整理 | Max

7月9日，2020世界人工智能大會(huì)開幕式在上海正式舉行，特斯拉首席執(zhí)行官埃隆·馬斯克在線接受了組委會(huì)的采訪。

馬斯克表示，對(duì)未來(lái)實(shí)現(xiàn)L5級(jí)別自動(dòng)駕駛或是完全自動(dòng)駕駛非常有信心，而且很快就會(huì)實(shí)現(xiàn)。特斯拉已經(jīng)非常接近L5級(jí)自動(dòng)駕駛，有信心將在今年完成L5級(jí)別的基本功能。

他認(rèn)為，實(shí)現(xiàn)自動(dòng)駕駛L5級(jí)目前不存在底層根本性的挑戰(zhàn)，但是有很多的細(xì)節(jié)問(wèn)題。而特斯拉面臨的挑戰(zhàn)就是要解決所有這些問(wèn)題，然后整合系統(tǒng)，持續(xù)解決長(zhǎng)尾問(wèn)題。

“你會(huì)發(fā)現(xiàn)可以處理大多數(shù)場(chǎng)景的情況，但是又會(huì)不時(shí)地出現(xiàn)一些奇怪的情況，你必須有一個(gè)系統(tǒng)來(lái)解決、來(lái)訓(xùn)練，解決這些奇怪的場(chǎng)景。你需要現(xiàn)實(shí)的場(chǎng)景，沒(méi)有什么比現(xiàn)實(shí)更復(fù)雜的。任何的模擬，都是現(xiàn)實(shí)世界復(fù)雜性的子集。"馬斯克說(shuō)。

他表示，特斯拉會(huì)在中國(guó)做很多原創(chuàng)性的工程開發(fā)，不僅僅是簡(jiǎn)單把美國(guó)的東西搬到中國(guó)。

馬斯克透露，特斯拉上海超級(jí)工廠的進(jìn)展很順利，預(yù)計(jì)未來(lái)在中國(guó)工廠中會(huì)運(yùn)用更多的人工智能和更智能化的軟件。但他認(rèn)為，在工廠真正地有效使用人工智能還需要花一些時(shí)間。

以下為問(wèn)答實(shí)錄：

Q：First, in terms of Tesla products, we know that Autopilot is one of its most popular features. How does it work in China?

問(wèn)：首先，我們都知道，Autopilot 自動(dòng)輔助駕駛是特斯拉純電動(dòng)車非常受歡迎的一項(xiàng)功能。它在中國(guó)市場(chǎng)的應(yīng)用情況如何？

A：Tesla Autopilot does work reasonably well in China. It does not work quite as well in China as it does in the US because still most of our engineering is in the US so that tends to be the local group of optimization. So Autopilot tends to work the best in California because that is where the engineers are. And then once it works in California, we then extend it to the rest of the world. But we are building up our engineering team in China. And so if you're interested in working at Tesla China as an engineer, we would love to have you work there. That will be great.

答：特斯拉自動(dòng)輔助駕駛在中國(guó)市場(chǎng)應(yīng)用還不錯(cuò)。但因?yàn)槲覀兣c自動(dòng)駕駛相關(guān)的工程開發(fā)集中在美國(guó)，所以自動(dòng)輔助駕駛功能在美國(guó)應(yīng)用得更好，在加州最好，這也主要是因?yàn)槲覀兿嚓P(guān)的工程師在加州。在我們確定這項(xiàng)功能在加利福尼亞運(yùn)作良好后，我們會(huì)將其推送到世界其他地區(qū)。目前，我們正在中國(guó)建立相關(guān)的工程團(tuán)隊(duì)。如果你想成為特斯拉中國(guó)的工程師，我們會(huì)非常歡迎，這將會(huì)非常好。

I really want to emphasize it is a lot that we are going to be doing original engineering in China. It's not just converting sort of stuff from America to work in China, we will be doing original design and engineering in China. So please do consider Tesla China, if you're thinking about working somewhere.

我想強(qiáng)調(diào)下，在中國(guó)我們要做的是進(jìn)行很多原創(chuàng)性的工程開發(fā)。所以并不是簡(jiǎn)單的將美國(guó)的東西直接照搬到中國(guó)，而是就在中國(guó)進(jìn)行原創(chuàng)的設(shè)計(jì)和原創(chuàng)的工程開發(fā)。所以，如果您考慮工作，請(qǐng)考慮在特斯拉中國(guó)工作。

Q：Great. How confident are you that level five autonomy will eventually be with us? And when do you think we will reach full level five autonomy?

