AI Day 2022 的答问环节进行到一半,马斯克说了一句不太像发布会结尾的话。有人问 Optimus 的手为什么用腱驱动,特斯拉工程师讲完线缆强度、零件数量和抗背隙之后,马斯克补了一句:
我们大概没有在所有的技术方案上都找对了树。这些方案没有刻在石头上。(马斯克)
这句话定下了整场答问的基调。台上刚花两个小时展示了 Dojo 芯片、占用网络、人形机器人,台下观众问的全是硬问题:腱会不会断,神经网络怎么做单元测试,Dojo 到底怎么赚钱,AGI 要不要监管。马斯克的回答比演讲时更放松,也更坦率。
这场答问透出的信息量,其实比主舞台的 PPT 更大。他把 Optimus 的量产时间说成 3 到 5 年,把 FSD 全球 beta 说成年底前技术上可行,把 Dojo 的商业模式定成 AWS 而不是英伟达,还顺手批评了几家硅谷公司是"蜂蜜陷阱"。下面按主题拆开讲。
手为什么用腱驱动,而不是一堆小连杆
第一个问题就直奔机器人最难的部位。观众问:腱不耐用,为什么 Optimus 的手要用腱驱动,还要加弹簧?
特斯拉工程师的回答分三层。先是材料,团队一开始试过合成腱,发现金属编织线缆强度更高。再是制造,腱驱动能大幅减少零件数量,而特斯拉的目标是造很多只手,一堆小连杆在量产时会变成噩梦。最后是精度,腱方案可以做到抗背隙(anti-backlash),手指动作不会出现间隙和卡顿。
弹簧的作用更直接:让手指能被动张开。如果用两个执行器分别负责握紧和张开,成本和重量都翻倍。现在一根腱拉合,弹簧负责伸展。人手也是这个逻辑,主动弯曲,被动伸展。
这套解释背后有一条贯穿全场的原则。马斯克反复说,Optimus 的目标是最快路径到一个能批量生产、真正有用的机器人。所有技术选择都服从这个目标,不追求一步到位的完美方案。他自己承认手会有第二版、第三版,架构可能大幅改动。
先在自家工厂搬零件,用现实闭环验证有用性
有人问 Optimus 什么时候能进新工厂、公众什么时候能买到。马斯克的回答把节奏说得很细。
起步阶段只做最简单的测试:把一个零件从 A 搬到 B,或者把零件装进现有的焊接机器人单元。目的很朴素,搞清楚怎么让它哪怕有一点点用。他提到一个词,close the loop on reality,必须在现实里闭环,才能确认机器人真的有用。
然后是有用场景的数量会指数增长。这句话听起来像画饼,但前面那句"先搬零件"给了它锚点。特斯拉的做法是把机器人扔进自己的工厂,用真实产线当试验场,而不是在实验室里堆功能。
关于交付时间,马斯克给了一个罕见的明确区间:
我觉得不会太远。大概 3 年内,最多不超过 5 年,你就能收到一台 Optimus。(马斯克)
3 到 5 年,这是全场唯一一个带数字的产品承诺。对照他过去对 FSD 时间的判断,这个数字要打个折扣看,但它至少说明了内部对量产节奏的预期量级。
使命从"加速可持续能源"扩到"让未来变得精彩"
有位观众问得很准:特斯拉的使命一直是加速可持续能源的到来,Optimus 跟这个使命对不上,公司会不会改使命?
