跟读练习: How I use AI to learn engineering (anti brain rot studying) - 通过视频学习英语口语

选择练习模式

每种模式都会打开全屏练习室。

197 句

如果句子过短或过长，请点击 Edit 进行调整。

Learning hard things used to take me weeks.

I used to look at slides, I used to look at these terrible lecture recordings from my professors with really bad audio

and then I would look at Indian men on YouTube giving me tutorials.

Now AI gets me unstuck in minutes and it's not by doing the work for me, it's by showing me the concepts in a visual manner,

cutting out all the noise from the signal and giving me those most important resources to learn ASAP.

Hi I'm Alex, I graduated with an electrical engineering degree two years ago i landed internships in ai hardware software engineering

and now i work as a member of technical staff at

a really cool ai startup working on training of large language models

so i literally learn hard things for a living here are

the four ways i use ai to learn at an incredible speed

and how you can steal this for your own purposes

and your educational needs let's get into the video back in the day

when i was a little kid i remember going on youtube

and watching free blue one brown he's a really cool mathematics youtuber

and he had fabulous fantastic exceptional presentations of maths in a

very visual manner he used an incredible like python tool

that he built to just make maths beautiful

and guess what now we can do this with anything we want not just maths

but with engineering too And I do this at work now

and it's honestly astonishing how good AI models are at showing us really stuff in an elegant manner.

Let me give you an example.

Imagine you're a mechanical engineering student and you need to do some CAD

but you don't really know what you should really even design and you don't really even know where to start.

Well, what you can do now is fire up something like claude code

or gpt codex

and start prompting away let's say you wanted to build a

robotics project you can type in something like hey claude i want to build a robotics project

but i'm not really too sure where to start give me

three ideas for a interactive presentation pitch them to me tell them why this one is the best one

and why should i build it and make a 3d model out of it

so i can see it this is super cool it will

generate a full presentation with a 3d model you can have

a look at it spin it around zoom in zoom out

try it out it's super super cool another example let's say

you're an electrical engineering student you have some electronics to do i remember for example i had an stm32

and i had to connect it to like an st link v3 to do some power measurements

and i remember it was very very confusing how to connect

these together the data sheet was terrible imagine a situation where you could just visualize how to connect stuff together

and you can just get to your work start getting your coursework done

that would be incredible huh so you can do

that now take a photo of your hardware

and tell claude make me an interactive dashboard

or presentation showing me how to wire the stm32 together to this other piece of hardware

so i can take power readings

and it will design for you this really cool interactive tool to show you how you wire this to

that this to that this to that step by step

so it's just so simple for you to follow

and it's incredible please try it out by the way

if you're going to do this i do recommend spending more of those expensive tokens on the better models

because they are much better at the 3d world

but this is just one element of it right this is just visualization that's a great way to learn

but in and of itself it's not enough and there are

so many resources out there which leads me to the next point of my video

and by the way i'm giving out my cv template for

free it got me interviews at google samsung hsbc my current

role as a member of technical staff at this ai startup

it's a really good cv template just go to alexgonick .com

resume link in the description go grab it use ai to decide what is actually worth reading

so in my job a quite a big element of it

can be reading research papers trying to figure out what's this

new little piece of architecture that's been slid into this transformer to make it operate in a cool way

but there's so much to read and you can't read everything

so you need to decide what you're going to read what

you're going to study what what information do you think is

valuable enough what's valuable for your time to actually go into your coranum

and i think the equivalent is at university i remember very

clearly you go into your lectures in the first few classes

the professor says yeah i have these book recommendations please read

please check out this book this book this book this book

they're all like 1 000 pages each one was written in

1970 what do you read should you even read them right

you should not read all five books that's for sure it

is going to be a complete waste of your time use

ai as a filter as a as a way to really

find exactly what you need from these books hey i got

this book recommendation list i'm not sure what this is about

i'm studying my course my course has this syllabus what book would be the most relevant to look at

and then once it's giving you a response guess what it's

probably not reasonable to read the whole book it's probably only one

or two or three chapters

that are actually going to be useful zoom in again ask it

which chapters of this book are like the best what's the 80 20 of this book

that will get me up to scratch with microwave communication systems

or stochastic control go and really do that filtering

so here's prompt for you

that you should use here's my here's my module slash class syllabus

and here is the reading list which books

and chapters from this book are actually worth my time considering my specific curriculum

