跟讀練習: Therapy for the Vibe-Coded Brain - 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。
126 句
如果句子過短或過長，請點選 Edit 進行調整。
1
There is a rat in your brain that craves only dopamine and thinks that 10 minutes is a long time.
2
That rat may not like this video.
3
This is because it is slightly more educational than usual.
4
I would even recommend taking notes.
5
I tried to make it fun, but I'm not Mr. Beast and I don't really want to be.
6
So this is a video about boredom and difficulty and learning to code.
7
And if there's anything I really wanted to teach you, it's that it is possible and advisable to So ignore that rat as best you can.
8
Last video, I talked about an experiment where novices were observed as they completed a beginner programming task.
9
Put simply, some of them were bad and some of them were good.
10
The difference was mostly in how much they actually thought about it before trying to solve it.
11
Shocking.
12
The thing is, if you look at what the students turned in, it's all pretty decent.
13
Every single student found a solution within the 30 minutes given to them, thanks to their AI tutor.
14
The thing about the study that made me a bit worried is
15
that the ones who had offloaded all problem solving to AI were completely unaware of their own incompetency.
16
They thought that they had solved it mostly on their own.
17
You've probably heard the term vibe coding to describe this practice of using AI without fully understanding the code it produces.
18
It's pretty great for doing, and pretty terrible for learning.
19
The following video is about how to actually use and develop the most important tool in programming.
20
Your brain.
21
What do you do when you get stuck on a programming problem?
22
Before AI tools, your options were look it up on Stack Overflow, ask a friend if you have any,
23
or pause to think about it.
24
Since the introduction of AI to answer every possible question that we have, it's increasingly the thing that we turn to when we get stuck.
25
Typing out a prompt and reading something feels more productive than an empty pause.
26
But is it?
27
In the experiment, we see that subject 7 read the question and started coding.
28
was distracted and overwhelmed by an autocomplete suggestion.
29
They abandoned their code, opened ChatGPT, and asked it what to do.
30
See what happened there?
31
They interpreted this moment of confusion as a sign that they wouldn't be able to solve the problem on their own.
32
So within 10 seconds of starting, they're already having the AI think for them.
33
And again, if you're a professional developer and you're a little eepy, then sure, vibe code away.
34
But if you're a student in school, your whole purpose is to learn and develop your thinking.
35
Give yourself that minute to think and try to do it on your own.
36
Otherwise, my friends, smooth brain is coming for us all.
37
Pauses in learning aren't necessary, even if they feel unproductive.
38
Struggling, getting stuck, getting frustrated, these aren't signs that you're not going to be able to solve the problem.
39
They are what solving a problem feels like most of the time.
40
Speaking of what things feel like, let's talk about psychology.
41
Researchers use a lot of words that most people don't need to know, but if there's one which I think should become common usage, it's the concept of metacognitive awareness.
42
Granted, the reason that it hasn't is probably because metacognitive awareness is not a very sexy term, but the idea is pretty simple.
43
It's about your ability to recognize common traps in your thinking and effectively problem-solve your way out of them.
44
In the example I just mentioned, the subject had just begun problem-solving when they were distracted by an AI autocomplete suggestion.
45
This is named the interruption trap by the researchers in the paper
46
and it's a metacognitive difficulty that is specifically introduced by AI use.
47
Real talk, when I first read this paper I was kind of mad.
48
This list is a such a succinct description of everything you struggle with
49
when you're first wrapping your head around learning to program.
50
If I'd known these from the beginning I think I probably would have felt a lot less confused and frustrated with myself.
51
And I wouldn't be the only one.
52
Research has shown that teaching beginners about common metacognitive difficulties like the interruption trap produces immediate and lasting changes to learners productivity,
53
independence, and confidence.
54
Given that, it might be good to take some time to learn them.
55
This is where I recommend you start taking notes.
56
The first five are forming, dislodging, assumption, location, and achievement.
57
Forming.
58
Right question, wrong answer.
59
You understand the problem, but you're using the wrong approach to solve it.
