2014年,牛津,金方向盘颁奖典礼。马斯克上台领终身成就奖。
台下第一排,一位提问者坐在德国某大车厂CEO旁边,凑过去低声问了句:这人是不是对你们很危险?
对方的回答很干脆:不用担心。第一,电动车这个想法永远不会成为大众市场。第二,硅谷这帮人根本不懂工程,也不懂怎么造一辆真正漂亮的好车。所以没什么好担心的。
那年特斯拉市值230亿美元,大众860亿。
六年之后,同一个话题的数字变成了:特斯拉5360亿,大众770亿。特斯拉比宝马、大众、戴姆勒三家加起来还大2.5倍。
这段往事是马斯克自己在2020年12月1日的一场访谈里翻出来的。那场对话从火星时间表一路聊到柏林工厂、自动驾驶、人工智能、人口崩溃、脑机接口,最后落到他四十岁生日派对上和相扑选手摔跤,颈椎间盘突出,五年才好。
信息量很大,值得慢慢拆。
火星这件事,他给了一个可以验算的时间表
提问者问得很直接:人类第一次踏上火星,现实点说,什么时候?
马斯克的回答没有绕弯:六年,我比较有信心。运气好,四年。
他还顺手把算法讲了。地球和火星的轨道同步窗口大约每26个月出现一次。2020年夏天刚有过一次,所以大约两年后会有下一次,再往后两年又一次。按这个节奏倒推,两年后先送一艘无人飞船过去探路,四年后如果一切顺利可以上人,六年是留了余量的稳妥答案。
至于SpaceX第一次把人送进轨道,他说:不好说,可能两三年。
注意这里的时间尺度。2020年底说两年后发无人飞船,意味着2022年前后;四年上人,意味着2024年前后;六年,意味着2026年前后。他后面所有的表述都围绕这个锚点展开。你可以不信,但这个时间表是他公开给出来的,可以拿来对照。
他反复强调的重点,落在“很多人”这三个字上。他要开发的是能让大批普通人前往火星的技术,一次象征性的登陆不算数。月球上要有基地,火星上要有城市,而且这座城市要能自我维持。他的原话是,这件事要尽可能早地发生。
我对地球上的未来是乐观的。但为生命整体买一份保险,这件事很重要。(马斯克)
提问者追问:这算商业旅游,还是地球万一出事的Plan B?
马斯克否掉了“Plan B”这个说法。他说这更像两件事叠在一起。一件是让人有一个值得兴奋的未来,另一件才是备份。而在他看来,第一件的分量一点都不比第二件轻。
一个成为太空文明、散落在群星之间的未来,每个孩子听到都会兴奋。你根本不用教他们,那是本能的。人需要有理由对生活感到兴奋,早上醒过来不能只面对一堆问题。(马斯克)
然后话题拐到一个略黑色幽默的地方。提问者问他是不是真的想葬在火星上。他说,是啊,反正人总有一死。接着补了一句:只是别是撞击造成的就行。“生在地球,死在火星”,在他看来是个挺酷的句号。
特斯拉和SpaceX,各自该拿什么尺子量
提问者提到两年前和马云的一场对话。马云当时的说法是,贝佐斯管轨道,他管地球。而马斯克好像两边都管。
马斯克的回答把两家公司的定位讲得很清楚:特斯拉是让地球上的未来变好,SpaceX是让地球之外的未来存在。
紧接着他说了一段关于“怎么衡量一家公司好不好”的话,这可能是整场访谈里最值得记的一段。
如果你从未来往回看,问特斯拉的根本价值是什么,我会说,应该看它把可持续能源的到来提前了多少年。这就是我衡量特斯拉的方式。对SpaceX也一样,看它在多大程度上提高了人类成为太空文明的概率。(马斯克)
这个尺子很有意思。它衡量的对象是“加速了多少年”,利润和市场份额不进入计算。按这个标准,一家公司哪怕一直不赚钱,只要把某个不可逆的进程往前推了十年,它的价值就是正的。
聊到这里,提问者抛出了那个大众CEO的故事,以及230亿对860亿、5360亿对770亿的对比。马斯克听完只是笑,说我还能怎么办呢。
提问者顺势问:股价是不是太高了?他答得也坦白:以前股价一百、八百美元(拆股前)的时候就有人说太高,别听我的。然后自嘲了一句,证券交易委员会又要来投诉了。
那特斯拉会不会去收购一家传统车企?