and the exams i'm going to sit

which tutorial problems are worth doing this prompt is super super

good it's really really good i strongly recommend you use it it will save you a lot of time

if you are that type of person that likes to go a bit deeper

and just just read and capture some of

that information i on the other hand

when i have a huge list of papers

that potentially i want to read

and maybe i found them on twitter i just tell it

hey i found these five papers i want to get to

i want to get a good update i want to learn

100

a bit more on this specific side of reinforcement learning

101

which of these papers is the best to read

102

which is the most legitimate and it tells me

103

and then i go and then i read it

104

and it's great i don't have to spend two hours reading papers

105

when i can just spend 15 minutes

106

but i know how it is to be a student i was there just two years ago

107

and sometimes you don't have a you don't have all this

108

fancy time to do a bunch of reading sometimes you got an exam next week

109

and you need to get a lot of stuff done so

110

that leads me to my next point which is find the 20

111

that is essential to do so story time when I was

112

when I was around 22 I was this was around 2023 I was an intern

113

and hardware intern chat gbt3 just came out

114

and I remember prompting it with something like hey could you

115

explain to this to me like I'm five years old

116

and I remember that was like an incredible thing to write

117

and it did a fantastic job actually

118

and it really made me fascinated by AI

119

and this prompt was the base for what i'm about to kind of speak to you about

120

and it's that usually in a topic

121

that you are working on you're studying you're building a project

122

on you don't need to know everything you don't even need

123

to know half of it you only probably need to know

124

20 to get moving to get going to get unstuck to go forwards

125

when i was working on my thesis in graph neural networks

126

for simulating airway flow in the lungs i got stuck okay i was doing all

127

that self -learning by myself

128

and i really really really didn't know what to do in certain moments

129

and in those moments you just need to just get to

130

the next step you know in deep learning you've got forward

131

passes you've got activation functions you've got soft max tan h

132

and relu you've got um you've got inputs outputs you've got matrices you've got

133

so much stuff going on so much stuff

134

that you can go really deeply on you've got different architectures

135

you've got transformer you've got neural normal normal neural networks you've

136

got graph neural networks what's the difference oh my god i

137

can go on forever is there's too much there's too much to study there's too much to learn

138

and i needed to really filter that down and i did i identified

139

that i just need to get a training loop going see a few epochs feed it some training data

140

and when i broke down

141

that problem into something way more simple i just focused on studying

142

that little tiny section on how do i use pytorch to run some data for a neural network

143

and i figured out and i kept going

144

and i still didn't know what how back pocket back propagation really works like back then

145

but it was enough to get going so use ai to figure out

146

what is the thing that you really need to know to get going then use

147

that llm to to to get that resource to to get you to study

148

that 20 out of the 100 that you really need

149

and then get it done and then go back

150

and learn the rest if you want to but don't think

151

that you need to study everything to get going you do not

152

and it's going to cost you many many hours

153

if you think that's the case and that's how you learn things quickly and once you're unstuck Well, you need the best things to learn from, which leads me to my next point.

154

By the way, if you're an engineering student and you're seriously stuck landing an internship, then you've landed in a great spot.

155

I have a coaching program where I work with ambitious engineering students to get you your dream role.

156

A, we create a fantastic portfolio of projects for you.

157

B, we freshen everything up, your LinkedIn, your CV, so it's actually serious.

158

And C, you get those interviews, you smash them and you get that dream job.

159

If you're interested in this, link in the description.

160

Make AI find you the best resource.

161

So I started to get into CAD recently.

162

It's actually a very interesting part of engineering.

163

And I know very little about it.

164

Okay, very, very little.

165

I did very little CAD at uni and I tried to avoid it.

166

I was really bad.

167

But now, because of my job, actually, I'm really involved in doing computer -aided stuff

168

and i've started using ai to really go deep

169

and find these fantastic resources in the cad world i

170

and i would never been able to figure this out by

171

myself okay is i got ai to run these sub agents to pretty much scatter the whole of internet

172

and find me really cool um data like cool templates of CAD stuff.

173

I found NASA 3D models of Mars rovers.

174

I found satellite files in .stl that I can just upload into my CAD software.

175

I could not believe it.

176

There is so much fantastic 3D models for free.

177

You can literally build your own Mars rover for free.

