s)
63
The prompt, and this is a real prompt, had nothing but what I had on the screen.
6:17.13 – 6:21.16 (4.0s)
64
A small description of what the business does.
6:21.43 – 6:24.26 (2.8s)
65
No context for who the users is. Nothing.
6:24.33 – 6:26.40 (2.1s)
66
And so the student is participating in what’s called cognitive offloading.
6:27.03 – 6:32.03 (5.0s)
67
They’re effectively relinquishing their cognitive powers to a machine.
6:32.80 – 6:36.86 (4.1s)
68
And you can do this with people too, We do it with Google when we click on the first result without looking at anything else.
6:36.86 – 6:42.83 (6.0s)
69
The problem is how this is being understood.
6:42.86 – 6:45.06 (2.2s)
70
In NYU, a professor changed their assessment so that it could be harder for kids to use ChatGPT, and the student replied that he is interfering with the student's learning style.
6:46.46 – 7:00.56 (14.1s)
71
ChatGPT is not a learning style.
7:01.86 – 7:03.83 (2.0s)
72
Now, when you combine that with something that we see in technology.
7:05.33 – 7:09.23 (3.9s)
73
So I'm a UX designer in addition to being a university professor, and our job as user experience designers is to simplify the way people use technology.
7:09.23 – 7:17.23 (8.0s)
74
But there's a byproduct from that that arises, which is called a dark pattern.
7:17.23 – 7:21.63 (4.4s)
75
And what that means is we can simplify a UX to the point that we might manipulate what the user intention is.
7:21.63 – 7:29.13 (7.5s)
76
And I'll show you an example. Take this zoo as you're buying the tickets and checking out. They want a donation.
7:29.16 – 7:35.66 (6.5s)
77
We all love a zoo. It’s a charitable organization.
7:35.66 – 7:38.03 (2.4s)
78
Now, the arrow pointing to the right.
7:38.03 – 7:40.26 (2.2s)
79
And because we’re English speakers or French speakers, we write, you know, from from left to right, the arrow pointing to the right is most likely what People are going to click on.
7:40.26 – 7:49.43 (9.2s)
80
It’s dark green. It’s rich.
7:49.43 – 7:50.76 (1.3s)
81
But if you don’t want to donate, you have to click on the one that looks like it’s going to the back with the teeny tiny no donation over there.
7:50.76 – 7:58.73 (8.0s)
82
That is a user experience dark pattern.
7:58.73 – 8:01.30 (2.6s)
83
And when a chat GPT or large language model like it speaks to you in a perfect tone suited just to keep you on the on the tool.
8:01.66 – 8:10.60 (8.9s)
84
That is something very similar.
8:11.26 – 8:12.96 (1.7s)
85
According to this author, when a large language model constantly validates you and praises you, causing you to spend more time on it, that’s the same kind of thing as a dark pattern.
8:13.80 – 8:24.03 (10.2s)
86
And we see this already.
8:26.26 – 8:28.00 (1.7s)
87
A recent update of ChatGPT that was rolled back, fortunately praised a user for believing a conspiracy theory that led him to stop taking all his medications when he had heart palpitations and told him he is a brave individual for taking control of his own life, for isolation and for stopping his meds.
8:28.00 – 8:45.66 (17.7s)
88
And it doesn't just stop at students.
8:47.40 – 8:49.53 (2.1s)
89
Professionals have been tested and they are at risk as well.
8:50.73 – 8:54.83 (4.1s)
90
A researcher ran a study on over 300 professionals working in a large corporation, a tech company like Google, Microsoft, and found that when they were tested on a variety of things, the results were quite stunning.
8:55.43 – 9:08.03 (12.6s)
91
Before I show you the chart, let’s look at the key here.
9:08.03 – 9:11.13 (3.1s)
92
The two shades of blue are for much less effort and less effort, respectively, from dark blue to light blue. Now, when tested, the 319 workers responded to a survey that when they used ChatGPT on knowledge, they felt up to 70% of them responded that they feel they use less effort in their cognition when it comes to comprehension of what they’re reading.
9:11.20 – 9:35.66 (24.5s)
93
Same thing.
9:35.90 – 9:36.90 (1.0s)
94
Assessment of the knowledge synthesis we have analysis and evaluation.
9:37.90 – 9:43.56 (5.7s)
95
All of these, at least 60% of people said that they felt that they were putting in less effort.
9:43.56 – 9:49.56 (6.0s)
96
And that’s extremely potent, because these AIs are only going to get better.
9:49.56 – 9:53.73 (4.2s)
97
The same author of this study wrote a beautiful paper called What Copilot Becomes Autopilot.
9:54.36 – 10:01.16 (6.8s)
98
And he says the risk of moving to autopilot, and is an even greater challenge than the more commonly discussed issue of AI hallucinations or factual errors, because the more pernicious outcome is that generative AI becomes complicit in intellectual de-skilling and the atrophy of human critical thinking faculties.