他的回答划了一条线:不会发起恶意收购。但如果有人主动上门说,我们觉得和特斯拉合并是个好主意,那这个对话当然可以谈。恶意的不做,友好的可以聊。
再往前推十几年,气氛完全两样。2007年第一代Roadster发布的时候,行业的普遍反应是:你们就是一群傻子。谁去创办一家汽车公司,谁就是疯了,钱会全部赔光。
马斯克自己的判断当时也没多乐观。他说:我觉得我们大概有10%的成功概率。别人跟他说你会失败、你会赔光一切,他的反应是:是啊,大概是真的,还有什么新鲜事。
全自动驾驶明年就来,但路上的车要换二十年
这是整场访谈里最容易被误读的一段,马斯克自己主动做了澄清。
先说他给的结论:十年后,几乎所有新车都会具备完全自动驾驶能力;十年后绝大多数车是电动的,大概70%到80%甚至更多。
然后他讲了一个大多数人没算过的账。全球在路上跑的汽车和卡车大约有20亿辆,而每年的新车产量只占整个车队的5%左右,也就是大约1亿辆。这意味着什么?即便从某一天开始,所有出厂的新车都是自动驾驶的,要把路上这20亿辆全部换掉,还得再花20年。
有些人习惯了手机两三年换一次,就以为全世界的车也会这样更新。不是的。汽车是贵得多的资产,寿命长得多。从所有新车都电动的那一天算起,还要再20年,整个车队才会完成替换。(马斯克)
这个数字上的错位,是很多关于自动驾驶的讨论跑偏的根源。技术拐点和保有量拐点,中间隔着二十年。
至于特斯拉自己的进度,他说得很满:我极度有信心在明年实现完全自动驾驶,并推送给特斯拉客户。后面那段不确定的,是监管批准要多久。
他的论据是数据。如果能积累几十亿公里的自动驾驶里程,把自动驾驶状态下的事故率和非自动驾驶状态下的事故率摆在一起,争论的空间就不大了。他说现在的统计已经显示出巨大差异,即使用了Autopilot的现有版本,安全性也明显更高。
欧洲会不会落后?他给了个不太客气的评价:欧盟监管机构是最保守的。他们大概每六个月才碰一次头。特斯拉的欧洲用户对此挺不高兴。但他判断,至少会有一些司法辖区在明年放行完全自动驾驶。
关于“人还该不该自己开车”,他的态度比很多人想象的温和。他说自己不反对开车,人们会一直开下去,能开多久开多久。只是自己开车这件事会变得越来越不寻常。
在漂亮的地形里、蜿蜒的山路上开一辆操控好的车,当然有意思。但在可怕的堵车里开车,一点意思都没有。(马斯克)
他给了个很具体的数字:人们平均每天花一个半小时,可能两个小时在车上,加州这种地方很常见,有些人一天通勤三个小时。让这段时间白白烧掉,才是他要解决的问题。
提问者顺势提起了他当年那个电梯的比喻:一百年前没人能想象一部没有电梯员的电梯,今天没人能想象一部有电梯员的电梯。这个类比放到车上,就是他现在押的注。
为什么选柏林,他顺带调侃了柏林机场
一年前在同一个场合,马斯克宣布要在柏林附近建厂。几个月后的六月正式动工,目标次年七月完工。提问者说,他们早上刚去工地转了一圈,进度惊人。在德国,尤其在柏林,一个工程能按时按预算完工,本身就够上新闻。
提问者说,你等于造了一个“反柏林机场”项目。柏林机场(BER)修了十四年,预算翻了几倍,2020年10月才通航。
马斯克先说了对德国的感情:我是德国的超级粉丝,很多朋友在这儿。然后是选址的实际理由,这段讲得很实在。
柏林是个很好玩的城市。年轻人能在市区用合理的价格租到公寓,有家庭的人也能负担得起房子。它对各个年龄段的人都友好。(马斯克)
他还顺口评价了一句:而且它确实性感。
关于工厂的开幕派对,他把计划描述得像一场音乐节:大家从白天开始,先是家庭友好的音乐,然后逐渐变硬,到午夜转成techno,一直放到天亮。所有在场的人都被邀请了。
提问者问,你今晚睡哪儿?