問(wèn)：您對(duì)于我們最終實(shí)現(xiàn)L5級(jí)別自動(dòng)駕駛有多大信心？您覺(jué)得這一天什么時(shí)候會(huì)到來(lái)？

A：I'm extremely confident that level five or essentially complete autonomy will happen, and I think will happen very quickly.

我對(duì)未來(lái)實(shí)現(xiàn)L5級(jí)別自動(dòng)駕駛或是完全自動(dòng)駕駛非常有信心，而且我認(rèn)為很快就會(huì)實(shí)現(xiàn)。

I think at Tesla, I feel like we are very close to level five autonomy.I think I remain confident that we will have the basic functionality for level five autonomy complete this year.

在特斯拉，我覺(jué)得我們已經(jīng)非常接近L5級(jí)自動(dòng)駕駛了。我有信心，我們將在今年完成開發(fā)L5級(jí)別的基本功能。

So yes, I think there are no fundamental challenges remaining for level five autonomy. There are many small problems. And then there's the challenge of solving all those small problems and then putting the whole system together, and just keep addressing the long tail of problems. So you'll find that you're able to handle the vast majority of situations. But then there will be something very odd. And then you have to have the system figure out a train to deal with these very odd situations. This is why you need a kind of a real world situation. Nothing is more complex and weird than the real world. Any simulation we create is necessarily a subset of the complexity of the real world.

我認(rèn)為實(shí)現(xiàn)自動(dòng)駕駛L5目前不存在底層的根本性的挑戰(zhàn)，但是有很多細(xì)節(jié)問(wèn)題。我們面臨的挑戰(zhàn)就是要解決所有這些小問(wèn)題，然后整合系統(tǒng)，持續(xù)解決這些長(zhǎng)尾問(wèn)題。你會(huì)發(fā)現(xiàn)你可以處理絕大多數(shù)場(chǎng)景的問(wèn)題，但是又會(huì)不時(shí)出現(xiàn)一些奇怪不尋常的場(chǎng)景，所以你必須有一個(gè)系統(tǒng)來(lái)找出并解決這些奇怪不尋常場(chǎng)景的問(wèn)題。這就是為什么你需要現(xiàn)實(shí)世界的場(chǎng)景。沒(méi)有什么比現(xiàn)實(shí)世界更復(fù)雜了。我們創(chuàng)建的任何模擬都是現(xiàn)實(shí)世界復(fù)雜性的子集。

So we are really deeply enmeshed in dealing with the tiny details of level five autonomy. But I'm absolutely confident that this can be accomplished with the hardware that is in Tesla today, and simply by making software improvements, we can achieve level five autonomy.

因此，我們目前非常專注于處理L5級(jí)別自動(dòng)駕駛的細(xì)節(jié)問(wèn)題上。并且，我相信這些問(wèn)題完全可基于特斯拉車輛目前搭載的硬件版本來(lái)解決，我們只需改進(jìn)軟件，就可以實(shí)現(xiàn)L5級(jí)別自動(dòng)駕駛。

Q：Great. If we look at the three building blocks of AI and robotics: perception, cognition, and action, how would you assess the progress respectively so far?

問(wèn)：您覺(jué)得人工智能和機(jī)器人技術(shù)的三大支柱：感知、認(rèn)知和行為，目前在各自領(lǐng)域的進(jìn)展如何？

A：I am not sure I totally agree with dividing it into those categories: perception, cognition, and action. But if you do use those categories, I’d say that probably perception we've made, if you can say like the recognition of objects, we've made incredible progress in recognition of objects. In fact, I think it would probably fair to say that advanced image recognition system today is better than almost any human, even in an expert field.

答：我不確定人工智能技術(shù)是否可以這樣分類。如果按照這個(gè)分類標(biāo)準(zhǔn)，在感知層面，以識(shí)別物體為例，目前的技術(shù)取得了巨大進(jìn)展?？梢哉f(shuō)，即便是在專業(yè)領(lǐng)域，當(dāng)今的高級(jí)圖像識(shí)別系統(tǒng)也比人類都要好。

So it is really a question of how much compute power, how many computers were required to train it? How many compute hours? What was the efficiency of the image training system? But in terms of image recognition or sound recognition, and really any signal you can say, generally speaking any byte stream, Can an AI system recognize things accurately with a given byte stream？Extremely well.