马斯克承认得很干脆,Optimus 严格来说不属于加速可持续能源。它提高做事效率,间接有助于能源转型,但关系是绕的。所以使命实际上已经扩大了,他用了一个很马斯克的表述:making the future awesome,让未来变得精彩。
接着他给了一个判断技术价值的方法:看这项技术五年、十年后还会不会有让人想看的变化。他说很多技术已经平台化了,停滞了,就不点名了。Optimus 属于另一类,五年十年后会"mind-blowing"。
这个判断标准值得记下来。它不问这项技术现在有多强,只问它的曲线还有多陡。
别走终结者路线:本地急停和不能联网更新的 ROM
有观众问 Optimus 会不会有对话能力、最终目标是什么。马斯克的回答先给了肯定,然后拐到了安全上。
他说 Optimus 肯定会能对话,感觉会很自然。最终目标会持续演化,落在某个有意思的地方。然后他提了一件事:本来想在开场放一段终结者的视频,骷髅被压碎那种,后来觉得观众可能会当真,算了。
玩笑之后是实打实的设计约束。特斯拉在 Optimus 里内置了几道保险:
- 本地急停,人可以在现场直接让机器人停下来
- 本地控制 ROM,不能通过互联网更新
- 遥控介入能力,训练阶段尤其需要
马斯克说这条"不能联网更新"是本质性的,不是可选项。逻辑很清楚,如果机器人的底层控制逻辑能被远程改写,那所有上层安全设计都是纸糊的。这个思路在后续回答开源问题时又出现了一次,他说给 Optimus 的指令要受某种类似机器人定律的规则约束,且用户无法绕过。
关于训练阶段的干预数据,工程师补充了两点。一是设计成遇到碰撞就停,人手按住机器人它就会停,不会把手压碎。二是仿真系统里可以检测碰撞,把危险动作标出来。这些干预记录本身就是训练数据。
Dojo 不卖芯片,要做神经网络训练的 AWS
一位观众问得很商业:Dojo 用了 7 纳米工艺,研发成本轻松超过一千万美元,这个生意怎么做,是卖芯片像英伟达,还是提供基础设施像 AWS?
马斯克的答案很明确,Dojo 是一台非常大的计算机,耗电量大,散热需求高,把它卖给别人的效率不如自己运营。
Dojo 更合理的形态是像亚马逊云服务那样运营,而不是卖给别家。最有效的方式就是让它成为一个在线可用的服务。(马斯克)
他的推理链条是:世界正在转向软件 2.0,神经网络训练的量会越来越大,那么训练方自然会想用最快、最便宜的训练系统。需求本身在膨胀,这块市场只会越做越大,存量竞争的顾虑可以放下。
这里有个细节:特斯拉自己就是 Dojo 最大的客户,自动驾驶训练需要的算力是个无底洞。先自用,把成本和稳定性跑通,再对外提供服务,这条路径和 AWS 当年从亚马逊内部孵化出来几乎一样。
车道预测为什么要借用语言模型
有人注意到 Autopilot 团队在车道建模上用了语言模型的思路,问这是怎么来的、提升有多大。这个问题触及了这套系统的架构核心。
答案是两个原因。
第一个是表达能力。以前用稠密三维方式预测车道,只能建模特定类型的车道。交叉路口内部那些纵横交错、互相连接的线,用稠密分割根本表示不出来,必须换成图预测。换成图预测是硬约束,没有可选项。
第二个是车道预测本身是多模态问题。在路口这头看过去,视觉信息不足以确定对面车道的确切走向。这时候需要的是一个能生成连贯结果的模型,让它承诺一种可能,而不是同时输出两条车道和三条车道两种答案。语言模型恰好提供了这种"生成并选定"的能力。
这个类比挺说明问题。让模型画车道线,本质上和让它写完一段话是一回事:给定不完整的前文,输出一个自洽的后文。
神经网络的单元测试:几十万个历史失败案例
一位观众问了个工程味道很重的问题:神经网络的单元测试怎么做?训练完了,怎么判断它能发布?