答:工厂里。
准确说,是工厂里的一间会议室。厂还没建完。他说这样能让他对现场的进展有直接的体感。提问者问一个人吗,他说应该是吧。然后补了一句,这也算一种邀请。
AI是他眼里最大的那个风险
话题转到人工智能。提问者提醒他,上次问哪个项目最重要,他给的答案既不是特斯拉也不是SpaceX,AI将来在社会里扮演的角色排在最前面。
马斯克的论证从一句话开始:人类长期以来是地球上最聪明的生物,这件事要变了。
他谈的对象是通用人工智能(AGI),在各个维度上都比人更聪明、甚至能模拟一个人的那种。今天市面上这些专用AI,不在他担心的范围内。他用的词是“应该担心”,紧接着的诉求是:AI的发展应该有政府监管,尤其是超级先进的AI。
他给的理由很朴素:任何对公众有潜在危险的东西,我们都默认它需要监管来保障公共安全。
提问者问,你是不是担心有一天人类要为机器服务,而不是反过来?
他的回答挺扎心:
说实话,当我看到人们盯着手机的样子,我觉得我们已经在为机器服务了。每次你搜索、每次你录入信息,你都在参与构建这个数字群体心智。(马斯克)
他把AGI的到来称为“奇点”,并解释了为什么用这个词:就像黑洞的奇点,你没法预测那里会发生什么。他强调,AGI的到来本身不一定是坏事,坏只是可能性之一。
至于时间,他提到雷·库兹韦尔(Ray Kurzweil)预测的2025年,评价是“相当准确”。
那怎么防止它变成威胁?他的答案还是监管,而且拿了几个现成的类比:汽车、飞机、食品、药品,都有监管机构盯着,确保公共安全。数字超级智能同样有公共安全风险,所以也该有人盯着。
我们不会让人为了好玩在后院造一颗核弹。数字超级智能可能比核弹更危险。不能让病人来管理精神病院。(马斯克)
数据归谁?他的态度很硬:每个人都该拥有自己的数据。不能因为某个网站的服务条款,稀里糊涂地就把数据交出去了,这太荒唐。他补了一句:那些条款本来也没人读。
提问者抛出了那个常见的反驳:你管得住自己,中国呢?中国的监管框架不一样,如果西方因为监管慢下来,中国就赢了。
马斯克的回答跟很多人的预期相反:
从我和中国官员的交流看,他们对AI也相当担心。事实上,他们可能比其他国家更有可能做出好的监管。(马斯克)
他还顺手划清了一条界限。特斯拉也在做AI,但那是很窄的一种,用于自动驾驶。那辆车不会哪天周日醒来就接管世界。
第二个风险,很多人根本没往那个方向想
提问者问,撇开AI,人类未来最大的威胁是什么。
马斯克的回答是:人口崩溃。他说这有点反直觉,因为大多数人觉得人已经太多了。他解释了这种错觉从哪来:你住在城市里,四周全是人,自然觉得拥挤。
如果你坐在飞机上往下看,扔一颗炮弹下去,多久能砸到一个人?基本永远砸不到。事实上每天都有东西从太空落下来,天然的陨石、旧的火箭残骸,一直都有。没人担心这个,因为人类在地球上的横截面实在是太小了。(马斯克)
他提到了Wait But Why的蒂姆·厄本(Tim Urban)算过的一笔账:地球上所有人,可以全部装进纽约市的一栋楼里,还只用一层。
所以风险的另一头站着出生率太低这件事。拥挤是一种视觉错觉,生育率是一条数学曲线。他说这算不上什么秘密,维基百科上就能查到各国的生育率数据,问题在于没人愿意认真看。
他对结局的描述相当不客气:
文明更可能是呜咽着结束,而不是轰然崩塌。那会是一个很悲伤的结局:平均年龄越来越高,年轻人实际上成了照顾老人的奴隶。这不是一个好的收场方式。(马斯克)
至于他自己有没有在做点什么,他的回答是:在生孩子这件事上,我试图做个好榜样。六个孩子。
提问者问他花多少时间陪孩子。他说:他们想跟我待多久,我就待多久。然后补了一段很真实的描述。一个还是婴儿,另外两个14岁和16岁,青少年通常不太想跟父母待在一起。感恩节周末孩子们都回来了,他问要不要多待一会儿,答案是不要。所以他判断,现在这个量大概正好。
Neuralink先救的是病人,健康人升级还早
聊到脑机接口,马斯克把预期压得很低。他说,短中期内,Neuralink真正要做的是帮助治疗脑损伤和脊柱损伤。
第一批植入人体的设备,是给四肢瘫痪的人用的,让他们只靠意念就能控制电脑或手机。
设想霍金能正常说话,用正常速度,甚至比正常速度更快。(马斯克)
这段话里没有未来学的宏大叙事,只有一个具体的人重新获得表达能力的场景。它不谈人类要进化成什么,只谈一个本来失语的人能重新把想法说出口。
访谈里还聊到医学的另一头。马斯克说,合成mRNA和DNA这块会有很多突破,因为它本质上就是一段程序。有了足够的技术和投入,阻止衰老、甚至逆转衰老,都不是太疯狂的想法。他的原话更夸张一点:只要DNA序列对了,你完全可以把一个人变成一只蝴蝶,就像毛毛虫那样。
这话听起来像玩笑,但背后的逻辑是认真的:生物在某种意义上是可以被编程的。
语言是人类最了不起的发明,写作是第二件
提问者问,过去一千年,人类最重要的发明是什么。
马斯克的答案是:语言。能说话,能表达概念,这是人类最大的发明。提问者说,这个答案出版行业很喜欢。马斯克接着补了一句:写作也了不起,古腾堡那家伙真的知道自己在干什么。
这里能看出他思考问题的一贯方式:把复杂的东西拆到最底层,然后看那个底层剩下什么。语言是把思想从一个脑袋搬到另一个脑袋的协议,写作是把协议固化成可以跨时间传输的格式。没有这两样,后面所有的技术都无从谈起。
提问者提到他和尼古拉·特斯拉的一个共同点:照相式记忆。这是天赋还是负担?
马斯克的回答有点出人意料地克制:有一部分确实如此,技术类的东西我记特别牢。然后立刻补了一句,但计算机的记忆比人好得多。他没有把这件事浪漫化。
techno、生日派对,和那五分钟的荣耀
访谈里穿插了不少私人话题,反而让整场对话松弛下来。
为什么techno对他这么重要?他的解释是,人有时候想体验到“最大程度的人”。
人们去狂欢,听着好音乐,那种感觉就像是,嘿,这是最大程度的人。你想真正地去感受它。和朋友一起玩,疯狂地跳舞,就是很好玩。(马斯克)
提问者顺势开玩笑:techno是不是你在柏林建厂的真正原因?马斯克答:老实说,这是个重要因素。
然后是生日。30岁那年在威尼斯办了一场化装舞会。40岁那场,技术上讲是“后末日化装舞会”,因为末日之后你还能剩几件衣服呢,总是有点破、有点烧焦的。
那场派对的主题是维多利亚时代的日式蒸汽朋克。派对上他和一位职业相扑选手摔跤,结果是颈椎间盘突出。他的总结是:五分钟的荣耀,换来五年的疼痛。
明年就50岁了,半个世纪。他的说法是:还没有计划,但通常都会想个疯一点的主题,得为这个“半个世纪派对”想点东西。毕竟在地球上待满五十年这件事本身,就够让人感慨一句:嘿,我还活着,真不错。
最后一个问题,生命的意义是什么。
两年前在一场晚宴上,他给的答案是42,《银河系漫游指南》里的那个梗。这次他补了后半段。
过一会儿,答案大概是那块好吃的法国奶酪。你要花点时间去欣赏生活里的东西和感受。食物很妙,有太多好的体验,其中一些根本不花钱。在大自然里走一走,或者好好吃一顿饭,就会觉得,哇,这真不错。我们应该花点时间欣赏这些小事,还有那些打动你的大事。这可能就是生命的意义,我能想到的最接近的定义。(马斯克)
整场访谈从火星到奶酪,跨度很大,逻辑却连得上。他做那些宏大工程的理由,最后落回到“让人有理由期待早上醒来”这件事上。城市、火箭、自动驾驶,都是为了让“活着”变得更值得。
这份访谈能拿来用的几件事
- 看技术拐点的时候,多问一句保有量拐点。全自动驾驶明年可能就实现,但路上那20亿辆车要换20年,机会和风险都藏在这段错位里。
- 衡量一个长期项目,试着换一把尺子。把计量单位从“赚了多少”挪到“把某件不可逆的事往前推了多少年”。这个标准能帮你分清真价值和短期噪音。
- 对“行业专家说这事不可能”保持警惕。2014年那位德国CEO的判断,六年后被5360亿对770亿的数字打了脸。专家的判断往往建在旧范式的成本结构上。
- 如果你在做新技术落地的事,把监管当成时间表里的一等变量,别放在事后补。欧盟每六个月开一次会的节奏,本身就是产品路线图的一部分。
- 关注AI风险,别只盯着“机器觉醒”。数据所有权、监管框架、各国的治理意愿,这些看起来琐碎的议题,才真正决定走向。
- 人口问题值得单独放进你的观察清单。它变化慢到几乎看不见,一旦形成趋势,所有关于增长、养老、创新的假设都要重算。
- 给自己留一点“法国奶酪时刻”。这话听着有心灵鸡汤的味道,说这话的人手上同时压着六家公司,一周工作一百小时。先把生活里那些不花钱的好东西认出来,你才有力气去做那些要花几十年的事。
0:00[Applause] well that was fun yeah i'm glad that you enjoyed it yeah it's like a ride i mean i think you could charge money for this this is great i mean yeah this really makes a difference to have this two screens and the angle change it's like that felt great like this you're right you know apart from this special trip to mars this evening when do you think realistically human beings will land on mars for the first time i think i i feel fairly confident about six years from now so every the the mars earth mod synchronization occurs roughly every 26 months so we had one this year this summer and so that means in roughly like about two years there'll be another one
0:48and then two years after that so i think i'd say if you say six years from now i think highly confident if we get lucky maybe four years and then we want to try to send a an uncrewed vehicle there in two years when will your first trip to orbit will take place i don't know possibly in two or three years i'm mostly concerned with developing the technology that can enable a lot of people to go to mars and make life multi-planetary have a base on the moon a city on mars