問(wèn)題的實(shí)質(zhì)在于需要多強(qiáng)的計(jì)算能力，多少計(jì)算機(jī)和多長(zhǎng)計(jì)算時(shí)間來(lái)訓(xùn)練感知能力？圖像識(shí)別訓(xùn)練系統(tǒng)的效率如何？就圖像識(shí)別或聲音識(shí)別而言，對(duì)于給定的字節(jié)流，人工智能系統(tǒng)能否準(zhǔn)確識(shí)別處理？答案是非常好。

Cognition. This is probably the weakest area. Do you understand concepts？Are you able to reason effectively? And can you be creative in a way that makes sense? You have so many advanced AIs that are very creative, but they do not curate their creative actions very well. We look at it and it is not quite right. It will become right though.

認(rèn)知可能是最薄弱的領(lǐng)域，人工智能是否可以理解概念？是否會(huì)有效推理？能否創(chuàng)造有意義的事物？目前有很多非常有創(chuàng)造力的技術(shù)先進(jìn)的人工智能，但是它們無(wú)法很好地控制其創(chuàng)造活動(dòng)。至少現(xiàn)在在我們看來(lái)不太對(duì)，不過(guò)未來(lái)它會(huì)看起來(lái)像樣些。

And then action, sort of like things like games, as maybe something part of the action part of thing. Obviously at this point, any game with rules, AI will be superhuman at any game with an understandable set of rules, essentially any game below a certain degree of freedom level. Let us say at this point, any game, it would be hard-pressed to think of a game where if there was enough attention paid to it, that we would not make it superhuman AI that could play it. That's not even taking into account the faster reaction time of AI.

然后是行為。這個(gè)可以以游戲打比方。在任何規(guī)則明確的游戲中，或者自由發(fā)揮空間比較有限的游戲，人工智能就像超人類一樣。就目前而言，很難想像有什么游戲，人工智能游戲玩家不能發(fā)揮超人類水平的，這甚至都不去考慮到人工智能更快的反應(yīng)時(shí)間。

Q：In what ways does Autopilot stimulate the development of AI algorithms and chips? And how do you does it refresh our understanding of AI technology?

問(wèn)：Autopilot自動(dòng)輔助駕駛在哪些方面推動(dòng)了AI算法和芯片的發(fā)展？它又如何改變了我們對(duì)AI技術(shù)的理解？

A：In developing AI chips for Autopilot, what we found was that there was no system on the market that was capable of doing inference within a reasonable cost or power budget. So if we had gone with a conventional GPUs, CPUs and that kind of thing, we would have needed several hundred watts and we would have needed to fill up the trunk with computers and GPUs and a big cooling system. It would have been costly and bulky and have taken up too much power, which is important for range for an electric car.

答：在為自動(dòng)輔助駕駛開發(fā)人工智能芯片時(shí)，我們發(fā)現(xiàn)市場(chǎng)上沒(méi)有成本合理且低功耗的系統(tǒng)。如果我們使用傳統(tǒng)的GPU、CPU或其他相似的產(chǎn)品，將耗費(fèi)數(shù)百瓦的功率，并且后備箱會(huì)被計(jì)算機(jī)，GPU巨大的冷卻系統(tǒng)占據(jù)，由此一來(lái)成本高昂，占用車輛體積，而且高耗能。要知道，能耗對(duì)于電動(dòng)汽車的行駛里程很關(guān)鍵。

So we developed our own AI chip, the Tesla Full Self-Driving computer with dual system on chips with the eight bit and accelerators for doing the dot products. I think probably a lot of people in this audience are aware of it. But AI consists of doing a great many dot products. This is like, if you know what a dot product is, it's just a lot of dot products, which effectively means that our brain must be doing a lot of dot products. We still actually haven't fully explored the power of the Tesla Full Self-Driving computer. In fact, we only turned on the second system on chip harshly a few months ago. So making full use of Tesla Full-Self Driving computer will probably take us at least another year or so.

