Prática de Shadowing: The AI Atrophy Problem How CIOs Fight It - Aprenda a falar inglês com vídeo
Criando lição...
1
(bright curious music) ABBIE LUNDBERG: AI is helping your team work faster, but is it hurting how they think?
2
I'm Abbie Lundberg, Editor in Chief at the MIT Sloan Management Review, and we are here at the annual MIT Sloan CIO Symposium asking IT and business leaders what one thing do you do to keep critical thinking sharp as AI takes over more and more work?
3
Let's hear what they had to say.
4
MONICA CALDAS: The way we think about it is, we put mechanisms in place to enable and remind people what we have to do.
5
So for example, when we think about the evolution of technology, we have an emerging technology radar that we incorporate into our ways of working where we talk about what are the emerging techs, where are they going?
6
So for example, quantum computing.
7
And then we engage in different ways to both go deeper on the topic, but also then extrapolate what is the meaning of that to our business aspirations.
8
That's a great example of a way where, yes, you can go and do research on an AI tool, fill in the blank, whatever your preferred chat engagement is, but you also have to have your own point of view.
9
So, how do you use a variety of different sources of information to develop your own point of view?
10
Critical thinking is now more important than it ever has been because general information is so easily available.
11
But how do you actually connect that to the problems that you are trying to solve?
12
MICHAEL SCHRAGE: When you get an output or an answer from a Claude or a Grok or a Perplexity that you really, really like, that you really, really resonate with, enjoy it for a moment and then stress it, conflict it, counter it, say, this is terrific, but what are the three most important arguments against what you've said?
13
Or are you sure you're not hallucinating?
14
When I'm in a more innovative or creative mood, I'll adopt certain kinds of persona.
15
What would the biggest criticism of Peter Drucker or a Marshall McLuhan might have for this?
16
The point is, you have to build in, forgive me, a dialectical stress test.
17
You cannot trust the output to be accurate and precise.
18
The overarching takeaway I would urge people to consider is, don't view these outputs as answers.
19
View them as insights.
20
View them as hypotheses that you should test, and stress test, and yes, iterate around.
21
MELISSA SWIFT: So, when I get an AI output, what I do is I actually put on this persona to evaluate it.
22
I pretend I'm a super critical outside reviewer, that I'm going to check every detail, right?
23
I'm going to vet the ideas, I'm going to look at the sources.
24
I put myself in that mindset every time I review AI output.
25
And that reengages my critical thinking skills.
26
So I don't get lulled by kind of a smooth, shiny output that might not be right.
27
KERI PEARLSON: What I've noticed is that there is a marked difference in the activities that people do when they start to use AI.
28
The promise is that AI will make our work faster and more streamlined.
29
But the flip side is that we don't practice some of the skills that the AI has taken over.
30
Now, how can we reverse that?
31
Well, first of all, we want to make sure that people who use AI tools actually have the ability to validate the output that comes out of AI.
32
If you don't know the basic skills for doing research, for example, and you use AI to do your research, then you have no ability to know if the outcome is valid.
33
Is it reasonable? Has it been hallucinated?
34
And we want to make sure that workers don't have that outcome.
35
So, validating your output, practicing the skills once in a while to make sure you know how to do it.
36
Following the thread, even asking AI, how did you come to this conclusion?
37
Probing AI as if it were a strong assistant, and not a replacement for the work, is one way we can make sure to keep our skillset strong.
38
MEGHNA SHAH: The one thing me and my teams do a lot is we emphasize that AI should accelerate thinking.
39
It should not really replace thinking.
40
That the output that you're getting from your AI model should really be a starting point.
41
It shouldn't be the final answer, right?
42
Because creative thinking is no longer going to be about, can you create content from scratch, right?
43
It's really going to be, can you ask better questions?
44
Can you apply reasoning?
45
Can you apply the business context and the judgment that is needed with the response that you're getting from the LLM?
46
So, it's going to be a lot of those pieces that feed into the critical thinking part of it.
47
I think in today's day and age, when data is available so freely, I don't think having access to intelligence is going to be a differentiator.
48
What's going to be a differentiator is how strategically and how responsibly you can actually apply that intelligence is going to be a critical differentiator.
49
We also encourage no blind trust flows.
50
So, even though we are using AI for coding, our engineers are still checking code and reviewing code before it goes into production.
51
Our product managers are still reviewing requirements and validating that this is really what the customers need.
52
And our architects are also reviewing architecture to see is this really fitting with what the organization needs and the long-term implications.
53
So, there's a lot of those pieces to ensure that there's always critical thinking and a human in the loop and the human brain in the loop.
54
GEORGE WESTERMAN: So much of what we do is data analysis, writing, summarizing information.
55
And so the first thing that my team and I always have to ask is, how do we know that that is correct?
56
We're always asking that question, always digging a little bit deeper to test a few things to make sure. But then, you can't ask the LLMs.