1:25and i think it's important that we strive to have a self-sustaining city on mars as soon as possible i mean i'm optimistic about the future on earth but it's important to have life insurance for life as a whole is it going to be a business kind of tourism in in in orbit or is it more kind of plan b if things on earth do not develop as well it's not exactly a plan b it's it's more that i think i think there's two two aspects to this one is that we want to have a future that is inspiring and exciting and what are the things that you find inspiring and exciting about the future i think one a future where we are a space-faring civilization and out there among the
2:07stars i think that's every kid gets excited about that you don't even need to teach them that they just get it's like instinctive and so we it's very important for us to have reasons to like reasons to be excited about life like when you wake up in the morning it can't just be about problems it's okay i know everyone in this room deals with a lot of tough problems but you know it's got to be more than that so you know i think a future where you can say hey even if it's not you there's going to be people out there they're going to we can have a base on the moon we're going to have a you know a city on mars maybe go further the moons of jupiter and
2:46everything i think that's a very exciting future and and then and i think most people do and you seriously want to be buried on mars just not on impact yeah i mean if you're gonna listen we're all gonna die someday so if you're gonna die someday i'm like okay don't be buried on mars i'm like mars sounds cool born on earth die on mars that's you know if you got the choice two years ago i had a conversation with jack ma and we spoke about jeff bezos plans with regard to orbit and he said well let's jeff bezos take care for the orbit i take care for the earth you seem to take care for both yeah basically tesla is about trying to make sure things are good
3:42for the future on earth and then spacex is about a good future beyond earth basically and so obviously we have to have sustainable energy both consumption and production of energy so like tesla does solar panels and batteries i think that's one of the key ways to have sustainable energy generation and also the batteries are useful for wind power so that and then you then you need to consume it by you consume electricity so electric vehicles and you know i think you look at these things like say okay if you look back from the future and say what's the fundamental good of tesla i would say it's probably should be assessed as by how many by how many years did tesla
4:27accelerate the advent of sustainable energy like that's like i would measure the goodness of tesla in that in that way and then for spacex it's like okay to what degree did we improve the probability of humanity being a space bearing civilization i remember very well the year 2014 when we were hosting the gold steering wheel here at oxford springer and you got the award for lifetime achievement and i was sitting in the first row with the then very successful and famous ceo of a very big german car company and i asked him while you were on stage isn't this guy dangerous for you i mean this looks really serious oh no don't worry first of all the whole
5:08idea of electric driving it's never going to be a mass market these guys these guys in silicon valley they have no clue about engineering about building really beautiful and great cars so we don't have to worry by then tesla's market cap was 23 billion today it's 536 billion us dollars
5:31the market cap of vw then was 86 and today it's 77 and you could you you are with tesla two and a half times bigger than bmw vw and daimler i even have you ever too high i mean what am i supposed to do you know like have you ever considered that too stock is too high a long time like when it was like a hundred eight hundred dollars pre pre-split and then don't listen to me but you know i tell you and the sec complained again i mean like i know
6:05is it a serious option to buy one of the incumbents one of the big car companies for you well i i think we're definitely not going to launch a hostile takeover so i suppose if there was a friendly one if somebody said hey we think it would be a good idea to merge with tesla we certainly have that conversation but you know we don't want to you know be a hostile takeover sort of situation did you feel a lot of complacency these days that the incumbents then let you feel that you are i mean the kind of hopeless disruptor but they know how to do it or where they're very polite and nice with you do you mean back then or not then oh no today everybody's super nice i would