為此我們開發(fā)了特斯拉自有的人工智能芯片，即具有雙系統(tǒng)的特斯拉完全自動(dòng)駕駛電腦，該芯片具有8位元和加速器，用于點(diǎn)積運(yùn)算。在座各位可能有很多人都有所了解，人工智能包含很多點(diǎn)積運(yùn)算，如果你知道什么是點(diǎn)積運(yùn)算，那么便知道點(diǎn)積運(yùn)算量巨大，這意味著我們的電腦必須做很多點(diǎn)積運(yùn)算。我們事實(shí)上還未完全發(fā)揮出特斯拉完全自動(dòng)駕駛電腦的能力。實(shí)際上，幾個(gè)月前我們才審慎地啟動(dòng)了芯片的第二套系統(tǒng)。充分利用特斯拉完全自動(dòng)駕駛電腦的能力，可能還需要至少一年左右的時(shí)間。

Then we also have the Tesla Dojo system, which is a training system. And that's intended to be able to process fast amounts of video data to improve the training for the AI system. The Dojo system, that's like an fp16 training system and it is primarily constrained by heat and by communication between the chips. We are developing new buses and new sort of heat projection or cooling systems that enable a very high operation computer that will be able to process video data effectively.

我們還開發(fā)了特斯拉Dojo訓(xùn)練系統(tǒng)，旨在能夠快速處理大量視頻數(shù)據(jù)，以改善對(duì)人工智能系統(tǒng)的訓(xùn)練。Dojo系統(tǒng)就像一個(gè)FP16訓(xùn)練系統(tǒng)，主要受芯片的發(fā)熱量和通訊的速率的限制。所以我們也正在開發(fā)新的總線和散熱冷卻系統(tǒng)，用于開發(fā)更高效的計(jì)算機(jī)，從而能更有效處理視頻數(shù)據(jù)。

How do we see the evolution of AI algorithms? I'm not sure how the best way to understand it, except that neural net seems to mostly do is to take a massive amount of information from reality, primarily passive optical, and create a vector space, essentially compress a massive amount of photons into a vector space. I am just thinking actually on the drive this morning, have you tried accessing the vector space in your mind? Like we normally take reality just granted in kind of analog way. But you can actually access the vector space in your mind and understand what your mind is doing to take in all the world data. What we actually doing is trying to remember the least amount of information possible.

我們是如何看待人工智能算法的發(fā)展呢？我不確定這是不是最好的理解方式，神經(jīng)網(wǎng)絡(luò)主要是從現(xiàn)實(shí)中獲取大量信息，很多來(lái)自無(wú)源光學(xué)方面，并創(chuàng)建矢量空間，本質(zhì)上將大量光子壓縮為矢量空間。我今天早上開車的時(shí)候還在想，人們是否能夠進(jìn)入大腦中的矢量空間呢？我們通常以類比的方式，將現(xiàn)實(shí)視為理所當(dāng)然。但我認(rèn)為，其實(shí)你可以進(jìn)入自己大腦中的矢量空間，并了解你的大腦是如何處理所有外部信息的。事實(shí)上它在做的是記憶盡可能少的信息。

So it's taking a massive amount of information, filtering it down, and saying what is relevant. And then how do you create a vector space world that is a very tiny percentage of that original data? Based on that vector space representation, you make decisions. It is like a really compression and decompression that is just going on a massive scale, which is kind of how physics is like. You think of physics out physics algorithms as essentially compression algorithms for reality.

它獲取并過(guò)濾大量信息，只保留相關(guān)的部分。那人們是如何在大腦中創(chuàng)建一個(gè)矢量空間呢？它的信息僅占原始數(shù)據(jù)很小一部分，卻可以根據(jù)這個(gè)矢量空間的表達(dá)做決策。這實(shí)際上就類似一個(gè)大規(guī)模的壓縮和解壓縮的過(guò)程，有點(diǎn)像物理學(xué)，因?yàn)槲锢韺W(xué)公式本質(zhì)上是對(duì)現(xiàn)實(shí)的壓縮算法。

That is what physics does. Those physics formulas are compression algorithms for reality, which may sound very obvious. But if you simplify what it means, we are the proof points of this. If you simply ran a true physics simulation of the universe, it also takes a lot of compute. If you are given enough time, eventually you will have sentience. The proof of that is us. And if you believe in physics and the arches of the universe, it started out as sort of quarks electrons. And there was hydrogen for quite a while, and then helium and lithium. And then there were supernovas, the heavy elements formed billions of years later, some of those heavy elements learned to talk. We are essentially evolved hydrogen. If you just leave hydrogen out for a while, it turns into us. I think people don't quite appreciate this. So if you say, where does the specialist come in? Where does sentience come in? The whole universe is sentience special or nothing is? Or you could say at what point from hydrogen to us did it become sentient?

