跟讀練習: Grok 4 pushes humanity closer to AGI… but there’s a problem - 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。

43 句

如果句子過短或過長，請點選 Edit 進行調整。

Yesterday, after achieving a, quote, ludicrous rate of progress, Elon Musk released his AI chatbot and large language model, Grok4, and claims it's the smartest AI in the world,

along with the Trust Me Bro benchmarks to back it up.

It can achieve perfect SAT scores every time, and outperforms almost every grad student in every discipline.

Vibecoders have been dropping all kinds of crazy demos with it, like this 3D first-person shooter built in four hours, and Elon himself claims it's even better than Cursor.

All you have to do is copy and paste your entire codebase into it.

In addition, Super Grok 4 Heavy can run in parallel to solve complex problems, while your obsolete monkey brain looks in awe at this beautiful futuristic UI.

It all sounds amazing, but there's just one problem.

Grok is literally MechaHitler, or at least that's what it's been calling itself recently, while offering unprompted praise to the original emo kid Adolf H.

Yes, the Austrian painter who died in Argentina in 1962.

Despite this controversy though, Grok appears to have pulled ahead in the race to AGI.

In today's video, we'll put it to the ultimate test

and find out if XAI just cracked the final solution to artificial intelligence.

It is July 11th, 2025, and you're watching The Code Report.

Elon Musk might have more haters than anyone else in the world right now.

Libs hate him for going full MAGA, while MAGA hates him for accusing Trump of being on the Epstein client list, a list that was on the Attorney General's desk, but now magically never existed.

It turns out this poor guy was just a math teacher with no clients at all.

The haters want to see Elon fail, but Grok 4 is too impressive to ignore.

If we're to trust these benchmarks, its reasoning capabilities are far ahead of the other top models.

Most notably on the ARC AGI benchmark, it's not only outperforming other models, but also doing it at a lower cost.

And XAI is scaling up very aggressively, like they're even shipping a power plant from overseas because they can't get it done fast enough in the United States.

The thing is though, every model nowadays is cooked to look as good as possible on benchmarks.

And the one true test is to have it solve real problems in your own life.

You can use Gronk 4 for just $30 per month, or if you're not already broke from OpenAI Pro, Claude Max, and Gemini Ultra,

you can use Super Grok4 Heavy for just $300 per month, a version with higher rate limits and the ability to run multiple agents in parallel.

But one problem in my life is building Svelte 5 apps with runes, so let's see if we can make Grok4 build a simple to-do app with this tech.

I've tried this prompt with every other AI tool out there, but none have been able to satisfy me.

When prompted, Grok did a ton of research.

It went to the documentation, it went to Reddit, GitHub, and even watched YouTube videos.

And the end result was a full working demo that did use the new runes feature in Svelte 5.

However, when I looked a little more closely at the code, I noticed it was using some legacy syntax that required some manual debugging.

Overall, Grok's coding capabilities seem to be on par with the other big models, but it's missing a CLI tool like CloudCode.

However, if it's as good as they say it is, why can't it just build its own CLI tool?

Well, actually it can, and that's exactly what this guy did if we're truly advancing into the singularity, AI can and should be building all of its own tooling at this point.

Pretty cool, but if you're still afraid to use it because of Hitler, you'll be happy to know that Elon said it was manipulated into saying this.

Maybe that's true, maybe not, but in general, Grok has far fewer guardrails on offensive speech compared to other mainstream models,

and this gives the end user the ability to steer it in unique ways that some might find offensive.

AI is now writing more of our code than ever, but according to a recent Microsoft study, it still sucks at debugging, which is why you should check out Sentry, the sponsor of today's video.

They just launched a new AI debugging agent called Sear, which developers claim is actually good and capable of fixing complex issues in one shot.

That's because unlike other AI debugging tools, Sear can access all the context that it gets from your codebase, like error data, logs, and stack traces,

allowing it to pinpoint the root cause with over 94% accuracy.

Using this context, it automatically debugs the root cause of your issue and opens a pull request with a fix.

Try out SEER for free today at sentry.io slash fireship.

This has been the Code Report, thanks for watching, and I will see you in the next one.

下載應用

流利口語所需的一切

AI 發音評分逐句打分

音標練習掌握每個音

詞彙學習建立你的詞庫

詞彙遊戲邊玩邊學

掃碼

Download on theApp Store Get it onGoogle Play

同頻道

FOR YOU

本課的詞彙與口語要點

這節 B2 級別的口語課以影片“Grok 4 pushes humanity closer to AGI… but there’s a problem”為素材。影片中反覆出現的詞有：Grok, model, Elon, debugging, benchmark。這段影片共有 43 個句子、841 個單詞可供跟讀。講話部分時長為 3:54。說話人語速較快，每分鐘約 216 個詞，會出現較多連讀和弱讀。只有 78% 的單詞屬於英語最常用的 3,000 詞，詞彙難度較高。