6:46not say they were super no it was difficult to characterize their responses super nice they used a lot of adjectives i don't think that any of them were positive so we tried we really tried hard to convince a lot of companies honestly i was in so many panels but they're generally worth yeah generally the settlement that was expressed that you mentioned earlier that was pretty much universal especially if back in say 2008
7:25or 2007 like when we first unveiled the roadster in 2007 yeah i mean it was just basically [Music] they just said well you're basically a bunch of fools well i mean generally they say like well who starting a car company is crazy you're going to lose all your money i was like i think i probably will lose all my money i agree it wasn't like i thought it would be successful i thought we had maybe a 10 chance of success so then people would say you're it's going to fail and you're going to lose lose everything it's like yeah probably true yeah what else is new a couple of years ago we we we saw each other in america and the guy asks you on a panel when
8:07autonomous driving will be approved and you said i do not care so much when it's going to be approved i care more when human beings in cars will be forbidden and then the guy said well that's really unrealistic it's never going to happen in in cars people wanted to do something actively and then you said well 100 years ago nobody could imagine elevator without a lift boy today nobody could imagine a lift with a lift boy yeah so when is autonomous driving really really going to happen and when when are you able to do it and when is it going to be approved okay just between us yeah it's a very discreet circle here yeah so well first of all i'm not i'm not
8:53against people driving to be clear so i think people will drive cars basically as far into the future as i can imagine it's just that it's gonna be increasingly unusual to to drive your own car and while it's fun to drive you know a well-handling car on a winding road in beautiful terrain of course that's that's fun but it's not fun to drive a car in terrible gridlock traffic like you know going through extreme traffic that's no fun driving a car so i think people are unlikely to most of the time want to commute or with with their and drive themselves and you know people are typically spending hour and a half a day maybe two hours on average driving
9:42especially say like california or something like that it's very common and some people will actually commute like three hours a day sometimes it's pretty crazy so so i so i think i think if you say fast forward to [Music] like 10 years from now i think 10 years from now almost all cars will be will have full autonomy capability that newton all new costs produced so there's there's about two billion cars and trucks in the in the existing fleet and the new vehicle production is about five percent of the fleet size it's about 100 million so even the point which all cars are autonomous they'll still take you know 20 years to replace all the cars assuming that the number of
10:28cars and structure trucks in the fleet stays constant it would like say 10 years from now i'd say vast majority of cars electric like maybe 7 80 or more and almost all cause autonomous electric autonomy is absolutely the future no question it's just a question of when but then like i said sometimes people think that that means the the global fleet gets replaced instantly and it's like nope you have to go 20 years beyond that point before it 20 years from the point at which all cars are new cars are electric then the fleet will be replaced this is just an important it's not like some people are used to like mobile phones and that kind of thing is like
11:12two year or three year replacement rate but cars are a much more expensive asset longer life anyway to actually answer your question i'm i'm extremely confident of achieving full autonomy and releasing it to the tesla customer base next year now the that there's a uncertain period of time for when regulatory approval will be will take how long will it take but i think if you are able to accumulate billions of kilometers of autonomous driving then it's difficult to argue and look at the accident rate when the car is autonomous versus non-autonomous and in fact our statistics already show a massive difference when the car is on autopilot or not on autopilot
