They were among the first to bring computerized manufacturing to my hometown in India.

他們是第一批將計算機化生產引入我的家鄉印度的企業之一。

Growing up as a young girl, I remember being fascinated how these computer programs didn't just reside within a computer, but touched the physical world and produced these beautiful and precise metal parts.

在我還是個小女孩的時候，我就對這些計算機程序感到著迷，因為它們不僅存在於計算機中，而且還接觸到物理世界，並生產出這些美麗而精密的金屬零件。

Over the last two decades,

在過去二十年裡

I've learned a great deal about the power of technology.

我學到了很多關於技術力量的知識。

Over the last two decades, as I pursued AI research, this memory continued to inspire me to connect the physical and digital worlds together.

在過去的二十年裡，當我從事人工智能研究時，這段記憶不斷激勵著我將物理世界和數字世界連接在一起。

I am working on AI that transforms the way we do science and engineering.

我正在研究人工智能，它將改變我們從事科學和工程工作的方式。

Scientific research and engineering design currently involves a lot of trial and error.

目前，科學研究和工程設計涉及大量的試驗和錯誤。

Many long hours are spent in the lab doing experiments.

在實驗室做實驗的時間很長。

So it's not just the great ideas that propel science forward.

是以，推動科學進步的不僅僅是偉大的想法。

You need these experiments to validate findings and spark new ideas.

您需要這些實驗來驗證研究結果並激發新的想法。

How can language models help here?

語言模型在這方面有何幫助？

What if I ask Chad Chippity to come up with a better design of an aircraft wing or a drone that flies under turbulent winds?

如果我讓查德-奇皮蒂設計出更好的飛機機翼或在亂風中飛行的無人機呢？

It may suggest something.

它可能暗示著什麼。

It may even draw something.

甚至還能畫出一些東西。

But how do we know this is any good?

但是，我們怎麼知道這是好東西呢？

We don't.

我們沒有。

Language models hallucinate because they have no physical grounding.

語言模型產生幻覺是因為它們沒有物理基礎。

While language models may help generate new ideas, they cannot attack the hard part of science, which is simulating the necessary physics to replace the lab experiments.

雖然語言模型可能有助於產生新的想法，但它們無法解決科學的難點，即模擬必要的物理學來取代實驗室實驗。

In order to model scientific and physical phenomena, text alone is not sufficient.

要模擬科學和物理現象，僅靠文字是不夠的。

To get to AI with universal physical understanding, we need to train it on the data of the world we observe.

要想讓人工智能具備普遍的物理理解能力，我們需要用我們觀察到的世界數據來訓練它。

And not just that, also its hidden details.

不僅如此，還有一些不為人知的細節。

From the intricacies of quantum chemistry that happen at the smallest level, to molecules and proteins that influence how all biological processes work, to ocean currents and clouds that happen at planetary scales and beyond, we need AI that can capture this whole range of physical phenomena.

從發生在最小層次的錯綜複雜的量子化學，到影響所有生物過程運作的分子和蛋白質，再到發生在行星尺度和更大尺度的洋流和雲層，我們需要能夠捕捉這一系列物理現象的人工智能。

We need AI that can really zoom into the fine details in order to simulate these phenomena accurately.

我們需要能夠真正放大細節的人工智能，以準確模擬這些現象。

To capture the cloud movements and predict how clouds move and change in our atmosphere, we need to be able to zoom into the fine details of the turbulent fluid flow.

要捕捉雲的運動並預測雲如何在大氣中移動和變化，我們需要能夠放大湍流流體流動的細節。

Standard deep learning uses a fixed number of pixels.

標準深度學習使用固定數量的像素。

So if you zoom in, it gets blurry, and not all the details are captured.

是以，如果放大，畫面就會變得模糊，無法捕捉到所有細節。

We invented an AI technology called neural operators that represents the data as continuous functions or shapes and allows us to zoom in indefinitely to any resolution or scale.

我們發明了一種名為神經算子的人工智能技術，它將數據表示為連續的函數或形狀，並允許我們無限放大到任何分辨率或比例。

Neural operators allow us to train on data at multiple scales or resolutions.