60
In the experiment, students were asked to write a program
61
that reported whether a user input a sequence of majority positive or majority negative numbers.
62
Participant 1 seemed to understand the problem in their verbal statements, but started writing code to sum the numbers, which was a flawed approach.
63
Right question, wrong answer dislodging or stuck in a rut even
64
when you realize your approach isn't working you struggle to change it
65
after participant 4 coded a complete solution to determine whether numbers were even odd they tested and edited it repeatedly
66
still confused why it wasn't working.
67
They failed to make the jump from knowing whether a number was even or odd, to counting even and odd numbers in a list.
68
Assumption.
69
Wrong question, right answer.
70
You perfectly solved a problem, just not the one that you were supposed to.
71
Participant 21 confidently declared variables num1, num2, num3, and num4, and wrote separate code blocks to handle exactly four numbers.
72
The solution worked, unless you wanted more than four numbers.
73
So close, but so far.
74
You skip important problem-solving steps in the beginning and think you're almost done, only to realize that you skipped something crucial, like a loop or a data structure.
75
Here participant 9 wrote a complete set of input and output statements without including a loop.
76
It took until testing for them to realize their omission, where they had to majorly rethink their program structure.
77
A band-aid on a broken bone.
78
They've written a lot of code and keep making small fixes in hopes that it'll start working, but actually it needs a total overhaul.
79
This sounds similar to location, but this doubling down on a bad strategy comes later in the problem solving process.
80
Participant 1 just kind of threw everything they knew as a problem without thinking through the logic.
81
They had nested for loops and conditionals and nested conditionals, and their code needed major structural changes.
82
Even after the AI suggested an alternative way to solve the problem,
83
they skimmed over that suggestion and kept tinkering instead of recognizing that starting over was probably the best next step.
84
Wow, turns out that when we start learning something new, we're usually bad at it.
85
Shocking.
86
It's easy to see where these participants went wrong from an outside perspective, but much harder to notice them when you're in the middle of the problem.
87
Remember, the goal isn't to never encounter these difficulties, it's to become aware of them in yourself as quickly as possible.
88
In an age before LLMs, this video would be over.
89
Those would have been the five metacognitive difficulties associated with learning to code.
90
Unfortunately, researchers found that rather than helping with these challenges,
91
AI tools actually introduced three new metacognitive difficulties for beginners.
92
Namely, progression, falling behind without noticing.
93
You think you're keeping up with the course, but because AI assistants can generate work in code that surpasses your current understanding, you're falling behind and you don't realize it.
94
This happened to participant 8, who struggled to set up a simple while loop correctly and needed the AI to tell them to initialize their variables,
95
despite these being concepts that they should have learned earlier in the course.
96
In reality, this student was probably several weeks behind in the core material without realizing it.
97
In the post-test interview, they said that the LLM was helpful in validating their own ideas.
98
In practice, though, they didn't have any.
99
They used the LLM to generate a passable solution, skipping the thinking part.
100
Interruption.
101
Distracting pop-ups.
102
Every time you try to to focus, an AI code completion tool throws a suggestion at you, breaking your train of thought.
103
This one's pretty obvious, but it can be a major problem for beginners.
104
Learners have to spend a lot of brain power to pause and think through a solution, and getting interrupted can really derail that mental process,
105
not to mention the strain of reading unfamiliar code.
106
It's worth noting that in the experiment, non-struggling students mostly ignored the AI autocomplete suggestions.
107
If you've made it this far, congrats for taking this step to learn programming the right way.
108
The final metacognitive difficulty introduced specifically by AI tools is...
109
Mislead.
110
Following bad advice.
111
You trust a suggestion from an AI, a tutorial, or even your own guess that seems right but actually takes you in the wrong direction.
112
In the study, participant 11 accepted a suggestion from the AI assistant
113
which pushed them in the direction of summing positive and negative numbers rather than counting them.
114
Thankfully, they were able to backtrack and fix the mistake, but it cost them a lot of extra time and energy.
115
At the moment, most AI coding tools are not built with these challenges in mind.