11:59the safety is much greater even with the current autopilot software and we are discussing level five autonomy so really full autonomy will europe lag behind or will it be approved here at the same time like in america or china it's hard to say exactly whether we approved i i mean just to and our customers already know this but the the the the eu regulators are the most conservative and i don't know if people want that to be the case or not our customers are sort of unhappy about it but yeah they only meet every six months
12:40maybe more often i don't know so yeah but i think at least some jurisdictions will allow full self-driving next year okay exactly a year ago you were announcing in this very building that you are planning to build a new site near berlin yeah and a couple of months later in june you started you want to finish it by july next year we did a little tour this morning it's impressive how advanced it is and it's almost unbelievable germany and picture particularly berlin is not world famous for finishing construction sites in time and in budget so you have created a kind of anti-berlin airport project why berlin why did you go to germany and to berlin
13:32to get that big project done sure well first of all i'm actually a big fan of germany i love germany it's great i've worked you know i have a lot of friends during friends and i think berlin is a very fun city and i think it's there's there's also from a location standpoint people like say young people can live in apartments at a reasonable price in the city of berlin but if somebody's got a family they can still have an affordable house so it's a good location offering you know good living for people of all ages and yeah it comes and and
14:22but it's definitely sexy could you imagine i mean so we're going to have like when we open the you're all invited by the way when we have the opening for gigabilin we're gonna have just a big party you know we're gonna have like start off from the day have more sort of family music and and then gradually get more hardcore and then go you know midnight techno till dawn do you plan to spend more time in berlin yourself you want to partly live here effect yes i will be spending a lot of time here where do you sleep tonight in the in the tonight's in the factory in a in the factory well technically a conference room in the factory but yeah you sleep in a
15:08conference room in the not finished factory tonight yeah it gives me a good feel for what's going on alone or yeah i presume so it's an invitation yeah okay elon you have so many projects it's not only tesla or spacex it's neuralink it's the boring company so many things and we discussed last time i asked you what is the most important project or the most important topic for you to deal with in the foreseeable future and you said that is truly the role that a.i is going to play in our society could you explain why and why that is a big opportunity but also seems to worry you yeah i think well i mean humans have been the smartest creature on earth for
16:02a long time and that is going to change with what's typically called artificial general intelligence so this is say an ai that is smarter than a human in every way it could even simulate a human so you know this is something we should be concerned about i think there should be government oversight of ai developments especially super advanced ai it's just this is anything that is a potential danger to the public we generally agree that this should have government oversight to ensure that the the public safety is taken care of because you feel that one day the mankind could serve the machines and not the other way around honestly when i see people on their
16:53phones i think we're already surveying the machine yeah it's like everyone's answering the questions you know every time you do a search or added information you're sort of building this the the the digital group mind but yeah the advent of artificial general intelligence is called the singularity for a reason because just like a black hole which is a singular singularity it's difficult to predict what will happen so it's not as though the advent of agi is necessarily bad but it's bad as one of the possible outcomes and what is singularity in the in the definition of ray quartz while going to happen
