Our last invention, the most powerful tool, weapon, or maybe even entity, artificial superintelligence.

我們最後的發明，最強大的工具、武器，甚至可能是實體--人工超級智能。

This sounds like science fiction, so let's start at the beginning.

這聽起來像科幻小說，讓我們從頭開始。

Intelligence is the ability to learn, reason, acquire knowledge and skills, and use them to solve problems.

智力是學習、推理、獲取知識和技能，並利用它們解決問題的能力。

Intelligence is power, and we're the species that exploited it the most.

智慧就是力量，而我們是最善於利用智慧的物種。

So much so, that humanity broke the game of nature, and took control.

以至於人類打破了大自然的遊戲規則，掌握了控制權。

But the journey there wasn't straightforward.

但這一旅程並不平坦。

For most animals, intelligence costs too much energy to be worth it.

對大多數動物來說，智力耗費的能量太多，不值得。

Still, if we track intelligence in the tree of species over time, we can see lots of diverse forms of intelligence emerge.

不過，如果我們對物種樹中的智能進行長期追蹤，就會發現出現了許多不同形式的智能。

The earliest brains were in flatworms 500 million years ago.

最早的大腦出現在 5 億年前的扁形蟲身上。

Just a tiny cluster of neurons to handle basic body functions.

只有一小簇神經元來處理身體的基本功能。

It took hundreds of millions of years for species to diversify and become more complex.

物種的多樣化和複雜化需要數億年的時間。

Life conquered new environments, gained new senses, and had to contend with fierce competition over resources.

生命征服了新的環境，獲得了新的感官，並不得不面對激烈的資源競爭。

But in nature, all that matters is survival, and brains are expensive.

但在大自然中，最重要的是生存，而大腦是昂貴的。

So for almost all animals, a narrow intelligence fit for a narrow range of tasks was enough.

是以，對於幾乎所有的動物來說，只要有適合狹窄任務範圍的狹窄智能就足夠了。

In some environments, animals like birds, octopuses, and mammals evolved more complex neural structures.

在某些環境中，鳥類、章魚和哺乳動物等動物進化出了更復雜的神經結構。

For them, it paid off to have more energy-consuming skills like advanced navigation and communication.

對他們來說，掌握高級導航和通信等更耗能的技能是值得的。

Until 7 million years ago, the hominins emerged.

直到 700 萬年前，類人猿出現了。

We don't know why, but their brains grew faster than their relatives.

我們不知道為什麼，但他們的大腦比他們的親戚長得快。

Something was different about their intelligence.

他們的智慧與眾不同。

Very slowly, it turned from narrow to general.

慢慢地，它從狹義變成了廣義。

From a screwdriver to a multi-tool.

從螺絲刀到多功能工具。

Able to think about diverse problems. 2 million years ago, Homo erectus saw the world differently from anyone before.

能夠思考各種問題。200 萬年前，直立人看待世界的方式與眾不同。

As something to be understood and transformed.

作為一種需要理解和改造的東西。

They controlled fire, invented tools, and created the first culture.

他們控制了火，發明了工具，創造了最早的文化。

We probably emerged from them around 250,000 years ago with an even larger and more complex brain.

我們可能是在大約 25 萬年前從它們中誕生的，擁有更大、更復雜的大腦。

It enabled us to work together in large groups and to communicate complex thoughts.

它使我們能夠在大型小組中一起工作，並交流複雜的想法。

We used our intelligence to improve our lives, to ask how things worked, and why things are the way they are.

我們用自己的智慧來改善生活，詢問事情是如何運作的，以及事情為什麼是這樣的。

With each discovery, we asked more questions and pushed forward.

每一次發現，我們都會提出更多的問題，並不斷向前推進。

Preserving what we learned, outpacing what evolution could do with genes.

保存我們所學到的知識，超越基因進化所能做到的。

Knowledge builds on knowledge.

知識建立在知識之上。

Progress was slow at first, and then sped up exponentially.

起初進展緩慢，後來成倍加快。

Agriculture, writing, medicine, astronomy, or philosophy exploded into the world. 200 years ago, science took off and made us even better at learning about the world and speeding up progress. 35 years ago, the internet age began.