特斯拉的回答是分层过滤。
最底层是软件本身的单元测试,跟普通软件开发一样。往上是为神经网络模型定义的 VIP 测试集,按失败模式分类。团队发现一件事,光有一个大的测试集不够用,必须针对不同失效类型建专门的集合,然后随着产品迭代不断扩充。
几年下来,这套集合里积累了几十万个曾经失败的例子。任何一个新模型发布前,都要跑一遍全部历史失败案例。这个机制的价值在于防止回归,模型在修好新问题的同时,不能把三年前修过的老问题再犯一遍。
再往上是影子模式,模型静默推送到车主车上,不参与控制,只回传它在哪些场景失败、哪些场景成功。最后是 QA 流程,马斯克说模型到用户手里之前要过九层过滤。
他还自曝了一个身份:
我自己就是 QA 测试员之一。我经常在车里测试最新的 alpha 版本,只要它不完全崩溃。能找出很多 bug。(马斯克)
这段回答里最有用的是那个思路:把历史失败案例变成一个只增不减的测试集。模型会遗忘,测试集不会。
教师模型、学生模型,和 PB 级的真实世界数据
有观众问基础模型(foundation model)对自动驾驶的意义,是不是应该先训一个巨大的教师模型,再蒸馏成学生模型装进车里。
特斯拉的做法基本就是这样,只是他们不太用这个词。车上跑的模型受算力和延迟限制,但服务器上没有这个限制。团队会训练一批完全离线、体积巨大的模型,用它们给数据打标签,再用这些高质量标签去训练能实时运行的车端网络。这是最标准的教师学生蒸馏。
关于基础模型本身,特斯拉建了几个 PB 级的数据集,某些任务在这个规模下表现非常好。他举的例子是运动学:从视频里推出每个物体的位置、速度、加速度,一直到四阶导数。
这个例子的分量在于精度要求。一阶导数是速度,二阶是加速度,再往上每求一次导,噪声都会被放大一次。能算出四阶导数还保持准确,说明底层的位置估计精度极高。很多人认为摄像头做不了可靠的深度检测,这个结果是对那个判断的反驳。
约翰补充了一句更朴素的观察:每次用更大的数据集训练,性能都会明显提升;每次用其他辅助任务做预训练初始化,性能也会提升。自监督和大数据集监督都有效。这话没有花哨的地方,但它是整个团队的技术信念。
AGI 会是副产品,而特斯拉可能握着最大的数据集
有人问特斯拉怎么定义 AGI、具体在攻哪个方向。马斯克的回答先划清界限:特斯拉没有专门做 AGI。他的判断是,AGI 可能是他们正在做的事情的涌现属性。
推理是这样的。AGI 需要的要素里,可获取的数据集极其重要。互联网数据能被扫描,但互联网给不了一样东西:几百万个摄像头在真实世界里同时采集的视频和音频。地球很大,现实很乱,这些数据只能靠车和机器人跑出来。如果路上有几千万辆自动驾驶车,再加上数量相当甚至更多的人形机器人,那就是人类历史上最大的真实世界数据集。
他顺带说了一句训练算力可能也是最多的。数据最多,算力最多,那么"特斯拉会对 AGI 有所贡献"这个结论就自然成立。
话题延伸到人类集体智能。马斯克说人类整体本来就算一种超级智能,互联网的出现相当于给人类装上了神经系统。一千年前书极少、极贵,识字的人很少,能拥有一本书的人更少。现在任何一本书都能瞬间拿到,任何东西几乎都能免费学。他认为互联网是历史上最大的信息平等器。
有人问他最想生活在哪个历史时期,答案就是现在。他说自己读了很多历史,这是历史上最有意思的时刻。
关于监管,他的立场很清楚:AI 应该在政府层面设一个监管机构,就像航空、汽车、食品、药品都有监管一样,因为这些领域影响公共安全,AI 也影响公共安全。他补了一句,政府目前还没理解这件事。至于特斯拉自己要不要建 AGI 安全方面的技术团队,他的回答是有条件的:如果看起来特斯拉要对 AGI 做出重大贡献,那一定会投入安全研究。
FSD:年底全球 beta,核心指标只有一个
关于 FSD 的那组问题最密集。观众问了三件事:什么时候推到美国和加拿大以外,当前最大的技术瓶颈是什么,高速和城市合并成单一栈的进展如何。
马斯克的回答分了几个时间点。技术层面,FSD beta 年底前具备全球推送的能力,但很多国家要看监管审批,卡在批文上。下个月会有一个大版本,特别擅长判断快速横穿车流的物体速度,还有一批其他改进。