57
The LLMs will tell you it's correct and they'll give you a really good reasoning, even though that may not be the situation.
58
So, that's the first step.
59
The other step we're trying to be really careful about is, is this the kind of question, the kind of prompt, that we can expect the engines to be good at?
60
Because we don't want to set ourselves up in a situation where it's trying to do something it's not built for, and then we have a giant mess in our hands.
61
MAX CHAN: At the end of the day, making sure that there is accountability owned by a human is critical.
62
And with that understanding, we make sure that anyone who is leveraging AI, who is using AI to deliver a piece of code, to change a design, they must have the foundation.
63
We make sure that employees and leadership do not forget the fact that, you know, being able to do something in AI does not mean that they have the skill to do it, they have the foundation to do it.
64
And so that is really one thing that we continue to ensure we keep an eye on.
65
And the other thing is with the speed of accelerations, right?
66
AI accelerations that is taking place.
67
We want to ensure that we have the ability to judge, applying judgment to make sure that we slow down where we have to.
68
It's not because everything needs to go so fast.
69
It's where do we need to slow down to gain that competitive edge?
70
VANESSA ESCRIVÁ GARCÍA: Okay, we believe more than ever that the key is the training of the people and to make sure that people understand that we need to make the right questions.
71
And AI, it's just the medium to obtain an objective.
72
And one of the examples for that is that we have started with the training with the top of the company.
73
The executive level have received this training, and we have a roadmap for it to make sure that 100% of the company is going to receive this training.
74
And to understand that the very, very important thing is that the human is going to be the one that decide, and the AI is going to be the medium for this decision.
75
KABIR NAGRECHA: You know, I think what we really push for on this team is questioning assumptions.
76
Questioning assumptions about what can and cannot be done by AI today.
77
What we assumed three months ago is impossible is now completely feasible, and that another part of our workflow could be completely automated.
78
That self-questioning, that self introspection, is a big part of critical thinking around AI.
79
THOMAS H. DAVENPORT: Hey, I'm a professor.
80
I really torture my students to get them to use AI in an appropriate fashion.
81
I'd say, "Every assignment that you do with AI," -- and I teach about AI, and of course I have to let them use AI, I've forced them use AI -- but they have to show the different prompts that they use.
82
They have to show how they added value to the output, the edits that they made.
83
They have to show that they checked the citations on the comments that the language model is making.
84
And they can't stand doing this.
85
One said to me, "It was easier when I could just paraphrase a Wikipedia article." But I think we all need to learn how to deal with AI in this way.
86
And as painful as it might be, as productivity-lowering as these activities are, I think they're critical to our success as a species.
87
(bright curious music)
✨ Vídeo recomendado
C13
159 frases
AI Is Stealing From You
Daily ConversationUniversity StudentsWorking Adults
C11
781 frases
How Do I Reverse Brain Rot?
Daily ConversationUniversity StudentsWorking Adults
C1762 frases
Building Anthropic | A conversation with our co-founders
Daily ConversationUniversity StudentsWorking Adults
C11
392 frases
Ultimate Claude Guide: How to Use Claude AI for Beginners in 2026
Daily ConversationWorking Adults
![The Great A.I. Debate: Should We Fear the Digital Revolution? [2015] | Intelligence Squared](https://img.youtube.com/vi/QWUeu3TLZN0/mqdefault.jpg)
C11
183 frases
The Great A.I. Debate: Should We Fear the Digital Revolution? [2015] | Intelligence Squared
Daily ConversationUniversity StudentsWorking Adults
C11
150 frases
Anthropic just proved AI isn't getting better
Daily ConversationWorking Adults
C12
39 frases
The Global Power Play
Daily ConversationWorking Adults
C186 frases
The End of Human Mathematics? AI's Recent Breakthrough -- And What's Next
Daily ConversationUniversity StudentsWorking Adults
C13
85 frases
AI Will End Philosophy, Physicist Claims
Daily ConversationUniversity StudentsWorking Adults
C12
155 frases
How to be a creative thinker | Carnegie Mellon University Po-Shen Loh
Daily ConversationUniversity StudentsWorking Adults
C17
75 frases
What AI can (and can't) currently do in a real lab | Jennifer Doudna
Daily ConversationUniversity StudentsWorking Adults
C11
105 frases
AI Governance Explained: Ethical, Transparent & Accountable AI for the Future
Daily ConversationUniversity StudentsWorking Adults
Vocabulário e dicas de fala para esta lição
Esta aula de conversação de nível C1 usa o vídeo “The AI Atrophy Problem How CIOs Fight It”. As palavras que mais se repetem: output, critical, thinking. Este vídeo tem 87 frases e 1651 palavras para praticar shadowing. A fala dura 9:55. O falante fala em ritmo natural, cerca de 166 palavras por minuto, próximo de uma conversa do dia a dia. 83% das palavras estão entre as 3.000 mais comuns do inglês; vale a pena estudar o restante antes de começar. 22 das 87 frases são perguntas, por isso é um bom treino para a entonação de perguntas e respostas curtas.