農業、文字、醫學、天文學或哲學在世界上大放異彩。200 年前，科學突飛猛進，讓我們更好地瞭解世界，加速進步。35 年前，互聯網時代開始。

Today, we live in a world made to suit our needs, created by us, for us.

今天，我們生活的世界是為了滿足我們的需求，由我們創造，為我們服務。

This is incredibly new.

這真是令人難以置信的新鮮事。

We forget how hard it was to get here, how enormous the steps on the intelligence ladder were, and how long it took to climb them.

我們忘記了來到這裡有多麼不容易，忘記了智力階梯上的臺階有多麼巨大，忘記了攀登這些臺階要花費多麼長的時間。

But once we did, we became the most powerful animal in the world in a heartbeat.

但一旦我們做到了，我們就會瞬間成為世界上最強大的動物。

But we may be in the process of changing this.

但我們可能正在改變這種狀況。

We're building machines that could be better at the very thing that gave us the power to conquer the planet.

我們正在製造的機器，可以在讓我們有能力征服地球的事情上做得更好。

Humanity's final invention.

人類最後的發明

Artificial intelligence.

人工智能。

Artificial intelligence, or AI, is software that performs mental tasks with a computer.

人工智能（Artificial Intelligence），簡稱 AI，是一種利用計算機執行智力任務的軟件。

Code that uses silicon instead of neurons to solve problems.

使用硅而不是神經元來解決問題的代碼。

In the beginning, AI was very simple.

最初，人工智能非常簡單。

Lines of code on paper, mere proofs of concept to demonstrate how machines could perform mental tasks.

紙上的一行行代碼，僅僅是演示機器如何執行智力任務的概念證明。

Only in the 1960s did we start seeing the first examples of what we would recognize as AI.

直到 20 世紀 60 年代，我們才開始看到人工智能的雛形。

A chatbot in 1964, a program to sort through molecules in 1965.

1964 年的哈拉機器人，1965 年的分子分類程序。

Slow, specialized systems requiring experts to use them.

速度慢，需要專家才能使用的專業系統。

Their intelligence was extremely narrow, built for a single task inside a controlled environment.

它們的智能極其狹窄，只能在受控環境中完成單一任務。

The equivalent of flatworms 500 million years ago doing the minimum amount of mental work.

相當於 5 億年前的扁蟲在做最少量的腦力勞動。

Progress in AI research paused several times when researchers lost hope in the technology.

當研究人員對人工智能技術失去希望時，人工智能研究的進展幾度停滯。

But just like changing environments create new niches for life, the world around AI changed.

但是，就像不斷變化的環境為生命創造了新的生存空間一樣，人工智能周圍的世界也發生了變化。

Between 1950 and 2000, computers got a billion times faster while programming became easier and widespread.

從 1950 年到 2000 年，計算機的運算速度提高了十億倍，而編程變得更加容易和普及。

In 1972, AI could navigate a room.

1972 年，人工智能可以在房間裡導航。

In 1989, it could read handwritten numbers.

1989 年，它可以讀取手寫數字。

But it remained a fancy tool, no match for humans.

但它仍然是一種花哨的工具，無法與人類相比。

Until in 1997, an AI shocked the world by beating the world champion in chess.

直到 1997 年，人工智能擊敗了國際象棋世界冠軍，震驚了世界。

Proving that we could build machines that could surpass us.

證明我們可以製造出超越我們的機器。

But we calmed ourselves because a chess bot is quite stupid.

但我們還是冷靜了下來，因為國際象棋機器人是非常愚蠢的。

Not a flatworm, but maybe a bee.

不是扁蟲，可能是蜜蜂。

Only able to perform a specialized narrow task.

只能完成專業性很強的任務。

But within this narrow task, it's so good that no human will ever again beat AI at chess.

但就在這一狹小的任務範圍內，它的表現是如此出色，以至於人類再也無法在國際象棋上戰勝人工智能。

As computers continued to improve, AI became a powerful tool for more and more tasks.

隨著計算機的不斷改進，人工智能成為越來越多任務的有力工具。

In 2004, it drove a robot on Mars.

