访谈·中文精读

马斯克BBC完整采访

Elon Musk - the full BBC interview

中文精读由 AI 基于原始视频与文稿改写,非官方翻译,可能存在偏差或错误;表述以原视频为准。说明

2016年,罗里·凯兰-琼斯(Rory Cellan-Jones)坐在一辆特斯拉里采访马斯克。采访过程中,记者反复追问一个有点扎人的问题:这车我买不起,我认识的人也都买不起,那它到底对普通人有什么意义?

马斯克没有回避。他把特斯拉的路线拆成了三个阶段,也承认平价车还没造出来。但他给了一个判断:特斯拉这家公司,如果把眼光放到历史尺度上看,可能把整个汽车工业向电动化转型的时间提前十年,甚至更多。

这场对话里,马斯克谈的东西比一辆车多得多。电动车、自动驾驶、人工智能的风险、三家公司背后统一的逻辑,还有他那句听起来有点反常的自我评价:我始终觉得自己是错的。

特斯拉想给整个行业当催化剂

采访一开始,记者问他到底想在汽车行业里干什么。

马斯克的回答没有绕弯子:加速可持续交通的到来。他的原话是,想办法让整个汽车工业比原本更快地转向电动化。他说特斯拉长期的价值,就在于当一个催化剂。

这个定位值得多琢磨一会儿。大多数人理解一家车企,是看它卖了多少车、赚了多少钱。马斯克对特斯拉的定义却是,成功与否不看自己卖多少车,看的是它把整个行业推得多快。

换句话说,如果特斯拉造出了好车,逼得宝马、奔驰、通用都加快电动化,那即使特斯拉自己只占一小块市场份额,这家公司的使命也算完成了。

我看到的特斯拉的长期价值,是充当一个催化剂,加速向可持续交通的转变。(马斯克)

为什么非得是电动车

记者追问:为什么是电动车?为什么这件事重要?

马斯克给的理由很直接。电动车在车辆层面是零排放的,它不产生二氧化碳,不产生氮氧化物,不产生硫氧化物,也不产生任何内燃机汽车会排出的有害气体。

他补了一个前提:如果能源生产本身也是可持续的,再把可持续能源和电动车结合起来,人类就有了一个长期可持续的未来。

这个前提他没有省略。电动车环保不环保,很大程度上取决于电从哪来。烧煤发的电充进车里,账不能这么算。马斯克心里清楚这一点,所以他把可再生能源也放进了同一个盘子里。电动化要成立,得跟能源结构的清洁化一起走。

记者接着问:十年之内,我们是不是都会开上电动车,还是说它永远只是少数人的玩具?

马斯克的回答比大多数人敢想:除了火箭,所有交通方式最终都会完全电动化。

他再一次强调特斯拉的角色:他要的是让这个未来来得更快一些,抢占这个未来不在他的计划里。

三步走,一步都不能跳

记者抛出的第一个尖锐问题,是关于价格的。这辆车很好开,很让人兴奋,但现在它只是给富人准备的。它真的能对大众市场有意义吗?

马斯克没有否认。他把特斯拉的战略从第一天起就说清楚了:分三步。

第一步,做一款产量低、价格高的车。那就是最早和路特斯(Lotus)合作打造的那辆跑车。第二步,做中等价位、中等产量的车,也就是 Model S。第三步,做高产量、低价位的车,就是 Model 3。

为什么不能一开始就造平价车?马斯克的解释是,任何一项新技术都需要时间打磨,需要好几代产品迭代,需要规模经济,才能让普通大众买得起。

他拿手机打比方。早期的手机又大又重又贵,可经过一轮又一轮的设计迭代,加上产量放大,今天一百美元就能买到一部口袋里的超级计算机。

技术的普及走不了直路,靠的是一代代产品和一次次放量,把成本一点点压下去。第一代产品往往又贵又粗糙,只有少数愿意尝鲜又付得起钱的人买。这些人付的钱,养活了下一代产品的研发。到了第五代、第十代,成本降下来了,性能稳了,普通人才进场。