17:38well i think you're saying he he's predicting 2025 i think that's reasonably accurate and how can it be avoided that is then more a threat for humanity than an opportunity is it a question of governance so that there's not too much power in one or in few hands or how would you how would you make sure that it goes into the right direction i think we should have a a government oversight just like we do we have government oversight and regulation of cars and aircraft and food and pharmaceuticals these are all you know there's like there are regulators that oversee these developments to ensure public safety and i think
18:25digital super intelligence would also be potentially a public safety risk and so it should be it's i think it's very important to for regulators to keep an eye on that who should own the data by then i think everyone should own their own data like individuals who own their data and they certainly shouldn't be tricked by some terms and conditions of a website and suddenly you don't own your data that's crazy now who reads those terms and conditions anyway so but i think it's just you know like we wouldn't let people develop a nuclear bomb in the backyard just for the hell of it you know that that seems crazy so digital superintelligence i think has
19:07the potential to be more dangerous than a nuclear bomb so yeah we should just somebody should be keeping an eye so we can't have the inmates running the asylum here which is a global issue because if we do well but china has other rules and a different regulatory framework that is another yeah i don't think yeah i generally like that this is one of the rebuttals i get from those developing ai and tesla is also developing a form of ai for self-driving but it's a very narrow form of ai just like like the car is not going to wake up sunday one day and take over the world so so it's
19:51but the rebuttal i guess like well you know china is gonna have unfettered ai development and so if if we have regulations and that slows us down then china will have it and i'm like look i from my conversations with government officials in china they are they they're quite concerned about ai as well and they in fact they're probably more likely to have a good oversight than i think other countries what is the biggest challenge ahead of us in general not only with regard to ai what is the biggest problem that needs to be solved what's the biggest threat to humanity's future or something well ai is certainly one of the biggest risks it could be the
20:43biggest risk i think we need to watch out about population collapse this is somewhat counterintuitive to most people they think that well there's so many humans maybe too many humans but that's just because they live in a city if you're an aircraft and you look down they say if you dropped a a cannonball how often would you hit a person basically never in fact there's stuff falling in from space all the time natural meteorites old rocket stages all the time but nobody worries about it because the the actual in fact it there's a good cool website called wait but y and scott tim urban like he actually just did the math and and all humans on earth
21:33could fit in the city of new york on one floor don't even need the upper floors so that's actually the cross-section of of humans as seen from earth is extremely tiny basically vanishingly small almost nothing so we need to watch out about population collapse low birth rates i think is a big risk
21:58and it's also not exactly top secret you can go look at the wikipedia you know birth rate so and and this this is actually this this is this is definitely the civilization ends with the with a whimper not a bang because it would be a sad ending where the the average age becomes very high and really the youth are effectively a de facto enslaved to take care of the old people this is not a good way to go end do you have any new projects dealing with these topics that you've just addressed well i'm trying to set a good example on the kid front six kids yes
22:40for now how much time do you spend with them i spend about as much time as they want to spend with me so they yeah i mean they're not well one's just a baby and now there's a 14 and 16 and teenagers don't usually want to hang out with their parents that much you know you know we had thanksgiving weekend so all the kids were over so you know they