2004 年，它在火星上駕駛了一個機器人。

In 2011, it began recommending YouTube videos to you.

2011 年，它開始向你推薦 YouTube 視頻。

But this was only possible because humans broke down problems into easy-to-digest chunks that computers could solve quickly.

但之所以能做到這一點，是因為人類將問題分解成易於消化的小塊，計算機可以快速解決。

Until we told AIs to teach themselves.

直到我們讓人工智能自學成才。

Rise of the self-learning machines.

自學機器的崛起

This is not a technical video, so we're massively oversimplifying here.

這不是一個技術視頻，所以我們在這裡過於簡單化了。

In a nutshell, the sheer power of supercomputers was combined with the almost endless data collected in the information age to make a new generation of AI.

一言以蔽之，超級計算機的強大功能與信息時代收集到的幾乎無窮無盡的數據相結合，造就了新一代人工智能。

AI experts began drastically improving forms of AI software called neural networks.

人工智能專家開始大幅改進名為神經網絡的人工智能軟件。

Enormously huge networks of artificial neurons that start out being bad at their tasks.

巨大的人工神經元網絡，一開始並不擅長完成任務。

They then used machine learning, which is an umbrella term for many different training techniques and environments that allows algorithms to write their own code and improve themselves.

然後，他們使用了機器學習，機器學習是許多不同訓練技術和環境的總稱，允許算法編寫自己的代碼並自我完善。

The scary thing is that we don't exactly know how they do it and what happens inside them.

可怕的是，我們並不清楚它們是如何做到這一點的，也不知道它們體內發生了什麼。

Just that it works and that what comes out the other end is a new type of AI.

只是它能工作，而且從另一端出來的是一種新型人工智能。

A capable black box of code.

一個能幹的代碼黑盒。

These new AIs could master complex skills extremely quickly with much less human help.

這些新的人工智能可以極快地掌握複雜的技能，而只需要更少的人類幫助。

They were still narrow intelligences, but a huge step up.

他們仍然是狹隘的智能，但已經向前邁進了一大步。

In 2014, Facebook AI could identify faces with 97% accuracy.

2014 年，Facebook 人工智能識別人臉的準確率高達 97%。

In 2016, an AI beat the best humans in the incredibly complex game of Go.

2016 年，人工智能在極其複雜的圍棋比賽中擊敗了最優秀的人類。

In 2018, a self-learning AI learned chess in four hours just by playing against itself and then defeated the best specialized chess bot.

2018 年，一個自學型人工智能僅通過與自己對弈，就在四個小時內學會了國際象棋，然後擊敗了最好的專業國際象棋機器人。

Since then, machine learning has been applied to reading, image processing, solving tests, and much more.

此後，機器學習被應用於閱讀、圖像處理、解題測試等領域。

Many of these AIs are already better than humans for whatever narrow task they were trained, but they still remained a simple tool.

這些人工智能中的許多已經比人類更好地完成了它們所訓練的狹窄任務，但它們仍然只是一種簡單的工具。

AI still didn't seem that big of a deal for most people.

對大多數人來說，人工智能似乎仍然不是什麼大事。

And then came the chatbot ChatGPT.

隨後，哈拉機器人 ChatGPT 誕生了。

The work that went into it is massive.

其工作量之大可想而知。

It trained on nearly everything written on the internet to learn how to handle language, which it now does better than most people.

它接受了互聯網上幾乎所有文字的訓練，學習如何處理語言，現在它比大多數人做得都好。

It can summarize, translate, and help with some maths problems.

它可以總結、翻譯和幫助解決一些數學問題。

It's incredibly more broad than any other system just a few years ago.

它比幾年前的任何其他系統都要廣泛得令人難以置信。

Not crushing any single benchmark, but all of them at once.

不是粉碎任何一個基準，而是同時粉碎所有基準。

Many large tech companies are spending billions to build powerful competitors.

許多大型科技公司正在斥資數十億美元打造強大的競爭對手。

AI is already transforming customer service, banking, healthcare, marketing, copywriting, creative spaces, and more.