特斯拉的三步走,本质上就是照着这个规律走。先造出让人相信电动车可行的高端产品,用利润和声誉为后续研发输血,再一层层往下够,直到够到普通人。

这套打法的前提是,第一步必须真的站得住。如果第一款高价车卖不动,后面的两步就全断了。所以特斯拉早期的每一款车,都背着一个任务:证明电动车可以做得让人想要。

“我买不起”引发的正面交锋

这段对话里最有意思的部分,是记者和马斯克关于价格的来回拉锯。

记者说得很实在:我买不起这辆车,我认识的人里也没有买得起的。

马斯克承认这车不便宜。他的回应是,如果你把账算全,结论会不一样。他建议用租赁或融资的方式来算,把电动车的低使用成本算进去,那么每个月的交通支出,其实比第一眼看上去要划算得多。

这个算法值得拆开看。一辆燃油车的开销,买车只是一部分,油费、保养、维修、税费在后面排着队。电动车的结构简单,没有变速箱,没有那么多需要定期更换的零件,日常用电的成本也远低于烧油。把三五年摊开算总账,价签上的那个数字就不是全部了。

但记者不吃这一套,继续问:你们到底有没有在造一款更多人买得起的车?

马斯克说这就是 Model 3,战略的第三步,高产量、低价格,预计第二年年底投产。记者追问是不是按时交付,马斯克的判断是,要想对交通产生实质影响,车必须便宜。没有平价车,特斯拉对世界的影响就很小。

这段交锋点破了一个常被忽略的事实:一家公司嘴上说的愿景,和它实际能触达的人群,中间隔着一道成本的门槛。马斯克自己心里清楚,所以他反复强调 Model 3 的重要性。愿景再宏大,最后都得落在一辆普通人买得起的产品上。

记者还问他一个更私人的问题:这件事对你个人重要吗?你在乎的到底是生意,还是让可持续交通的愿景成真?

马斯克的回答是,整个特斯拉存在的意义就是加速这个转变。如果一开始就能造出平价车,他们肯定会那么做。只是技术需要时间成熟,需要迭代,需要规模。

S曲线,以及人们为什么总低估新技术的普及速度

关于电动车多快能普及,马斯克给了一个判断:它会比人们想象的更快。

他讲了一个技术普及的规律。新技术的采用曲线通常是S形的,起步很慢,然后加速,最后趋缓。问题在于,人们做预测时习惯用直线外推。在S曲线的早期,你用直线去猜,结果一定低估实际速度。

他给了特斯拉自己的数据。2015年年底,全球路上总共有大约五万辆特斯拉。2016年,特斯拉又生产了五万辆,也就是说整个车队一年翻了一倍。2017年,大概还会再翻一倍。

这个翻倍的速度放在S曲线的语境里就很好理解了。曲线早期看着不起眼,一旦过了拐点,增长会显得很猛。用直线思维看,会觉得每年涨一点;用曲线思维看,会发现翻倍才是常态。

直线思维为什么这么顽固?因为人的直觉是从过去推未来。看到去年涨了一万辆,就假设明年也涨一万辆。可技术扩散在早期靠的是早期用户,数量本来就有限;一旦产品进入主流价格带,触达的人群一下子放大几十倍,增速就变了。

马斯克没有说电动车什么时候能占到新车销量的一半。他只说了一件事:每年翻倍这个节奏如果持续下去,结果会比大多数人的预测来得早。

底特律为什么会睡着

记者问了一个挑战性的问题:传统车企好像醒过来了,宝马这类公司是不是最终更有可能把电动车带给大众,而不是特斯拉?

马斯克的反应是,如果他们真能做到,那是好事,全力支持。他甚至希望所有厂商都尽可能快地多造电动车,因为这对世界是好事。

但被问到传统车企过去的表现时,他不客气了。

他说这些公司不太相信电动车,原因有三个。第一,他们觉得技术上不可能造出有吸引力的电动车。第二,他们觉得即便有人造出长续航、高性能的电动车,人们也不会买,因为大家对汽油有某种深爱。第三,他们认为五分钟就能加满油这件事是不可替代的刚需。

马斯克说,特斯拉做的,就是把这几个“公理级的错误”一个个拆掉。它的跑车证明了一件事:电动车可以很快、很好看、性能强、操控好。如果你造出这样一辆车,而且续航足够,人们是会买的。

他还讲了一个流传很广的故事。2007年年中特斯拉宣布跑车之后,通用汽车的一位高管把新闻稿拿给自己的工程师看,问了一句:加州一家小公司都能做到,我们为什么做不到?这件事后来促成了通用的电动车项目,也就是雪佛兰沃蓝达(Chevy Volt),再之后又带动日产做聆风(Nissan Leaf)。