关于技术栈合并,工程师给了背景。以前量产版 Autopilot 和 FSD beta 差异很大,几个月前开始,两者在所有车辆上共用同一套纯视觉目标检测栈。剩下的主要差别在车道建模,量产版还在用简化模型。现在 FSD beta 的车道模型正在扩展到高速场景。
马斯克自己开的就是合并后的版本,城市道路和高速都用 FSD 栈,他的评价是"开得相当好"。还需要验证大雨、大雪、沙尘这些天气,确认在各种环境下都优于量产栈。时间上他说"肯定在今年之内,可能十一月"。
泊车场景也会在年底前并入 FSD 栈。目标是从停车场入口上车,一直开到车位,全程一个栈。
至于衡量标准,他给了一个数字:
最根本的指标是两次必要干预之间能跑多少英里,尤其是安全相关的干预。我们每周都在测这个数,进步非常大。(马斯克)
关于半挂卡车 Semi 的传感方案,他的回答回到了第一性原理。人开车靠的是什么?一个生物神经网络,加两个装在极慢云台上的摄像头,那个云台就是脖子。既然这样的配置能开卡车,那么八个摄像头、连续 360 度视野、更高帧率、更快反应速度,开得比人好是显然的。
给 20 多岁的自己:偶尔停下来闻闻玫瑰
一位观众问了个非技术问题:如果能回到 20 多岁,会给自己什么建议。
马斯克想了半天,说想找点有用的话讲。然后给了几条:多接触聪明人,多读书(他说这两条自己当年倒是做了),还有一条是别绷得太紧,多享受当下。
接着讲了个具体的故事。当年在夸贾林环礁开发猎鹰一号火箭,那是个很漂亮的小岛,整个开发周期里他一次都没在海滩上喝过一杯。他说:
我应该在海滩上喝一杯的。那没什么问题。(马斯克)
这个细节比任何励志金句都实在。一个以极端工作强度著称的人,回头看觉得自己错过了什么,说出来的不是"应该更努力",是"应该松一点"。
硅谷的"蜂蜜陷阱",和特斯拉的反面
有人问,特斯拉凭什么能同时做电池、电机、电控、芯片、算法、机器人这么多领域。马斯克的回答分两层。
第一层是历史。特斯拉本质上是一连串创业公司叠起来的,为了造电动车必须造电池和电机,为了造自动驾驶车必须造芯片和神经网络训练系统。每一步的能力都是被逼出来的,公司从没画过一张覆盖这些领域的版图。这一串创业公司到现在几乎都成功了。
第二层是他对自己职责的定义:创造一个让优秀工程师能施展的环境。然后话锋一转,开始批评硅谷的一类公司。
他说很多公司里,真正有才华、有驱动力的工程师,才能是被压制的。这种压制不一定看起来像坏事,恰恰相反,薪水高、环境舒服、要求低,人的产出极低但过得很愉快。他给这类地方起了个名字,honey trap,蜂蜜陷阱。
硅谷有几家蜂蜜陷阱公司。看起来对工程师不坏,但你得问一句:一个很好的工程师进去了,出来的是什么?产出似乎很低。(马斯克)
他的结论是特斯拉不属于蜂蜜陷阱,要求高,会让你做成很多事,过程不轻松。SpaceX 也一样。这段话既是对应聘者的喊话,也透露了他判断组织效率的标准:看输入的天赋和输出的成果之间的比值。
占用网络:一年前还不存在的东西
一位研究者问,车道检测已经从实例分割换成了 Transformer 建图,还有哪些硬骨头,学术界能帮上什么忙。
特斯拉工程师拿占用网络(Occupancy Network)举例。这个东西一年前不存在,团队只用一年时间就把它做出来,一年里迭代了十几个版本。
他描述了一年前的困境。那时候车在二维世界里开车,一堵墙和一条路缘石用同样的静态边缘表示。可开车时面对墙和面对路缘石,做的决定完全不同,一个要停,一个可以压过去。意识到必须上三维之后,整个问题要重新想一遍。
这个例子说明的是自动驾驶的难点所在:很多错误的原因不在模型不够大,在世界表示的方式本身不够。换表示方法,等于重做一遍系统。
关于数据引擎怎么找到特定场景,比如那辆该停但容易漏掉的车,工程师给了两个触发机制。一是信号内部出现分歧就回传,比如某个物体的"停放"状态在停放和行驶之间闪烁。二是利用影子模式,车主忽略了某个物体但系统认为应该停,这个分歧也回传。靠这些触发逻辑,定向收集数据。
控制带宽没有标准答案,八条手臂以后再说
最后一个技术问题问的是 Optimus 现在和理想的控制器带宽。工程师的回答是不给数字:先看你让它做什么任务,对这个任务做频率变换,需要肢体做什么动作,带宽从任务里推出来,不能凭空指定一个数。