23:14they upset if they want to spend more time with me i said like oh you should i actually asked somebody we don't want to hang out more like no so i think it's probably the right amount then since they that's about the they don't want to hang out more so i think we really should take this seriously the population collapse artificial intelligence obviously sustainable energy is important the faster we transition to sustainable energy the less of a gamble we're taking with climate and i think there's going to be a lot of breakthroughs on the medical front particularly around the synthetic mrna you can basically do anything with the synthetic rna dna
24:02it's like a computer program so i mean i think with enough with with with effort that's not too crazy you could probably stop aging reverse it if you want these are you can basically join you can turn someone into a freaking butterfly if you want with the right dna sequence so when caterpillars do it so your project neuralink is in a way empowering human intelligence versus artificial intelligence that's the purpose of it is that correct yeah so neurolink the in the short to medium term neural link is really just going to help cure brain injuries and rain and spine injuries so it's like if somebody is in fact our first implanted devices in humans will be for
24:52quadriplegic tetraplegics allowing them to control a computer or a phone just using their mind so like you can imagine like if stephen hawking could just talk and at normal speed or even faster than normal speed looking back for the last thousands of years what is the most important invention of mankind so far in the past thousand years i guess it's millions well i think language being able to talk and express concepts and this this is probably the biggest invention of humanity's language it's an answer that we like very much in the publishing yes absolutely no writing is exactly just incredible right writing really made a big difference good that guy gutenberg you
25:43really know what he's doing you have one thing in common with nicolas tesla that's a photographic memory is that only a gift or sometimes a burden because you're memorized too much i have a photographic memory in some respects for technical stuff i have a very good memory so for a human yeah you know computers are much better at memory computers are really good memory why is music so important for you techno music in particular well it's pretty fun i think it's you know you want to i don't know feel maximum human you know and so i think when people have like sort of a rave and good music it can be like hey maximum human you know you want to really feel
26:36you know it's like but like what where it really gets you to feel you know and i think that you know having fun with friends and you know just crazy dancing is fun perhaps you know techno music is the secret reason why you are building big projects in berlin yeah honestly that's a it's a significant factor okay elon last question you you celebrated your 30th birthday with a masked ball in venice for your 40th birthday i was told you had a fight with a samurai sword fighter what is your plan for your 15th birthday next year well so my my 40th birthday was the was in venice it was it was technically a post-apocalyptic masked wall because you know after the apocalypse
27:24how much clothing do you really have you know it's not going to be a little ragged a little burnt you know so no plans for the 50th yeah 50th the half century party i have to think of something usually go with some kind of crazy theme the the the the party where i ended up wrestling with world champion tumor wrestling which by the way also caused me to burst a disc in my neck so you're five minutes of glory for five years of pain was this that really hurt so that party was victorian japanese steampunk so that was cool i have to think of something for the half century party you have a little time to think being on earth for a half century that's
28:24like hey i'm still alive wow cool okay one very last question when i asked you what is the meaning of life during a dinner 42 you said after a while after a while well probably this wonderful french cheese could you please explain well i was just saying that you know you want to take a moment to appreciate things in life and the sensations food's incredible and like there's so many good things that you can experience some of them cost nothing really you know have a walk in nature or
29:05just a nice meal and it's like wow that's pretty great you know and we should take a moment to appreciate these these little things the big things the things that move your heart i think that's probably the meaning of life most close definitions as i can think thank you very much elon [Applause]
资源与出处
- 视频在 YouTube 收看
- 档案silicon-valley.tech 档案页本页视频与文稿数据来自 Elon Musk Archive
- 精读本篇精读 Markdown 版