不过,马斯克对传统车企的评价并不高。他认为这些项目绝大多数产量很低,除了聆风之外,更多只是在满足监管要求,真心想把电动车做大的没几个。直到特斯拉在正常的细分市场里卖出了相当多的车,局面才开始改变。

这段批评里藏着一个行业规律:大公司判断新技术,往往拿现有的成本结构和用户习惯当参照。参照系没变,结论就不会变。只有当一家小公司真的把产品做出来、卖出去,参照系被强行改写,大公司才会动。

Model S 卖赢了奔驰宝马奥迪

马斯克在这里给了一个很能说明问题的数据。在豪华轿车这个细分市场里,Model S 是当年美国卖得最好的车,销量超过奔驰、宝马和奥迪在同一个细分市场的车型。

这个事实的分量在于,赢下来的地盘是一个正常的、跟燃油豪华轿车正面竞争的细分市场,不是什么“环保小众市场”。买这辆车的人,本来可以在奔驰 S 级、宝马 7 系、奥迪 A8 之间挑,最后挑了特斯拉。

特斯拉证明的不只是电动车能造出来,它还能在传统车企最赚钱的地盘上打赢他们。豪华轿车是整个汽车工业利润最厚的一块,谁在这里站稳了,谁就有资格往下打。

苹果造车,藏不住的秘密

记者把话题转到科技公司造车上。谷歌在做自动驾驶,苹果是不是也在造车?这会不会对特斯拉构成威胁?

马斯克的态度是鼓励更多人参与,因为造车这事很难。但他判断,像苹果这样的公司大概率会造出一辆有吸引力的电动车,因为这看起来就是件显而易见该做的事。

记者问他是不是得到什么风声。马斯克的回答带着一点调侃:如果你雇了一千多名工程师去做一件事,其实很难把它藏住。

记者追问,你觉得苹果是认真的?他说,是。记者说这是公开的秘密,他承认。

至于威胁,马斯克看得比较开。他认为苹果的进入会扩大整个行业,而特斯拉的目标始终是造出最有吸引力的电动车,同时也尽力帮别的公司把电动车造好。

他举了一个例子。前一年特斯拉把所有的专利都开源了,任何人都可以免费用特斯拉的专利。这个动作放在普通商业逻辑里很难理解,专利通常是护城河。一家把“加速行业转型”当作使命的公司偏偏反着来:把门打开,让更多人冲进来一起把这个市场做大。

开源专利这件事,短期看是让利,长期看是改变游戏规则。电动车要真正普及,靠一家公司造车是不够的,得有几十家、上百家公司一起把供应链做起来,把充电网络铺开,把成本压下去。特斯拉把专利放出去,等于在自己的使命上押了重注。

电动化加自动驾驶,两大革命同时发生

采访的后半段,话题转向自动驾驶。记者问这条路会走多远。

马斯克给了一个很高的定位。他说,交通领域最大的两场革命,一个是电动化,一个是自动驾驶。这是自流水线诞生以来最大的两个创新,而且它们几乎在同一时间发生。

他对自动驾驶的判断很绝对:长期来看,没有人会买一辆不能自动驾驶的车。他的类比是手摇电梯,今天没人会想买一部要自己摇上去的电梯。

记者问,十年后我是在开车,还是只在手机上按个按钮,车自己开过来把我送到目的地?

马斯克说,你只会在想开的时候才开。不自动驾驶的车,长期来看就像养一匹马,你会出于感情养它,但不会拿它做日常通勤。

这个类比有点刺人,但逻辑是通的。一百年前,马是主要交通工具,汽车刚出来的时候,很多人觉得它吵闹、危险、加油不方便。等到汽车比马便宜、比马快、比马可靠,马就退到了周末骑一骑的位置。自动驾驶对人工驾驶的替代,路径大概也是这样。

别怕车,该怕的是别的东西

记者注意到马斯克一直对人工智能表达担忧,于是问了一个很自然的问题:这辆车越来越智能,你是不是担心它哪天发展出自己的意识?