马斯克补了个大致范围。肯定不是 1 赫兹,也用不着做到 100 赫兹,大概 10 到 25 赫兹这个量级,往后会继续提高。翻译成用户能感知的词,就是灵巧度和延迟,方向是延迟越低越好、灵巧度越高越好。
最后一位观众问了个很有意思的问题:为什么 Optimus 必须左右对称?人有惯用手,长期使用会有磨损,某些关节和执行器会更早失效。人类还在幻想里给自己加了各种超能力,为什么机器人不能做得更自由,比如八条手臂。
马斯克的回答先开了个玩笑,说把神探加杰特造出来挺酷的。然后回到现实:现在只想做一个基础人形机器人,把它做好。
他接着讲了一段关于"有用"的话,是整场答问里最值得记的部分。真正难的事情是有用。要把效用算出来:帮了多少人,平均每个人帮了多少,总效用是多少。能做出大量人喜欢、真正有用的产品,难到超出想象。发货的公司和没发货的公司之间是天壤之别。发完货之后,还要让输出价值高于输入成本,硬件领域尤其难。
至于八条手臂、附件生态、给 Optimus 加电源接口像给手机加配件一样,他说这些以后都可能,甚至会有一批公司专门做 Optimus 的扩展件。但要先把基础的那台做出来。
回到开头那句"我们可能没有找对所有的树"。整场答问里,马斯克一共承认了三次不确定:手的设计会改、AGI 的终点不知道在哪、Optimus 的具体形态会演化。这些坦白和那些具体数字放在一起,构成了这场发布最可信的部分。有明确的时间表,也有明确的无知边界。
对读者来说,这场答问里有几件事可以直接拿去用:
- 判断一项技术值不值得投入,看它五年、十年后还有没有变化空间,平台化的领域不要再挤
- 评估一家公司对工程师好不好,别只看薪水和舒适度,看输入的天赋和输出的成果之间的比值,警惕蜂蜜陷阱
- 做机器学习系统,把历史失败案例建成只增不减的回归测试集,新模型发布前全量跑一遍
- 遇到表达能力不够的问题,先怀疑表示方法,别急着加参数;二维表示分不清墙和路缘石,换三维才能解决
- 任何涉及物理世界安全的系统,底层控制逻辑要做成本地、不可远程更新的形式,这是其他安全设计的前提
- 衡量自动驾驶这类系统的进展,盯住一个可量化的核心指标(比如两次必要干预之间的里程),每周测,别用演示视频代替
- 做硬件产品,先算清楚输出价值能不能覆盖输入成本,再谈生态和扩展件
0:00All [Applause] right. All right. Let me know. All right. So, we hopefully that was enough detail. and now we can move to questions. and guys like I think the whole the team come out come out on stage and we really wanted to show the the depth and breadth of Tesla in artificial intelligence comput hardware robotics
0:47actuators and and and try to really shift the perception of the company away from you know a lot of people think we're like just a car company or we make cool cars, whatever. But they don't have most people have no idea that Tesla is arguably the the leader in real world AI hardware and software. and that we're building what is arguably the first the most radical computer architecture since the the Crayon supercomputer.
