跟讀練習: Did Google just kickstart the intelligence explosion? - 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。

54 句

如果句子過短或過長，請點選 Edit 進行調整。

In 1965, a straight British mathematician named I.J.

Goode, who had spent World War II fighting Nazis next to a gay Alan Turing in Bleschley Park, wrote that the first ultra-intelligent machine would be the last invention man ever needed to make,

because once an AI gets good enough to improve itself, every improvement makes it better at improving, which makes it better at improving, which, you get the idea.

This is called RSI, and it's been the wet dream of AI researchers ever since.

Well, last week, 33 researchers from ByteDance, Tsinghua, and a few other Chinese labs published a paper called The Last AI Built by Humans.

It laid out a five-stage roadmap for RSI, where in the final stage, the AI rewrites the process it used to improve itself, and we all get reassigned to the countryside for agricultural work.

Then on Sunday, Google DeepMind

and the University of Maryland responded by dropping a similar paper of their own called Wet Dream RSI.

In it, they claim that by turning an AI's old discovery logs into a simulator

and letting it dream up thousands of new search strategies inside of it, the AI got better at discovering things without anyone ever touching the model itself.

In today's video, we'll break down how Dream RSI works under the hood, and decide whether an agent being able to rewrite its own exploration policy is actually RSI,

or just more hype slop.

It is September 17th, 2026, and you're watching the Code Report.

Every time ai has made a mathematics breakthrough the process has been the same one

that alpha evolved popularized last year you take a coding agent handed a problem

and a scoring function then run a loop where it proposes a solution evaluates it reads the feedback

and tries again a few thousand times this process is what discovered the jacobian conjecture this summer

and what openai used to front run the Navier-Stokes problem earlier this month.

But the most interesting part of the process is one no one really talks about called exploration policy.

The idea is that at every step of the loop, there's a decision to make about what the agent tries next.

For example, say the loop is optimizing for a matching algorithm for horses.

If one attempt pairs up horses slightly better than others, does the agent keep building on it, or does it start over with something that could be better?

And if an attempt crashes before a single horse gets matched, is it the whole idea that's bad, or just the implementation.

Until this week, this exploration policy was hard coded into the loop by whoever set it up.

But what the DeepMind team figured out is that if you save everything from every attempt, like the code it wrote, the score it got, and whether or not it crashed,

you don't actually need to touch the model again to test a new policy.

You can just show the new policy the old runs

that are cached on the disk and let it decide where to go from there.

And because that costs nothing, the agent can test thousands of different policies against the same run

and then whichever one would have reached the best result in the fewest attempts.

Then it deploys that policy on the next run, saves that run too, and repeats the process again.

The paper calls this process Dreaming, and to test it, they pointed Gemini at eight different problems across algorithm design and mathematics,

then ran the same setup with a fixed policy to see if the Dreaming version could beat it.

I won't bore you with the TMBBs, but the most impressive one was that it wrote a lasso solver

that beats Python's standard machine learning library in about 300 tries, where the static policy needed 550, and the previous record holder needed about 51,000.

And because we're living in hell, the most interesting part was the prompt.

It basically begs the agent to read every past attempt before writing any code, to stop making tiny tweaks to the same idea over and over, and to pinky promise not to kill any processes.

So is any of this actually RSI?

By IJ Good's definition, no, because the whole point is that the thing doing the improving gets smarter each round.

In this case, the model that writes each new exploration policy is still the same Gemini, so it can never find a solution that it wasn't already capable of writing.

It just finds them faster and with fewer wasted attempts.

But with that said, that's also true of every AI math breakthrough we've had this year.

The Jacobian conjecture, Navier-Stokes, and progress on the Ryman hypothesis all came from static models wrapped in a custom harness,

with sub-agents, swarms, and orchestration doing most of the heavy lifting, and the weights only get better when a human goes back and trains the next model on what the swarm found.

So if you gave IJ mushrooms and convinced him that a human in the loop still counts, he might agree that this is the last invention man ever needed to make, but when he came down,

he'd probably argue that this is just a cool search algorithm with some caching, And that's why you need to check out Blacksmith, the sponsor of today's video.

It's a drop-in replacement for GitHub Runners that lets you run your GitHub actions twice as fast while costing 75% less.

And they just launched Codesmith, a cloud coding agent that knows your repos and CI runs, so you can ask it to build something from GitHub, the web, or Slack like I'm doing here.

I'm asking it to add a new AI provider to my app

and wire up the API key in my infrastructure repo and it can open a PR in each one,

then fix failing tests or address review comments without relying on messages between bots.

You can also ask it to recommend better runner sizes for your CI history and turn those changes into a pull request, so you're not renting a supercomputer to check your semicolons.

Try it out for free and get 3,000 GitHub Actions minutes at the link below.