马斯克的回答把两件事分得很清楚。他说这辆车是“窄人工智能”(narrow AI),它没有深度智能,只是看车道线、控制方向。他甚至认为这个问题基本上已经解决了,剩下的只是打磨技术细节、把它推向市场、把可靠性提到足够高。

他给了一个具体的门槛:自动驾驶需要很多个九的可靠性,大概99.9999%这个量级。而且,一辆自动驾驶车要比人类驾驶安全一个数量级,也就是十倍,到那时候关于“哪个更安全”的争论才会彻底消失。

很多个九是什么概念?99%听起来不错,但那意味着每一百次里有一次出错。99.9999%意味着每一百万次里才有一次出错。自动驾驶每天要处理几十亿次判断,门槛必须定在这个量级,才有资格谈上路。

但这一切仍然只是窄用途。车不会发展出意识,也不会决定接管世界。

他真正担心的是另一件事。如果出现某种庞大的智能,它要么发展出自己的意志,要么被少数人掌控,那人类的未来可能就不太妙。

这个担忧可以拆成两半。第一半是失控:一个目标跟人类不一致的超级智能,会做出人类不希望看到的事。第二半是集中:即便智能本身没有恶意,只要它掌握在极少数人手里,权力结构就会失衡。两半加起来,风险都不小。

记者问他为什么会有这种担忧。马斯克提到了哈兰·埃里森(Harlan Ellison)的短篇小说《我没有嘴,但我必须尖叫》(I Have No Mouth, and I Must Scream),说这篇小说会让人做噩梦。记者问他读科幻是不是读出了噩梦,他承认,确实做过噩梦。

记者提到自己前一年采访过史蒂芬·霍金,霍金也有类似的担忧,然后问这场担忧是眼下的事,还是离我们很远、现在不用操心。

马斯克的回答是:它会比任何人以为的都快。

他给了几个观察。计算机智能的复杂度每年都在戏剧性地增长,他判断人类正处在一条人工智能指数级进步的曲线上。同时,投入AI研发的聪明人数量也在急剧增加。AI会议的出席人数每年翻倍,会场都坐满了。

他还讲了一件私人的事。他有个年轻的表弟刚从伯克利毕业,学的是计算机科学和物理。他问表弟,聪明学生里有多少在学AI?答案是,全部。

“全部”这两个字,比任何数据都更能说明趋势。一个领域如果聚拢了最聪明的一批年轻人,它的进展速度就不会按常规节奏走。

三家公司,一个底层逻辑

记者把话题拉回到马斯克本人。他同时身处三个行业:电动车、火箭、太阳能。记者问,这三件事背后有什么共同点?

马斯克的回答是,他试图做的是把未来可能出现的生存性威胁降到最低,或者尽自己所能确保未来是好的。

这个回答把三家公司串到了一起。电动车和太阳能,对应的是气候和能源的长期风险。火箭和太空,对应的是人类只待在一个星球上的脆弱性。三条线指向同一个目标:让人类文明的存续概率高一点。

他坦承,自己当初并不觉得这些公司会成功。他以为它们大概率会失败,尤其是特斯拉和 SpaceX。他甚至给过一个具体的数字:一开始他觉得 SpaceX 和特斯拉大概只有10%的成功率。SolarCity 成功的概率他觉得更高一些。所以今天看到它们还活着,他自己都觉得意外。

记者问,现在是不是更有信心了?

马斯克说,他确实感觉这些公司到了一个阶段,短期内不太可能死掉。特斯拉和 SpaceX 都积累了大量客户,赢得了很多人的信任。他说了一句挺朴素的话:当你的客户真心希望你成功的时候,你就会成功。

“我始终有点错”

采访快结束时,记者问了一个带点挑衅的问题。你是个很有竞争心的人,跟贝索斯在火箭上竞争,跟底特律在汽车上竞争,跟监管者在太阳能上较劲。驱动你的是不是一种“我是对的,他们都错了”的心态?