Vocabulário principal deste vídeo
As 15 palavras mais avançadas do vídeo, com pronúncia e significado:
| Palavra | Pronúncia | Significado |
|---|---|---|
| validate verbo | /ˈvæl.ɪ.deɪt/ | validar |
| distress substantivo | /dɪˈstɹɛs/ | desconforto |
| hallucinate verbo | /həˈl(j)uːsɪneɪt/ | alucinar |
| persona substantivo | /pɝˈsoʊnə/ | personagem |
| acceleration substantivo | /əkˌsɛl.əˈɹeɪ.ʃən/ | aceleração |
| prompt verbo | /pɹɑmpt/ | incitar, impelir |
| emerge verbo | /ɪˈmɝd͡ʒ/ | emergir |
| assumption substantivo | /əˈsʌm(p).ʃ(ə)n/ | suposição, pressuposição |
| aspiration substantivo | /ˌæspəˈɹeɪʃən/ | aspiração |
| automate verbo | /ˈɔ.təˌmeɪt/ | automatizar |
| compute verbo | /kəmˈpjuːt/ | computar, calcular |
| dialectical adjetivo | dialético | |
| extrapolate verbo | /ɛkˈstɹæp.əˌleɪt/ | extrapolar |
| introspection substantivo | /ɪntɹəˈspɛkʃən/ | introspecção, introspeção |
| paraphrase verbo | /ˈpæɹəfɹeɪz/ | parafrasear |
Phrasal verbs que você vai ouvir
| Palavra | Significado |
|---|---|
| take over verbo | tomar posse de, apoderar-se de |
| slow down verbo | desacelerar |
Frases que vale a pena repetir
Frases curtas e completas do vídeo que você pode usar na conversa do dia a dia:
- Let's hear what they had to say.
- Or are you sure you're not hallucinating?
- It's really going to be, can you ask better questions?
- Can you apply reasoning?
Gramática neste vídeo
As estruturas que o falante mais usa, com as palavras exatas do vídeo:
| Estrutura | No vídeo |
|---|---|
| Present perfect have/has + particípio passado — uma ação passada que ainda importa agora | has been · you've said · I've noticed |
| Voz passiva be + particípio passado — o foco está no que acontece, não em quem faz | been hallucinated · is needed · cannot be done |
| Orações relativas who / which + oração — informação extra sobre uma pessoa ou coisa | people who use · anyone who is |
Pronúncia para ficar de olho
O falante usa 29 contrações e formas reduzidas, como don't, I'm, you're. Diga-as na forma curta, do jeito que você ouve.
- Os sons de “sh” e “zh”: differentiator /dɪf.əˈɹɛn.ʃi.eɪ.tə/, acceleration /əkˌsɛl.əˈɹeɪ.ʃən/, assumption /əˈsʌm(p).ʃ(ə)n/, aspiration /ˌæspəˈɹeɪʃən/, introspection /ɪntɹəˈspɛkʃən/
- Palavras longas — acerte a sílaba tônica: differentiator /dɪf.əˈɹɛn.ʃi.eɪ.tə/, hallucinate /həˈl(j)uːsɪneɪt/, acceleration /əkˌsɛl.əˈɹeɪ.ʃən/, aspiration /ˌæspəˈɹeɪʃən/, extrapolate /ɛkˈstɹæp.əˌleɪt/
Sons difíceis para falantes de português:
- Consoante final — sem acrescentar um “i” depois: validate /ˈvæl.ɪ.deɪt/, hallucinate /həˈl(j)uːsɪneɪt/, prompt /pɹɑmpt/, automate /ˈɔ.təˌmeɪt/, compute /kəmˈpjuːt/
- /h/ — um sopro suave, diferente do “r”: hallucinate /həˈl(j)uːsɪneɪt/, hypothesis /haɪˈpɒθɪsɪs/
Como praticar com este vídeo
- Ouça o vídeo inteiro uma vez sem falar e anote as palavras que você não conhece.
- Comece na velocidade 0,75×, faça shadowing frase por frase e volte à velocidade normal quando ficar fácil.
- Grave a sua voz e compare com o original, prestando atenção a palavras como validate, distress, hallucinate.
O que é a Técnica de Shadowing?
Shadowing é uma técnica de aprendizado de idiomas com base científica, originalmente desenvolvida para o treinamento de intérpretes profissionais. O método é simples, mas poderoso: você ouve áudio em inglês nativo e repete imediatamente em voz alta — como uma sombra seguindo o falante com 1-2 segundos de atraso. Pesquisas mostram melhora significativa na precisão da pronúncia, entonação, ritmo, sons conectados, compreensão auditiva e fluência na fala.