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

這段影片共有 54 個句子、1063 個單詞可供跟讀。講話部分時長為 4:56。說話人語速較快，每分鐘約 215 個詞，會出現較多連讀和弱讀。只有 79% 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 15 個不太常見的單詞，附發音和釋義：

loop /luːp/ (名詞) — 圈, 環 /環. A length of thread, line or rope that is doubled over to make an opening.
exploration /ˌɛkspləˈɹeɪʃən/ (名詞) — 勘探, 勘察. The process of exploring.
mathematics /mæθ(.ə)ˈmæt.ɪks/ (名詞) — 數學 /數學, 算學 /算學. An abstract representational system studying numbers, shapes, structures, quantitative change and relationships between them.
algorithm /ˈælɡəɹɪðm̩/ (名詞) — 演算法, 演算法. A collection of ordered steps that solve a mathematical problem. A precise step-by-step plan for a computational procedure that possibly begins with an input…
discover /dɪˈskʌvə/ (動詞) — 發現 /發現, 發覺 /發覺. To find or learn something for the first time.
invention /ɪnˈvɛnʃən/ (名詞) — 發明 /發明. Something invented.
researcher /ˈɹiˌsɝ.t͡ʃɚ/ (名詞) — 研究員 /研究員, 研究者. One who researches.
breakthrough /ˈbɹeɪk.θɹuː/ (名詞) — 突破. An advance through and past enemy lines.
cache /kæʃ/ (名詞) — 快取 /快取, 快取. Such a store of physical supplies, placed by humans or other animals for practical reasons.
impressive /ɪmˈpɹɛsɪv/ (形容詞) — 令人印象深刻. Making, or tending to make, a positive impression; having power to impress.
argue /ˈɑɹ.ɡju/ (動詞) — 辯論 /辯論, 爭辯 /爭辯. To show grounds for concluding (that); to indicate, imply.
cloud /ˈklaʊ̯d/ (名詞) — йүн, 雲 /雲. A visible mass of water droplets suspended in the air.
infrastructure /ˈɪnfɹəˌstɹʌkt͡ʃə/ (名詞) — 基礎設施 /基礎設施. An underlying base or foundation for a building, organization, or system.
discovery /dɪˈskʌv.ɹi/ (名詞) — 發現 /發現. Something discovered.
custom /ˈkʌstəm/ (名詞) — 習慣 /習慣, 習俗 /習俗. Frequent repetition of the same behavior; way of behavior common to many; ordinary manner; habitual practice; method of doing, living or behaving.

影片中出現的短語動詞

break down (動詞) — 壞了 /壞了, 出故障. To stop functioning.
check out (動詞) — 退房. To record the departure or withdrawal of someone or something (such as guests, employees, books, etc.).
come down (動詞) — 下來 /下來, 降. To descend, fall down, collapse.
come from (動詞) — 來自 /來自. To have as one's origin, birthplace or nationality.
figure out (動詞) — 弄清楚. To come to understand; to discover or find a solution; to deduce.
go back (動詞) — 回去, 返回. To return to a place or state after having been there at a previous time.

需要注意的發音

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

“th” 音: mathematics /mæθ(.ə)ˈmæt.ɪks/, algorithm /ˈælɡəɹɪðm̩/, breakthrough /ˈbɹeɪk.θɹuː/, hypothesis /haɪˈpɒθɪsɪs/, mathematician /ˌmæθ.(ə.)məˈtɪʃ.ən/
“sh” 和 “zh” 音: exploration /ˌɛkspləˈɹeɪʃən/, invention /ɪnˈvɛnʃən/, cache /kæʃ/, implementation /ˌɪmplɪmənˈteɪʃən/, mushroom /ˈmʌʃˌɹuːm/
長單詞——注意重音位置: exploration /ˌɛkspləˈɹeɪʃən/, mathematics /mæθ(.ə)ˈmæt.ɪks/, infrastructure /ˈɪnfɹəˌstɹʌkt͡ʃə/, implementation /ˌɪmplɪmənˈteɪʃən/, intelligent /ɪnˈtɛlɪd͡ʒənt/

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 loop, exploration, mathematics 這類單詞。

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
現在完成時 have/has + 過去分詞 — 過去發生但與現在仍有關聯的事	it's been · has made · we've had
被動語態 be + 過去分詞 — 強調發生了什麼，而不是誰做的	is called · are cached

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: Did Google just kickstart the intelligence explosion? - 通過影片學習英語口語

選擇練習模式

下載應用

The last real programmers are dying out...

PostgreSQL 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...

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

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

Godot in 100 Seconds

Grok 4 pushes humanity closer to AGI… but there’s a problem

Array Map in 100 Seconds

Docker in 100 Seconds

The most interesting "hack" in history...

OpenAI’s new browser feels familiar…

10 weird OSS projects you need right now...

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

DHH has gone completely off the rails...

Rust in 100 Seconds

The Limitless AI Lie. The Bubble Is Slowly BURSTING.

The most interesting hack in history just got weirder...

Google’s AI Course for Beginners (in 10 minutes)!

Alexandr Wang on what matters in an age of abundant intelligence

A.I. ‐ Humanity's Final Invention?

Very excited to share that I now have a column in New Scientist

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

The Alignment Problem Explained: Crash Course Futures of AI #4

The Great A.I. Debate: Should We Fear the Digital Revolution? [2015] | Intelligence Squared

Introducing Automations: always-on coding agents

DeepSeek R1 Explained in 2 Minutes