這便是物理學(xué)的作用。很明顯，物理公式是現(xiàn)實(shí)的壓縮算法。簡(jiǎn)言之，我們?nèi)祟惥褪俏锢韺W(xué)作用的證據(jù)。如果你對(duì)宇宙做一個(gè)真正物理學(xué)意義上的模擬，就需要大量的計(jì)算。如果有充足時(shí)間，最終會(huì)產(chǎn)生覺(jué)知。人類便是最佳證明。如果你相信物理學(xué)和宇宙的演化史，便知道宇宙一開始是夸克電子，很長(zhǎng)一段時(shí)間是氫元素，然后出現(xiàn)了氦和鋰元素，接著出現(xiàn)了超新星。重元素在數(shù)十億年后形成，其中一些重元素學(xué)會(huì)了表達(dá)。那就是我們?nèi)祟悾举|(zhì)上由氫元素進(jìn)化而來(lái)。若將氫元素放一段時(shí)間，它就會(huì)慢慢轉(zhuǎn)變?yōu)槲覀?。我覺(jué)得大家可能不太贊成這一點(diǎn)。所以有人會(huì)問(wèn)，specialist的作用是什么？覺(jué)知的作用又是什么？整個(gè)宇宙是一種特殊的覺(jué)知或者不存在特殊性？又或者，在氫元素轉(zhuǎn)變?yōu)槿祟惖倪^(guò)程中何時(shí)產(chǎn)生了知覺(jué)?

Q：Great. Our last question, congratulations on an incredible year so far at Tesla. How are things going at Gigafactory Shanghai? Is there any application of AI to manufacturing specifically at Giga Shanghai?

問(wèn)：最后一個(gè)問(wèn)題。祝賀特斯拉今年出色的業(yè)績(jī)，我們也想知道，特斯拉上海超級(jí)工廠目前的進(jìn)展怎么樣？在上海超級(jí)工廠有沒(méi)有一些制造業(yè)相關(guān)的AI應(yīng)用？

A：Thank you. Things are going really well at Giga Shanghai. I'm incredibly proud of the Tesla team. They're doing an amazing job. And I look forward to visiting Giga Shanghai as soon as possible. It's really an impressive work that's been done. I really can't say enough good things. Thank you to the Tesla China team.

答：謝謝，特斯拉上海工廠進(jìn)展順利，我為特斯拉團(tuán)隊(duì)感到無(wú)比自豪，他們做得很棒！我期待能盡快訪問(wèn)上海超級(jí)工廠，他們出色的工作確實(shí)讓我深感欣慰。我不知道該如何表達(dá)，真的非常感謝特斯拉中國(guó)團(tuán)隊(duì)。

We expect over time to use more AI and essentially smarter software in our factory. But I think it will take a while to really employ AI effectively in a factory situation. You can think of a factory as a complex, cybernetic collective involving humans and machines. This is actually how all companies are really.

預(yù)計(jì)未來(lái)我們的工廠中會(huì)運(yùn)用更多的人工智能和更智能化的軟件。但我認(rèn)為，在工廠真正有效地使用人工智能還需要花費(fèi)一些時(shí)間。你可以將工廠看作一個(gè)復(fù)雜的集合體，控制論集合體，其中涉及人也涉及機(jī)器。實(shí)際上所有公司都是如此。

Q：Thanks again you on for your time and joining us at this year's world artificial intelligence conference. We hope to see you next year in person.

問(wèn)：再次感謝您參加世界人工智能大會(huì)，也感謝您的精彩分享，我們期待著明年的大會(huì)能在現(xiàn)場(chǎng)見(jiàn)到您！

A：Thank you for having me in virtual form. I look forward to visiting physically next year, and I always enjoy visiting China. I am always amazed by how many smart, hardworking people that are in China and just that how much positive energy there is, and that people are really excited about the future. I want to make things happen. I cannot wait to be back.

答：謝謝您的線上采訪。我希望明年能有機(jī)會(huì)能親自參加，我很喜歡到中國(guó)。中國(guó)總是給我驚喜，中國(guó)有很多既聰明又勤奮的人，中國(guó)充滿了正能量，中國(guó)人對(duì)未來(lái)滿懷期待。我會(huì)讓未來(lái)成為現(xiàn)實(shí)，所以我非常期待再次回來(lái)。