人工智能已經在改變客戶服務、銀行業務、醫療保健、市場營銷、文案寫作、創意空間等領域。

AI-generated content has already taken hold of social media, YouTube, and news websites.

人工智能生成的內容已經在社交媒體、YouTube 和新聞網站上佔據一席之地。

Elections are expected to be inundated by propaganda and misinformation.

預計選舉將充斥著各種宣傳和錯誤信息。

No one is sure how much good or harm can come from adopting AI everywhere.

沒有人知道，到處採用人工智能會帶來多少好處或壞處。

Change is scary.

變化是可怕的。

There will be winners and losers.

有贏家，也有輸家。

One of the biggest questions governments and corporations have now is how to manage the transition to an AI-boosted economy.

政府和企業目前面臨的最大問題之一是如何管理向人工智能推動的經濟過渡。

All these potential gains or risks are just the result of today's AI.

所有這些潛在的收益或風險都是當今人工智能的成果。

ChatGPT's intelligence is a major step up, but it remains narrow.

ChatGPT 的智能是一大進步，但它的範圍仍然很窄。

While it can write a great essay in seconds, it doesn't understand what it's writing.

雖然它能在幾秒鐘內寫出一篇好文章，但它並不瞭解自己在寫什麼。

But what if the AIs stopped being narrow?

但如果人工智能不再狹隘呢？

General AI What makes humans different from current AI is our general intelligence.

通用人工智能 人類與當前人工智能的不同之處在於我們的通用智能。

Humans can technically absorb any piece of knowledge and start working on any problem.

人類可以從技術上吸收任何知識，並開始解決任何問題。

We're great at many very different skills and tasks, from playing chess to writing or solving science puzzles.

我們擅長許多不同的技能和任務，從下棋到寫作或解決科學難題。

Not equally, of course.

當然，並不是一視同仁。

Some of us are experts in some fields and beginners in others, but we can technically do all of them.

我們中的一些人在某些領域是專家，而在另一些領域則是初學者，但從技術上講，我們可以勝任所有領域的工作。

In the past, AI was narrow and able to become good at one skill, but was rather bad in all the others.

過去，人工智能的範圍很窄，能夠精通一種技能，但其他技能卻相當糟糕。

Simply by building faster computers and pouring more money into AI training, we'll get us new, more powerful generations of AI.

只需建造更快的計算機，併為人工智能培訓投入更多資金，我們就能獲得更強大的新一代人工智能。

But what if the next step for AI is to become a general intelligence like us, an AGI?

但是，如果人工智能的下一步是成為像我們一樣的通用智能，即 AGI 呢？

If the AI improvement process continues as it has been, it's not unlikely that AGI could be better in most or even all skills that humans can do.

如果人工智能的改進進程一如既往，AGI 在大多數甚至所有人類可以做到的技能方面都能做得更好，這並非不可能。

We don't know how to build AGI, how it will work, or what it will be able to do.

我們不知道如何建造 AGI，不知道它將如何工作，也不知道它能做什麼。

Since narrow AIs today are capable of mastering one mental task quickly, AGI might be able to do the same with all mental tasks.

既然現在的狹義人工智能能夠快速掌握一項智力任務，那麼 AGI 也許也能完成所有智力任務。

So, even if it starts out stupid, an AGI might be able to become as smart and capable as a human.

是以，即使一開始很愚蠢，人工智能也有可能變得和人類一樣聰明能幹。

While this sounds like science fiction, most AI researchers think this will happen sometime this century, maybe already in a few years.

雖然這聽起來像科幻小說，但大多數人工智能研究人員認為，這將在本世紀某個時候實現，也許幾年後就會實現。

Humanity is not ready for what will happen next, not socially, not economically, not morally.

人類還沒有準備好應對接下來發生的一切，社會上沒有，經濟上沒有，道德上也沒有。

Earlier, we defined intelligence as the ability to learn, reason, acquire knowledge and skills, and use them to solve problems.

前面，我們把智力定義為學習、推理、獲取知識和技能並利用它們解決問題的能力。

All things humans excel at.

人類擅長的所有事情。

An AGI as intelligent as even an average human would already disrupt modern civilization because they're not bound by the same limitations as we are.