116
Even if they were, you are the sole expert of your own mind, and there's only so much an external resource can do to see and protect you from the traps in your thoughts.
117
By taking a bit of time to learn about these thinking traps, you're already scientifically better off.
118
But try not to let these lessons slip away.
119
Find your own way to remind yourself of them while you're coding.
120
Write them on a post-it note, point them out to someone else when you're pair coding, respectfully, and try to reflect on which ones you might be the most susceptible to.
121
Most of all, I encourage you to remember to enjoy the process of learning to program, even if it can be a bit annoying sometimes.
122
I'll also remind you that we have free licenses for education
123
and free programming courses you can complete right in your IDE at academy.jefferns.com.
124
Please subscribe if you enjoy these videos.
125
They take a lot of work and I'd love to be able to continue making them.
126
That's all for this one and enjoy your learning.
FOR YOU
推薦影片
C12
171 句
Learn 10X Faster In 10 Minutes
Daily ConversationUniversity StudentsWorking Adults
C11
78 句
The last real programmers are dying out...
Daily ConversationUniversity StudentsWorking Adults
C11
167 句
Mental Debugging: How to Build Resilience After Failure
Daily ConversationUniversity StudentsWorking Adults
C110
93 句
Some brains are built for philosophy. Is yours?
British AccentDaily ConversationUniversity StudentsWorking Adults
C11
111 句
Spec-Driven Development: AI Assisted Coding Explained
Daily ConversationUniversity StudentsWorking Adults
C11
347 句
Doctor Explains: How to Train Your Brain to Crave Doing Hard Things
Daily ConversationUniversity StudentsWorking Adults
C13
159 句
AI Is Stealing From You
Daily ConversationUniversity StudentsWorking Adults
B21
12 句
How I’d Learn Coding in 2026 (Without Wasting Years)
Daily ConversationUniversity StudentsWorking Adults
B21
72 句
Fight brain-rot with Worked Examples
Daily ConversationUniversity StudentsWorking Adults
B21
169 句
Can This Vibe Coder Beat a Senior Developer?
Daily ConversationUniversity StudentsWorking Adults
B2298 句
Forget Your Reading Journal, Do This Instead
Daily ConversationUniversity StudentsWorking Adults
B21
331 句
Dear Higgsfield, I reverse engineered you in 12 hours
Daily ConversationUniversity StudentsWorking Adults
這個影片適合誰？
如果你是想提升英語口語的學習者，尤其是喜歡通過看影片學英語的朋友，這個影片超合適！它不僅內容有趣，還包含大量實用表達，非常適合用來練習shadow speech（影子跟讀法）。無論是程式設計新手還是想鍛鍊思維的人，都能在學英語的同時get到乾貨，一舉兩得！
值得"偷學"的表達和習語
- vibe-coded：字面意思是"氛圍編碼"，影片裡指依賴AI生成程式碼卻不理解原理的行為，類似"摸魚式編碼"。
- smooth brain：不是"光滑的大腦"，而是調侃人"不動腦子"，比如過度依賴AI會變"傻"。
- eepy：口語化表達，指"有點累、沒精神"，比"tired"更俏皮。
這些表達生動又接地氣，記下來能讓你的口語更native！
如何練好口音？試試shadow speak！
影片 speaker 的發音清晰，語速適中，超適合英語口語練習。建議用shadowspeaks方法：播放一句，暫停，立刻跟讀，模仿語調、重音和節奏。比如"Pauses in learning aren't necessary"，注意"pauses"的濁化和"necessary"的弱讀。多練幾遍，你會發現自己的口語越來越流暢！
記住，學習英語和程式設計一樣，別怕"卡殼"。哪怕慢一點，也要多思考、多模仿，這樣才能真正提升。現在就開啟影片，開始你的shadow speech之旅吧！
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 |
|---|---|
| 被動語態 be + 過去分詞 — 強調發生了什麼，而不是誰做的 | were observed · was distracted · were distracted |
| 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 | You've probably heard · has shown · They've written |
| 定語從句 who / which + 從句 — 補充說明人或事物 | ones who had · numbers, which was |
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。