马斯克的回答出人意料。他说,他实际上真的认为自己始终在某种程度上有错。他的目标,是少错一点。

不管你是什么人,我们始终都在某种程度上有错。我想做的,是随着时间推移,把自己犯错的成分尽量减小。(马斯克)

这句话放在一个被称为“钢铁侠”的人嘴里,有点反差。但如果把整场采访连起来看,它其实很自洽。特斯拉的三步走,本质是承认技术需要迭代,一步到不了位。自动驾驶要很多个九的可靠性,本质是承认错误必须被压到极低。他对AI的担忧,本质也是承认人类可能选错方向。

他最后说,有几件事对未来的走向很重要。可持续能源、可持续交通,最终成为多行星物种,走向星辰。这些事情让他对未来的想象是明亮的。如果这些事情不发生,未来在他看来就相当暗淡。

这场采访发生在2016年。回头看,马斯克当时说的很多东西,有的兑现了,有的比预想慢,有的比他当时描述得更曲折。但那份底层判断,也就是技术会以S曲线的速度推进、成本会被规模一点点压下来、行业需要有人去当催化剂,这些放在今天依然成立。

这场访谈真正有用的地方,在于他看问题的那套方法:别用直线猜未来,算账要算全,承认自己会错,然后把错误的成分一年年往下压。至于马斯克是不是永远正确,那倒没那么重要。

  • 评估任何一项新技术时,先用S曲线思维想一遍,而不是直线外推。早期看起来慢的东西,可能只是还没到拐点。
  • 看一家公司,先看它有没有在改变整个行业的成本结构。只卖高价产品的公司,影响力有天花板。
  • 遇到“我买不起”的质疑,别急着接受或反驳,把账算全:购置成本、使用成本、融资成本,分开放,再比总账。
  • 看到“AI威胁论”时,先分清是窄AI还是通用AI。车看车道线不叫危险,被少数人掌控的庞大数据智能才值得警惕。
  • 判断一个行业的转型速度,去看头部公司车队的年增长率,翻倍往往比百分比更能说明处在曲线的哪一段。
  • 给自己留一条“我始终有点错”的底线:每次决策后复盘,把错误的成分记下来,下一次少犯一点。
YouTube 平台字幕(自动或人工),可能有识别错误,人名与数字请以视频为准。已按语句合并成段落,并去掉 um、uh 等口头填充词。 文稿来源:YouTube(自动/人工字幕)。

0:00Elon Musk what's their Vision here what are you trying to do to the car industry well I'm trying to accelerate the Advent of sustainable transport so try to get try to get the econ industry to move towards electrification faster than they would otherwise move so the the the fundamental value long-term value that I see for Tesla is serving as a catalyst to accelerate the transition to sustainable transport why electric cars what why why is that important well electric car is are zero emission at the car level so they're not producing CO2 or nitrous oxide or sulfur oxides or any of the sort of the noxious gases that any kind of combustion engine car

0:46would produce and so if if we can have sustainable energy production and combine that with electric cars we have a long-term sustainable future and how how's that going I mean you've made made a deal of progress but do you see us all having electric cars in 10 years time or is it going to be always a minority Pursuit no I think All Transport with the exception of rockets will go fully electric so that's why I mean I see the the value of of Tesla as an as an accelerant as a catalyst in that transition I think Tesla maybe when one looks back on it from an historical perspective it it might accelerate that transition by a decade maybe maybe more now this is a a great car it's an

1:37exciting drive but at the moment it's just for the rich isn't it is this really going to mean anything for the mass Market yeah I mean the the the Tesla strategy from the beginning has been to start off with a low volume high- price car that was the sports car that we first did in partnership with lotus and then we have the model S which is kind of a mid-priced mid volume C car and and actually when you when you look at the the price of the model S inclusive of the cost of petrol it it actually is a lot more competitive than it would than you'd think I'm not going to be able to afford one of these and nobody I know is well I think

2:16it's it's certainly not a low pric car U I'm just saying that if you were to lease or Finance the car and compare the the lease or Finance cost with a much lower cost of operation of an electric car your monthly cost of Transport is a lot more competitive than it may seem at first glance just looking at the price of the car but are you on the road to producing a model that far more people could so the the model 3 that's the third part of our strategy which is produce a high volume lowcost car

2:53and we expect to be in production with that at the end of next year is that all on schedule and what what are the Ambitions for that car is is that the one that will really make this revolution that you've been talking about happen I no question in in order to have a a substantial effect on on

3:10transportation we have to make the cars affordable so I think the model 3 is extremely important as part of that strategy unless there's an affordable car we will as you alluded to earlier only have a small impact on the world so it it needs we need to make a car that most people can afford in order to have a really substantial impact how important is it for you that people's lives are changed in this way obviously it's important for this business because you've lost a lot of money so far you need to have a mass market car but is it important for you as a thinker that that people have electric cars that that that sustainable Vision comes true yeah