178

Everything is there open sourced download the files print it out

179

wow guys it's it's incredible i was mind blown okay

180

and i will and i would have never been able to figure

181

that out without the use of ai

182

so remember you searching through the internet for 10 hours is

183

probably still going to be slower than 10 sub agents crawling the internet

184

and they'll find you much better stuff

185

if you just specify exactly what you're looking for just make sure you click on the links

186

and double check that everything is legit

187

because ai does hallucinate still just a big fyi

188

so a little recap make stuff visual okay filter out all the bs

189

which is on the internet 80 20 okay figure out what the 20

190

that you need to know in your head right now to get stuff done

191

and finally make sure when you are studying you use the best resource to do

192

so and use ai to do these four things

193

because it's better at you than us okay it's better now go learn something hard now

194

if you want to learn more about ai tools

195

that are actually good for you as an engineering student then

196

go check out this video over here i go through my favorite tools

197

that i would use as an engineering student if i was there now in 2026 and beyond that's everything see ya

下载应用

流利口语所需的一切

AI 发音评分逐句打分

音标练习掌握每个音

词汇学习建立你的词库

词汇游戏边玩边学

扫码

Download on theApp Store Get it onGoogle Play

FOR YOU

你能学到什么

这段视频能帮你提升两项关键英语口语技能：一是清晰表达复杂想法的能力，比如像演讲者那样用简洁语言解释AI学习工具的作用；二是自然运用连接词，比如视频中频繁出现的“and”“but”，让口语更流畅。此外，还能学会如何在对话中合理插入例子，使表达更有说服力。这些技能对英语口语练习至关重要，能让你在日常交流或正式演讲中更自信。

注意这些发音现象

视频中存在不少连读和弱读现象，需要重点关注。比如“used to”常弱读为/ˈjuːst tə/，而非逐个发音；“look at”会连读成/lʊk æt/；“start getting”则是/stɑːt ˈɡetɪŋ/，前一个单词的尾音与后一个单词的首音自然衔接。这些连接 speech 的技巧能让你的发音更接近母语者，也是提高英语发音的重要环节。

像母语者一样表达

要模仿演讲者的节奏和重音，需注意以下两点：一是节奏快慢有致，在解释关键概念（如“AI模型”“3D模型”）时适当放慢，而在讲述例子时加快语速，形成自然的语流；二是重音突出重点，比如“incredible”“astonishing”等形容词要重读，以强调情感。建议使用 shadowing site 进行英语影子跟读练习，即边听视频边同步跟读，模仿其语调、节奏和重音。长期坚持 shadow speech，能有效提升口语的流利度和自然度。

视频中的语法

说话人最常用的结构，并附上视频中的原话：

结构	视频中的用法
“used to” used to + 动词 — 过去的习惯或状态，现在已不再如此	used to take · used to look
现在完成时 have/has + 过去分词 — 过去发生但与现在仍有关联的事	that's been · you've landed · i've started

什么是跟读法？

跟读法 (Shadowing) 是一种有科学依据的语言学习技巧，最初开发用于专业口译员的培训，并由多语言者Alexander Arguelles博士普及。这个方法简单而强大：您在听英语母语原声的同时立即大声重复——就像是一个延迟1-2秒紧跟说话者的影子。与被动听力或语法练习不同，跟读法强迫您的大脑和口腔肌肉同时处理并模仿真实的讲话模式。研究表明它能显着提高发音准确性，语调，节奏，连读，听力理解和口语流利度——使其成为雅思口语备考和真实英语交流最有效的方法之一。

影子跟读法: 阅读完整分步指南 →

跟读练习: How I use AI to learn engineering (anti brain rot studying) - 通过视频学习英语口语

选择练习模式

下载应用

推荐视频

A Practical Guide To Becoming An AI Engineer (2026)

How I Learned Procedural Generation

Learn 10X Faster In 10 Minutes

MIT Introduction to Deep Learning | 6.S191

We Let $200 GPT-Pro Think for 7 Days on Unsolved Math

Harnessing generative AI in architecture, engineering & construction

Your Life as Every Industrial Engineer Rank

What AI can (and can't) currently do in a real lab | Jennifer Doudna

Dear Higgsfield, I reverse engineered you in 12 hours

LangChain in 5 Minutes (Explained Clearly)

DeepSeek R1 Explained in 2 Minutes

Fight brain-rot with Worked Examples

你能学到什么

注意这些发音现象

像母语者一样表达

视频中的语法

什么是跟读法？