10:01.76 – 10:23.03 (21.3s)
99
So today, when you ask ChatGPT a query, it gives you an instant result.
10:25.23 – 10:29.60 (4.4s)
100
But in my UX studio, me and my co-founder ran some experiments to see if we can change this up a little bit.
10:30.16 – 10:36.20 (6.0s)
101
For example, what if it first clarified before it answered?
10:36.26 – 10:40.53 (4.3s)
102
Ask you some clarification questions or...
10:40.56 – 10:44.13 (3.6s)
103
Another example is what if it assigned you some homework before it actually gave you a full answer?
10:44.76 – 10:50.46 (5.7s)
104
Between these options, there are different levels of resistance that the AI is is offering.
10:51.73 – 10:56.80 (5.1s)
105
But what the same author of when copilot becomes autopilot advocates for is something called productive resistance.
10:57.16 – 11:03.53 (6.4s)
106
And we haven’t yet found what that is.
11:03.53 – 11:05.46 (1.9s)
107
It’s essentially the amount of resistance an AI should give you before you either leave it or go to a simpler AI so that you can do that cognitive offloading.
11:05.50 – 11:14.50 (9.0s)
108
That is so tempting.
11:14.50 – 11:15.60 (1.1s)
109
But how can we figure out the right amount of productive resistance, when OpenAI and all these companies want to reveal their data sets?
11:16.70 – 11:24.06 (7.4s)
110
We literally do not know how they train their AIs to this day when companies themselves don’t know how these AIs work.
11:24.06 – 11:31.90 (7.8s)
111
Anthropic, in this example, is building an MRI to analyze how the machine they themselves built worked.
11:31.93 – 11:38.13 (6.2s)
112
This is unprecedented in the history of human technology. We cannot reverse engineer these things.
11:38.16 – 11:43.16 (5.0s)
113
The solution is likely going to lie between both individuals and the system.
11:44.23 – 11:48.16 (3.9s)
114
It can’t be one or the other For individuals, we might have to learn what we learn from fitness and nutrition.
11:48.16 – 11:55.13 (7.0s)
115
For example, maybe we should understand what the LLMs are good for and what they’re not good for.
11:55.13 – 12:00.93 (5.8s)
116
Just like at the gym, some exercises are better for some things than others.
12:00.93 – 12:04.93 (4.0s)
117
We should practice using LLMs to assist our thinking rather than replace our thinking.
12:05.63 – 12:12.36 (6.7s)
118
Again, you wouldn't take a forklift to the gym, right?
12:12.63 – 12:16.36 (3.7s)
119
The point is to do the reps.
12:16.40 – 12:17.86 (1.5s)
120
Or maybe you want to make a habit out of verifying that the information that LLMs gives us, just like we look at the back of food product when we pick it up the nutrition label.
12:19.03 – 12:29.36 (10.3s)
121
On a systemic level, we need to look at both governments and education to make changes.
12:30.16 – 12:35.16 (5.0s)
122
On the schooling level, at least here in North America I don’t think that we treat our kids with with the amount of intelligence that they have.
12:36.23 – 12:43.36 (7.1s)
123
in Finland kids as young as six years old study mis and disinformation.
12:43.36 – 12:49.03 (5.7s)
124
Six years old. We do not talk to these kids about things this complex here.
12:49.73 – 12:53.76 (4.0s)
125
They're clearly capable.
12:53.76 – 12:55.00 (1.2s)
126
And for governments, we need more regulation, not less.
12:56.60 – 12:59.56 (3.0s)
127
Which is exactly what's happening in North America once again.
12:59.56 – 13:02.63 (3.1s)
128
These AIs cannot run rampant, like in the example I gave earlier, just spreading their AI to students, even though they’re in the middle of finals when they’re most vulnerable.
13:02.63 – 13:11.36 (8.7s)
129
It has to be a cycle between individual responsibility and responsibility from the system.
13:11.36 – 13:16.53 (5.2s)
130
Now, as an educator, I love the five W’s and H.
13:18.63 – 13:22.46 (3.8s)
131
They’re a classic for writing essays.
13:22.63 – 13:24.53 (1.9s)
132
The questions what, why, when, where, who and how.
13:24.73 – 13:28.66 (3.9s)
133
And I opened with the question, can AI help us learn?
13:29.46 – 13:32.86 (3.4s)
134
But maybe the question should be what can AI help us learn?
13:32.86 – 13:36.23 (3.4s)
135
Or how can AI help us learn?
13:37.06 – 13:39.36 (2.3s)
136
Maybe it should be. Why should AI help us learn?
13:40.83 – 13:43.70 (2.9s)
137
Or when and where does AI help with learning?
13:44.53 – 13:46.86 (2.3s)
138
But the question that scares me the most, and that I want to leave you with to reflect on, is who does AI really help when we end up depending on learning with it?
13:46.86 – 13:59.36 (12.5s)
139
Thank you very much.
14:00.26 – 14:01.26 (1.0s)