3:51absolutely mean the like I said the whole point of Tesla is to accelerate the transition to sustainable transport if we could have made an affordable car right off the the bat we definitely would have it's just that it takes time to refine the technology when you think of any new technology it takes multiple versions and economies of scale in order to make the technology accessible to the general public you can look at say cell phones when the early days of cell phones they were very big and heavy and very expensive but with successive design iterations and scaling up the production we've brought it to the point where for $100 you can buy basically a

4:31supercomputer in your pocket how soon will most people have an electric car is this a very long-term Vision or is it going to happen soon well I think it's going to happen sooner than people think I mean the typical curve of new technology adoption it it's is like an S curve so it starts off quite slow and

4:55generally people tend to predict to base their predictions on a straight line as opposed to a curve fit and but when you have an an S curve of adoption if you make a straight line extrapolation at the beginning of that S curve it's always going to underestimate the actual adoption rate so but if

5:18you look at say what Tesla is doing every year we are doubling our total cumulative production so at the of last year we had 50,000 cars in total on the roads worldwide and then last year we produced another 50,000 cars so the total Fleet of Tesla vehicles doubled last year it will approximately double again this year meanwhile Detroit the traditional motor makers seem to have woken up is

5:51is it not true that the likes of BMW are eventually more likely to bring electric vehicles to the masses than Tesla well I say say more power to them if they I mean I would certainly encourage their actions on electric vehicles and and for all all manufacturers I hope they produce as many electric vehicles as soon as possible that's the the right thing for the world what do you think their record has been like so far has has has Detroit and the traditional motor industry been caught napping well they they didn't really believe in electric cars they didn't think that it was technologically possible to create a a compelling electric car and they felt that even if even if

6:33someone did create an electric car with long range and high performance that people wouldn't want to buy it because they have some deep love of gasoline and and that it's vital to be able to refuel in 5 minutes so with with Tesla we were able to disprove those axiomatic errors so with the the Tesla Roadster we're able to show that you can in fact create a very fast attractive high performance great handling electric car and that if you made such a car with with long range and if you were if you make such a car people would buy it so with with the with the Tesla roader in fact the that's really what got General Motors to do the Chevy Vault and Nissan to do

7:21the leaf so we we announced the Tesla rooster in mid2 2007 and blah BL he's told the story many times took the took out press release to his engineers and said if a little company in California can do this why can't we and that's what got General Motors to do their electric vehicle program which in turn encouraged the other manufacturers to do through electric vehicle programs as well but but those programs for the most part have been or really I'd say almost entirely with the exception of maybe the leaf have been of quite low volume

8:00and more more more inclined towards satisfying Regulators but but now I think we are seeing a change because people are seeing that Tesla is able to sell quite a few quite a few electric cars in a normal car segment in in in the premium sedan segment in in fact we were the the bestselling car in the in our segments in the United States last year so more than you know I miss more

8:36than Mercedes or BMW or Audi in in our segment the technology industry generally suddenly seems incredibly enthusiastic about getting into the motor industry we've obviously had Google with it self-driving car do you see that continuing and do you do you see apple building a car and that maybe being a threat to you well I I think I I I would encourage

9:04more participation by whoever it is to create electric vehicles it it's it's quite hard to do but I think companies like apple will probably make a compelling electric car it seems like the obvious thing to do are you betting that that's going to happen have you heard anything well it's pretty hard to hide something if you hire over a thousand Engineers to do it so you think apple is serious about it yeah I do this is an Open

9:37Secret and will that be a threat to you or will that just expand the industry I I think that I mean I think it will expand the industry certainly Tesla would aspire to still make the most compelling electric vehicles and that would be our goal while at the same time trying to help other companies make electric cars as well so for example last year we open sourced all of our patents so anyone can use any of our patents for free you've also been a leader in autonomous driving in this road to the self-driving car how far do you think that's going to go I mean I think the two biggest Revolutions in transport are electrification and autonomy th those

10:22are the and it's you those are the two biggest Innovations since the moving production line and they're both happening at about the same time I think autonomy is extremely important and I think in the long term nobody will buy a car unless it is autonomous it would be like having a manually operated elevator or something like that so it's a strange agonism in 10 years time what will I be driving or will I be driving at all will I just be pressing a button on an app a car will drive up and and take me where I want it yeah you will you will only drive if you you want to drive it'll be like owning a owning a car that is not self-driving in in the long term