神經運算符允許我們在多種規模或分辨率的數據上進行訓練。

And also allows us to incorporate the knowledge of mathematical equations to fill in the finer details when only limited-resolution data is available.

在只能獲得有限分辨率數據的情況下，我們還可以利用數學方程的知識來填補更精細的細節。

Such learning at multiple scales is essential for scientific understanding.

這種多尺度的學習對科學理解至關重要。

And neural operators enable this.

而神經運算器可以實現這一點。

With neural operators, we can simulate physical phenomena such as fluid dynamics as much as a million times faster than traditional simulations.

有了神經運算器，我們模擬流體動力學等物理現象的速度可以比傳統模擬快一百萬倍。

Last year, we used neural operators to invent a better medical catheter.

去年，我們利用神經運算器發明了一種更好的醫用導管。

A medical catheter is a tube that draws fluids out of the human body.

醫用導管是一種將液體引出人體的管子。

Unfortunately, the bacteria tend to swim upstream against the fluid flow and infect the human.

不幸的是，細菌往往會逆流而上，感染人體。

In fact, annually, there's more than half a million cases of such health-care-related infections.

事實上，每年此類與醫療保健相關的感染病例超過 50 萬例。

And this is one of the leading causes.

這是主要原因之一。

Last year, we used neural operators to change the inside of the catheter from smooth to ridged.

去年，我們使用神經運算器將導管內部從光滑變為脊狀。

With ridges, now we have vortices created as the fluid flows.

有了脊，流體流動時就會產生渦流。

And we can hope to stop the bacteria from swimming upstream because of these vortices.

我們可以希望阻止細菌因為這些漩渦而逆流而上。

But to get this correct, we need the shape of the ridges to be exactly right.

但要做到這一點，我們需要脊線的形狀完全正確。

In the past, this would have been done by trial and error.

在過去，這需要反覆試驗。

Design a version of the catheter, build it out, take it to the lab, observe a hypothesis if something went wrong, rinse and repeat and redesign again.

設計一個版本的導管，製造出來，拿到實驗室，觀察假設是否出了問題，然後沖洗，重複，再重新設計。

But instead, we taught AI the behavior of the fluid flow inside the tube.

不過，我們教人工智能的是管內流體流動的行為。

And with it, our neural operator model was able to directly propose an optimized design.

有了它，我們的神經運算模型就能直接提出優化設計方案。

We 3D-printed the design only once to verify that it worked.

我們只對設計進行了一次 3D 打印，以驗證其是否可行。

In the video, you're seeing our catheter being tested in the lab.

在視頻中，您將看到我們的導管正在實驗室中進行測試。

The bacteria are not able to swim upstream, are instead being pushed out with the fluid flow.

細菌無法逆流而上，而是被液體推了出去。

In fact, we measured the reduction in bacterial contamination by more than 100-fold.

事實上，我們測得細菌汙染減少了 100 多倍。

So in this case, the neural operators were specialized to understand fluid flow in a tube.

是以，在這種情況下，神經運算器專門用於理解管中的流體流動。

What other applications can AI tackle and help us solve such pressing problems?

人工智能還能處理和幫助我們解決哪些緊迫問題？

Can deep learning beat numerical weather models?

深度學習能否打敗數值天氣模型？

A group of leading weather scientists asked this question in February 2021 in a Royal Society publication.

2021 年 2 月，一組頂尖氣象科學家在英國皇家學會的一份出版品中提出了這個問題。

They felt that AI was still in its infancy and that a number of fundamental breakthroughs would be needed for AI to become competitive with traditional weather models.

他們認為，人工智能仍處於起步階段，要想與傳統氣象模型競爭，人工智能還需要一些根本性的突破。

And that would take years or even decades.

而這需要幾年甚至幾十年的時間。

Exactly a year later, we released ForecastNet.

整整一年後，我們發佈了 ForecastNet。

Using neural operators, we built the first fully AI-based weather model that is high-resolution and is tens of thousands of times faster than traditional weather models.