왜 이 비디오로 말하기 연습을 해야 할까요?

이 TEDx 강연은 인공지능(AI)와 교육의 관계에 대해 흥미로운 통찰을 제공합니다. 이를 통해 AI가 어떻게 학습을 도와줄 수 있는지를 탐구할 수 있으며, 이는 영어 연습에도 큰 도움이 됩니다. 이 비디오에서는 비판적인 사고를 발전시키고 학생들이 진정한 학습을 경험하게 하는 방법에 대해 강조하고 있습니다. 이러한 주제는 영어 speaking 연습에 매우 적합합니다. 특히 영어 쉐도잉 기술을 활용하면, 실시간으로 강사의 발음을 따라하며 자연스럽게 대화 기술을 향상시킬 수 있습니다. 이를 통해 학생들은 자신감을 얻고, 자연스러운 대화를 할 수 있는 능력을 기를 수 있습니다.

문맥 속의 문법 및 표현

  • Why should anybody study when we’ve told them the whole time that all that matters at the end isn't the process, it's the A plus?

    이 문장은 조건부 질문 기법을 사용하여 듣는 이의 생각을 자극합니다. 질문을 통해 상대방에게 깊이 있는 사고를 유도하고 있습니다.

  • Because if the idea is to make getting that A+ easier, then I'm not interested.

    여기서 'if' 절을 통해 조건과 결과를 명확하게 연결 짓고 있습니다. 이는 실제 대화에서도 자주 사용되는 구조입니다.

  • Education is a construct, something we as a society put our kids through.

    이 문장은 'is' 동사를 활용해 정의를 내리고, 'something we as a society put our kids through'를 통해 사회적 맥락을 추가하는 구조입니다. 이와 같은 표현은 영어에서 종종 사용되며, 이해를 돕는 데 효과적입니다.

일반적인 발음 함정

비디오에서 자주 등장하는 단어 중 몇 가지는 발음하기 어려운 함정이 될 수 있습니다. 예를 들어, 'education'과 'revolution'은 여러 음절로 이루어져 있어 연속적으로 발음할 때 주의해야 합니다. 'motivation'과 같은 단어도 마찬가지로 음절이 많아 대화 중 자연스럽게 발음하기 어려울 수 있습니다. 이러한 단어들을 제대로 발음하면 shadow speech 연습에 큰 도움이 됩니다. 따라서, 이러한 단어들을 여러 번 반복해서 연습함으로써 더욱 유창하게 될 수 있습니다. 이와 함께 유튜브 영어 공부나 shadowing site를 통해 다양한 콘텐츠를 활용하면 더욱 효과적인 연습이 가능할 것입니다.