即使與普通人一樣智能的 AGI 也會破壞現代文明，因為它們不受我們的限制。

Today's AIs like ChatGBT already think and solve the tasks they were made for at least 10 times faster than even very skilled humans.

如今，像 ChatGBT 這樣的人工智能在思考和解決任務時的速度已經比非常熟練的人類至少快 10 倍。

Maybe AGI will be slower, but it may also be faster, maybe much faster.

也許 AGI 會更慢，但也可能更快，也許快得多。

And since AGIs are software, you could copy them endlessly as long as you have enough storage and run them in parallel.

由於 AGI 是軟件，只要有足夠的存儲空間，就可以無休止地複製它們，並並行運行。

There are 8 million scientists in the world.

世界上有 800 萬科學家。

Now imagine an AGI copied a million times and put to work.

現在想象一下，一個人工智能被複制了一百萬次，然後投入工作。

Imagine 1 million scientists working 24-7, thinking 10 times faster than humans, without being distracted, only focused on the task they've been given.

想象一下，100 萬名科學家全天候工作，他們的思維速度是人類的 10 倍，而且心無旁騖，只專注於他們被賦予的任務。

What if, suddenly, AGI could do all intelligence-based jobs in the world, from interpreting law to coding to creating animated YouTube videos, better, faster and much cheaper than humans?

如果突然之間，AGI 能夠比人類更好、更快、更便宜地完成世界上所有基於智能的工作，從解釋法律到編碼，再到製作 YouTube 動畫視頻，會怎樣呢？

Would whoever controls this AGI suddenly own the economy?

控制 AGI 的人會突然擁有經濟嗎？

And thinking bigger, human progress is our intelligence applied to problems.

往大了想，人類的進步就是我們將智慧應用於解決問題。

So what could a million AGIs achieve?

那麼，一百萬個 AGI 能實現什麼呢？

Solve fundamental questions of science, like dark energy.

解決科學的基本問題，如暗能量。

Invent new technology that gives us limitless energy, fix climate change, cure aging and cancer.

發明新技術，為我們提供無限能源，解決氣候變化問題，治癒衰老和癌症。

But then again, sadly, humans apply their intelligence not just for the benefit of all.

但可悲的是，人類運用自己的智慧並不只是為了造福所有人。

What if the AGIs are tasked to guide drones or pull the triggers in war?

如果 AGI 的任務是引導無人機或在戰爭中扣動扳機呢？

Or to engineer a virus that only kills people with green eyes?

或者設計一種只殺死綠眼睛人的病毒？

Or to create the most profitable social media, so addictive that people starve in front of their screens?

還是創建最賺錢的社交媒體，讓人們沉迷其中，在螢幕前忍飢挨餓？

The creation of AGI could reasonably be as big of an event as taming fire or electricity, and give whoever invents it equally as much power.

按理說，創造 AGI 就像馴服火或電一樣重要，無論誰發明了它，都會擁有同樣強大的力量。

But now let's go one step further.

但現在，讓我們更進一步。

What if the potential of AGI doesn't stop here?

如果 AGI 的潛力不止於此，那該怎麼辦？

Intelligence explosion.

情報爆炸

Intelligence and knowledge build and accelerate each other, but humans are limited by biology and evolution.

智力和知識相互促進，但人類受到生物和進化的限制。

Once we evolved the right hardware, our software outpaced evolution by orders of magnitude, and within a heartbeat, we ruled this planet.

一旦我們進化出了合適的硬件，我們的軟件就會在數量級上超越進化，並在瞬間統治這個星球。

But our software basically hasn't changed much since then, which is why we have obesity and destroy the climate for short-term gains.

但從那時起，我們的軟件基本上就沒怎麼變過，所以我們才會為了短期利益而肥胖，破壞氣候。

Since AGI is software on a computer, once it's smart enough to do AI research, the rate of AI progress should speed up a lot.

由於 AGI 是計算機上的軟件，一旦它足夠聰明，可以進行人工智能研究，人工智能的進步速度就會大大加快。

And that results in better AI that's better at AI research without much human involvement.