利用神經運算器，我們建立了首個完全基於人工智能的天氣模型，該模型具有高分辨率，速度比傳統天氣模型快數萬倍。

What used to take a big supercomputer can now run on a gaming PC that you may have at home.

過去需要大型超級計算機才能完成的工作，現在只需一臺家用遊戲電腦就能完成。

This model is also running at the European Center for Medium-Range Weather Forecasting, one of the premier weather agencies of the world.

該模型也在世界主要氣象機構之一的歐洲中期天氣預報中心（European Center for Medium-Range Weather Forecasting）運行。

And our AI model is not just tens of thousands of times faster than traditional models.

我們的人工智能模型不僅比傳統模型快數萬倍，而且還比傳統模型更快。

It's also more accurate in many cases.

在很多情況下，它也更準確。

On September 16 last year,

去年 9 月 16 日

Hurricane Lee hit the coast of Nova Scotia, Canada.

李 "颶風襲擊了加拿大新斯科舍省海岸。

A full 10 days earlier, our ForecastNet model correctly predicted that the hurricane would make landfall.

整整 10 天前，我們的預測網模型正確預測到颶風將登陸。

But the traditional weather model predicted the hurricane would skip the coast.

但傳統的氣象模型預測颶風將跳過海岸。

Only five days later, on September 11, did the traditional weather model correct its forecast to predict landfall.

直到五天後的 9 月 11 日，傳統氣象模型才修正了預測，預測到了登陸時間。

Extreme weather events such as Hurricane Lee will only increase further unless we take action on climate change.

除非我們對氣候變化採取行動，否則像李颶風這樣的極端天氣事件只會進一步增加。

Such as finding new, clean sources of energy.

比如尋找新的清潔能源。

Nuclear fusion is one of them.

核聚變就是其中之一。

But unfortunately, there are still big challenges with it.

但遺憾的是，它仍然面臨著巨大的挑戰。

The fusion reactor heats up the plasma to extremely high temperatures to get fusion started.

聚變反應堆將等離子體加熱到極高的溫度，以啟動聚變。

And sometimes, this hot plasma can escape confinement and can damage the reactor.

有時，這些熾熱的等離子體會逃脫束縛，從而損壞反應堆。

We train neural operators to simulate and predict the evolution of plasma inside the reactor.

我們訓練神經運算器來模擬和預測反應堆內等離子體的演變。

And with it, we can use this to predict disruptions before they occur and take corrective action in the real world.

有了它，我們就能在干擾發生之前利用它進行預測，並在現實世界中採取糾正措施。

We are enabling the possibility of nuclear fusion becoming a reality.

我們正在使核聚變成為現實。

So neural operators and AI broadly are enabling us to tackle hard scientific challenges, such as climate change and nuclear fusion.

是以，神經運算器和人工智能正在幫助我們應對氣候變化和核聚變等艱鉅的科學挑戰。

To me, this is just the beginning.

對我來說，這僅僅是個開始。

So far, these AI models are limited to the narrow domains they're trained on.

到目前為止，這些人工智能模型還侷限於它們所訓練的狹窄領域。

What if you had an AI model that could solve all and any scientific problem, from designing better drones, aircrafts, rockets, and even better drugs and medical devices?

如果你有一個人工智能模型，可以解決所有科學問題，包括設計更好的無人機、飛機、火箭，甚至更好的藥物和醫療設備，你會怎麼想？

Such an AI model would greatly benefit humanity.

這種人工智能模式將極大地造福於人類。

This is what we are working on.

這就是我們的工作。

We are building a generalist AI model with emergent capabilities that can simulate any physical phenomena and generate novel designs that were previously out of reach.

我們正在建立一個具有突發性能力的通用人工智能模型，它可以模擬任何物理現象，併產生以前無法實現的新穎設計。

This is how we scale up neural operators to enable general intelligence with universal physical understanding.

這就是我們如何擴展神經運算器，以實現具有普遍物理理解能力的通用智能。

Thank you.

謝謝。

Thank you.

謝謝。

Thank you.

謝謝。

Thank you.

謝謝。