11:07will be like owning a horse you would own it for and you would use it for sentimental reasons but not for you know not for daily use really unlikely now you you've taken a lot of interest in artificial intelligence H this is becoming an artificially intelligent car well I would quote narrowly narrow artificial intelligence you've also expressed concern about where that's heading is that because you've looked at what that car can do and have thought what happens if it develops a mind of its own why why are you wor we don't need to be concerned about the cars the car this is I mean car a car is not deep AI it's

11:49a it's a narrow use case you know we're not trying to build sentients into the car it's just trying to look at the lines on the road and steer correct and I would consider that to be essentially a solved problem it's just a question of refining the details of the technology and bringing that to Market and then improving the nines of reliability so in order to have a self-driving car you really have to have many nines of reliability so it's 99.9999% something like that is is is how good it needs to be or you know if let's say to First

12:33approximation you would want a self-driving car to be in order of magnitude safer than a human driven car and if if if you're like okay it's 10 times safer than it's like there's no more doubt there's no more debate about which one is safer but it's it's still a narrow use case the the cars not going to develop Consciousness or decide that they want to take over the world or something like that I think we really need to be more concerned about

13:08D yeah and why do we need to be concerned about that well I mean because there are I think scenarios where if there's some vast intelligence that either develops the will of its own or is subject to the will of a small number of people then we could have an undesirable future if you want to read a real scary one I'd say Haren Ellison's I have

13:47no mouth and I must scream we'll give you nightmares you read a lot of Science Fiction is it giving you nightmares at the moment this it has given me nightmares yeah and are you genuinely what when you you've been one of the Prime movers about AI I interviewed Steven Hawking last year right had similar concerns is that a a short-term concern or is it something way off that we don't need to worry about for a long time it's going to come faster than anyone appreciates I think it's with with each passing year the sophistication of of computer intelligence is is growing dramatically I I mean I really think we're on an exponential Improvement path of

14:30artificial intelligence and and the number of smart humans that are developing AI is also increasing dramatically I mean if you look at like the attendance at the AI conferences they're they're doubling every year they're getting full I have a a sort of a young cousin of mine who's graduating from Berkeley in computer science and physics and I asked him like well how many of the smart smart students are studying AI in computer science and the answer is all of them now you're in three extraordinary Industries you're in electric cars you're in Rockets and you're in

15:13solar energy what unites you what unites those those three interests well the what I'm trying to do is to minimize existent future existential threats or to take whatever action I can to ensure the future is good I didn't expect these companies to succeed I thought they would most likely fail particularly Tesla and facex I thought Sol City had a higher probability of success but I in the beginning I thought SpaceX and Tesla maybe had a 10% chance of success and so I'm quite surprised really to see that they're not that we're alive it's great I wasn't expecting that

15:59and are you now more confident that they each of them will be a success I I I do feel it's though they've reached a level of progress that makes it unlikely that they will unlikely that will they will die in the near term both Tesla and SpaceX have a lot of customers sort of built really earn the trust of a large number of customers and and these are really solid organization or or like really you know I think thoughtful end customers for the car and I and I think like whenever

16:37your customers really want you to succeed then you then you you will succeed finally you're obviously a very competitive person you're competing with the likes of Jeff Bezos in Jeff who in rocket you're competing with Detroit in the motor industry you're competing with regulator when ites to Solar City what drives you is it just I am right and they're all wrong well no I mean I I actually really take the position that I'm always to some degree wrong and the aspiration is to be less wrong we're always to some degree wrong all it doesn't matter who you are and I think Trying to minimize the wrongheadedness over time is that I believe in that philosophy

17:31and and I mean I I I think these there's some things that are important for the future sustainable energy obviously sustainable transport ultimately ultimately becoming a multi-planet species and and traveling out there Among the Stars I think those are those are great things those are things that make me like the future and or be inspired about the future whereas if those things don't happen the future I think looks quite dim and

18:06and I I just feel quite fortunate that we've made the progress that we have on on those fronts and we'll aspire to make more progress in the future you don't know

资源与出处

相关视频

同频道、同主题或相邻年份