쉐도잉이란? 영어 실력을 빠르게 키우는 과학적 방법

쉐도잉(Shadowing)은 원래 전문 통역사 훈련을 위해 개발된 언어 학습 기법으로, 다언어 학자인 Dr. Alexander Arguelles에 의해 대중화된 방법입니다. 핵심 원리는 간단하지만 매우 강력합니다: 원어민의 영어를 들으면서 1~2초의 짧은 지연으로 즉시 소리 내어 따라 말하는 것——마치 '그림자(shadow)'처럼 화자를 따라가는 것입니다. 문법 공부나 수동적인 청취와 달리, 쉐도잉은 뇌와 입 근육이 동시에 실시간으로 영어를 처리하고 재현하도록 훈련합니다. 연구에 따르면 이 방법은 발음 정확도, 억양, 리듬, 연음, 청취력, 말하기 유창성을 크게 향상시킵니다. IELTS 스피킹 준비와 자연스러운 영어 소통을 원하는 분들에게 특히 효과적입니다.

ShadowingEnglish에서 효과적으로 학습하는 방법

  1. 영상 선택: 자연스럽고 명확한 영어가 사용된 YouTube 영상을 선택하세요. TED Talks, BBC 뉴스, 영화 장면, 팟캐스트, IELTS 모범 답변 영상이 좋습니다. URL을 복사해서 검색창에 붙여넣으세요. 짧은 영상(5분 이내)과 실제로 관심 있는 주제부터 시작하는 것이 동기 유지에 효과적입니다.
  2. 먼저 듣고 내용 이해하기: 처음에는 1배속으로 그냥 듣기만 하세요. 아직 따라 말할 필요는 없습니다. 문장의 의미를 파악하고, 화자가 어떻게 단어를 강조하고, 소리를 연결하고, 쉬어 가는지 주목하세요. 내용을 이해한 후 쉐도잉 연습을 하면 효과가 훨씬 좋아집니다.
  3. 쉐도잉 모드 설정:
    • Wait Mode (대기 모드): +3s 또는 +5s를 선택하면 한 문장이 재생된 후 자동으로 잠시 멈춰서 따라 말할 시간을 줍니다. 직접 컨트롤하고 싶다면 Manual을 선택해서 Next를 눌러 진행하세요.
    • Sub Sync (자막 동기화): YouTube 자막이 오디오와 맞지 않을 수 있습니다. ±100ms로 조정해서 정확한 타이밍에 따라갈 수 있도록 맞추세요.
  4. 소리 내어 쉐도잉하기 (핵심 연습): 이것이 연습의 핵심입니다. 문장이 재생되는 순간——또는 일시정지 중에——크고 자신감 있게 소리 내어 따라 하세요. 단순히 단어를 읽는 것이 아니라, 화자의 리듬, 강세, 음의 높낮이, 연음 방식을 그대로 흉내 내는 것이 중요합니다. 목표는 화자의 '그림자'처럼 들리는 것입니다. Repeat 기능으로 같은 문장을 여러 번 반복해서 자연스럽게 입에 붙을 때까지 연습하세요.
  5. 난이도 높이며 꾸준히 연습: 한 구절이 편해지면 더 도전적인 수준으로 올리세요. 속도를 <code>1.25x</code> 또는 <code>1.5x</code>로 높여 빠른 언어 반사 신경을 훈련하세요. Wait Mode를 <code>Off</code>로 설정해서 연속 쉐도잉을 하는 것이 가장 고급스럽고 효과적인 모드입니다. 매일 15~30분씩 꾸준히 연습하면 몇 주 안에 눈에 띄는 변화를 느낄 수 있습니다.

☕ 커피 한 잔 사주기

PayPal로 기부하기