這樣就能產生更好的人工智能，在沒有太多人類參與的情況下更好地進行人工智能研究。

It may even be possible that AI could learn how to directly improve itself, in which case some experts fear this feedback loop could be incredibly fast.

人工智能甚至有可能學會如何直接改進自己，在這種情況下，一些專家擔心這種反饋循環的速度會快得驚人。

Maybe just months or years after the first self-improving AGI is switched on.

也許就在第一個自我完善的人工智能啟動後的幾個月或幾年。

Maybe it would actually take decades.

也許真的需要幾十年。

We simply don't know.

我們根本不知道。

This is all speculative.

這些都是推測。

But such an intelligence explosion might lead to a true super-intelligent entity.

但是，這種智能爆炸可能會導致真正的超級智能實體。

We don't know what such a being would look like, what its motives or goals would be, what would go on in its inner world.

我們不知道這樣的生物會是什麼樣子，它的動機或目標是什麼，它的內心世界會發生什麼。

We could be as laughably stupid to a super-intelligence as squirrels are to us, unable to even comprehend its way of thinking.

對超級智能體來說，我們可能就像松鼠對我們一樣愚蠢可笑，甚至無法理解它的思維方式。

This hypothetical scenario keeps many people up at night.

這種假設情景讓許多人徹夜難眠。

Humanity is the only example we have of an animal becoming smarter than all others.

人類是我們所掌握的動物變得比其他動物更聰明的唯一例子。

And we have not been kind to what we perceive as less intelligent beings.

我們對那些我們認為智力低下的生物並不友好。

AGI might be the last invention of humanity.

AGI 可能是人類最後的發明。

It's possible that it could become the most intelligent and therefore most powerful being on Earth.

它有可能成為地球上最聰明、是以也是最強大的生物。

A god in a box that could exercise its power to bring unimaginable wealth and happiness to humans while securing our future.

一個裝在盒子裡的神，可以行使它的力量，為人類帶來難以想象的財富和幸福，同時確保我們的未來。

Or it could subvert civilization and bring about our end, with humanity unable to come up with a way to stop it.

或者，它可能顛覆文明，給我們帶來滅頂之災，而人類卻無法想出阻止它的辦法。

We'll look at some of these potential futures in more videos, but for now, let's wrap up.

我們將在更多視頻中瞭解這些潛在的未來，但現在，讓我們來總結一下。

The only thing we know for sure is that today, right now, many of the largest and richest companies in the world are racing to create ever more powerful AIs.

我們唯一可以確定的是，今天，就在此時此刻，世界上許多最大、最富有的公司都在競相創造更加強大的人工智能。

Whatever our future is, we are running towards it.

無論我們的未來是什麼，我們都在奔向它。

Who knows how long we have until we must confront our AI future.

誰知道我們還有多久必須面對人工智能的未來？

Luckily, you still have plenty of time to prepare for it, if you're learning on Brilliant, that is.

幸運的是，如果您是在 Brilliant 上學習，您還有充足的時間為此做準備。

Brilliant will make you a better thinker and problem solver in just minutes a day, with thousands of bite-sized hands-on lessons on just about anything you may be curious about, including AI.

Brilliant 每天只需幾分鐘，就能讓你成為更好的思考者和問題解決者，它提供了數千個小巧實用的課程，內容涉及你可能好奇的任何事情，包括人工智能。

Their latest course, How LLMs Work, takes you under the hood of real language models.

他們的最新課程 "LLM 如何工作 "將帶您瞭解真實語言模型的原理。

It demystifies technologies like ChatGPT with interactive lessons on everything from how models build vocabulary to how they choose their next word.

它通過互動課程揭開了 ChatGPT 等技術的神祕面紗，從模型如何積累詞彙到如何選擇下一個單詞，無所不包。

You'll learn how to tune LLMs to produce output with exactly your desired tonality, whether it's poetry or a cover letter.

無論是詩歌還是求職信，您都將學會如何調整 LLM，使其輸出的內容完全符合您的要求。

And you'll understand why training is really everything by comparing models trained on Taylor Swift lyrics and the legal speech of big tech's terms and conditions.