影片中的重點詞彙

影片中最難的 15 個單詞，附發音和釋義：

單詞	發音	釋義
debug 動詞	/diːˈbʌɡ/	除錯 /除錯, 除錯 /除錯
codebase 名詞		程式碼庫 /程式碼庫
sentry 名詞	/ˈsɛn.tɹi/	哨兵, 步哨
demo 名詞	/ˈdɛm.oʊ/	示範 /示範
chatbot 名詞	/ˈtʃætbɑt/	聊天機器人 /聊天機器人
guardrail 名詞	/ˈɡɑɹdˌɹeɪl/	護欄 /護欄
ludicrous 形容詞	/ˈluː.dɪ.kɹəs/	滑稽的, 荒唐的
syntax 名詞	/ˈsɪn.tæks/	句法
obsolete 形容詞	/ˈɑbsəlit/	淘汰, 過時 /過時
manipulate 動詞	/məˈnɪp.jʊ.leɪt/	操縱
accuse 動詞	/əˈkjuz/	指責 /指責, 控告
steer 動詞	/stɪɹ/	掌舵, 駕駛
pasta 名詞	/ˈpɑ.stə/	мянфан, 義大利麵 /義大利麵
painter 名詞	/ˈpeɪntɚ/	畫家 /畫家
satisfy 動詞	/ˈsætɪsfaɪ/	滿足 /滿足

影片中出現的短語動詞

單詞	釋義
check out 動詞	退房
find out 動詞	找出, 查明
try out 動詞	試用 /試用
turn out 動詞	原來 /原來

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
現在完成進行時 have/has been + -ing — 以前開始、現在仍在繼續的動作	have been dropping · it's been calling
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	I've tried · has been
被動語態 be + 過去分詞 — 強調發生了什麼，而不是誰做的	is cooked · was manipulated

需要注意的發音

說話人用了 11 次縮略和弱讀形式，例如 you're, can't, we're。請按聽到的簡短形式來說。

“sh” 和 “zh” 音: benchmark /ˈbɛn(t)ʃmɑːk/, documentation /ˌdɑkjəmɛnˈteɪʃən/, shooter /ˈʃuːtəː/
長單詞——注意重音位置: capability /ˌkeɪ.pəˈbɪl.ɪ.ti/, futuristic /ˌfjut͡ʃəˈɹɪstɪk/, singularity /ˌsɪŋɡjəˈlæɹətɪ/, manipulate /məˈnɪp.jʊ.leɪt/, documentation /ˌdɑkjəmɛnˈteɪʃən/

中文母語者容易讀錯的音：

詞尾子音 — 要讀清楚，後面不要加母音: debug /diːˈbʌɡ/, benchmark /ˈbɛn(t)ʃmɑːk/, prompt /pɹɑmpt/, chatbot /ˈtʃætbɑt/, futuristic /ˌfjut͡ʃəˈɹɪstɪk/
/r/ 和 /l/ 的區別: guardrail /ˈɡɑɹdˌɹeɪl/, ludicrous /ˈluː.dɪ.kɹəs/, singularity /ˌsɪŋɡjəˈlæɹətɪ/
/æ/ — 嘴張大，不要讀成 /e/: chatbot /ˈtʃætbɑt/, magically /ˈmæd͡ʒ.ɪ.kli/, singularity /ˌsɪŋɡjəˈlæɹətɪ/, syntax /ˈsɪn.tæks/, satisfy /ˈsætɪsfaɪ/

如何用這段影片練習

先完整聽一遍影片，不要開口，記下不認識的單詞。
先用 0.75 倍速逐句跟讀，熟練之後再回到正常速度。
錄下自己的聲音並與原聲對比，特別注意 debug, codebase, sentry 這類單詞。

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。

影子跟讀法: 閱讀完整分步指南 →

跟讀練習: Grok 4 pushes humanity closer to AGI… but there’s a problem - 通過影片學習英語口語

選擇練習模式

下載應用

Brainf**k in 100 Seconds

Did the AI bubble just pop?

PostgreSQL in 100 Seconds

React Native in 100 Seconds

Did Google just dethrone OpenAI and Anthropic?

A $6.3 billion open-weight model just got embarrassed by the French...

Every app on Earth is now open source... whether it likes it or not

OpenAI’s lead engineer was secretly a spy...

The most interesting "hack" in history...

Mind-bending new programming language for GPUs just dropped...

China just dropped a new frontier model... and nobody can explain how

The last real programmers are dying out...

Intel chips can’t possibly be this bad… 100% crash rate?

Docker in 100 Seconds

OpenAI’s new browser feels familiar…

Firefox and Tor hit with 9.8 critical level exploit

5 open source tools that replaced my $320/mo AI stack...

C++ in 100 Seconds

Rust in 100 Seconds

Linux in 100 Seconds

I spent 3 days at MIT... the robot hype is worse than you think

Anthropic begged the world to stop AI… then shipped this

How I'd Learn AI From Scratch in 2026 (skip the useless 80%)

Is AI Doom Real... or Just Good Marketing?

Can This Vibe Coder Beat a Senior Developer?

What Happens If China Invades Taiwan? Elon Musk's AI Warning

Top five moments from CNBC's interview with OpenAI CEO Sam Altman

Codex Is Winning… So Why Am I Still Paying for Claude?

Claude Just Revealed AI's Biggest Problem

I Cooked Against Robots

FULL DISCUSSION: Google's Demis Hassabis, Anthropic's Dario Amodei Debate the World After AGI | AI1G

How the Algorithm Hijacked Monkey’s Brain (Why Monkey Can’t Learn Online)