1:23And I think if you're interested in developing some some of the most advanced technology in the world that's going to really affect the world in a positive way, Tesla's the place to be. So yeah, let's fire away with some questions. I think there's there's a mic at the front and a mic at the back.
1:46Chris on this side. Just just throw mics at people. Jump all for the mic. Yeah. Hi. Thank you very much. I I was impressed here. Yeah. I was impressed very much by Optimus. But I wonder why tendon driven the hand. Why did we choose a tendon-driven approach for the hand? Because tendons are not very durable and why springloaded.
2:15Yeah. Hello. Is this cool? Awesome. Yes, that's a great question. you know, when it comes to any type of actuation scheme, there's trade-offs between, you know, whether or not it's a tendon-driven system or some type of linkage based system. Just keep the mic close to your mouth.
2:30A little bit closer. Yeah. Hear me? Cool. Yeah. so yeah, the main reason why we went for a tendon-based system is that, you know, first we actually investigated some synthetic tendons, but we found that metallic boating cables are, you know, a lot stronger. one of the advantages of these cables, is that it's very good for part reduction.
2:52we do want to make a lot of these hands. So, having a bunch of parts, a bunch of small linkages ends up being, you know, a problem when you're making a lot of something. one of the big reasons that, you know, tendons are are better than linkages in a sense is that you can be anti-backlash. so anti-backlash essentially, you know, allows you to not have any gaps or, you know, stuttery mo motion in your fingers.
3:17spring-loaded, mainly what spring-loaded allows us to do is is allows us to have active opening. so instead of having to have two actuators to drive the fingers closed and then open, we have the ability to, you know, have the tendon drive them closed and then the springs passively extend. And this is something that's seen in our hands as well, right? We have the ability to actively flex and then we also have the ability to extend.
3:42yeah, I mean our goal with Optimus is to have a a robot that is maximally useful as quickly as possible. So there's there's a lot of ways to solve the various problems of of a humanoid robot. and I we're probably not barking up the right tree on on all the technical solutions. And I should say that we're we're open to evolving the technical solutions that you see here over time.
4:06We're not they're not locked in stone. but we do we have to pick something in and we want to pick pick something that's going to allow us to produce the robot as quickly as possible and have it like I said be useful as quickly as possible. We're trying to follow the the goal of fastest path to a useful robot that can be made at volume. and we're going to test the robot internally at Tesla in in our factory and and
4:33just see like how useful is it because you have to have a you got to close the loop on reality to confirm that the robot is in fact useful and yeah so we're just going to use it to build things and we're confident we can do that with the hand that we have currently designed but there I'm for sure there'll be hand version two version three and we may change the architecture quite significantly over time.
5:02Sorry. Hi. your the Optimus robot is really impressive. You did a great job. bipedal robots are really difficult. but what I noticed might be missing from your plan is to acknowledge the utility of the human spirit. And I'm wondering if Optimus will ever get a personality and be able to laugh at our jokes while they while it folds our clothes.
5:29Yeah, absolutely. I think we want to have really fun versions of Optimus. and so that optimist can both do be utilitarian and do tasks, but can also be kind of like a friend and a buddy and and hang out with you. And I'm sure people will think of all sorts of creative uses for this robot. and
5:56you know the thing on once you have the core intelligence and actuators figured out then you can actually you know put all sorts of costumes I guess on on the robot. I mean you can make the robot look you can skin the robot in many different ways. and I'm sure people will find very interesting ways to to yeah versions of Optimus.
6:26So thanks for the great presentation. I wanted to know if there was an equivalent to interventions in Optimus. It seems like labeling through moments where humans disagree with what's going on is important. And in a humanoid robot that might be also a desirable source of information.
6:52That's worth saving. yeah, I I think we we'll have ways to remote operate the robot and intervene when it does something bad. especially when we are training the robot and bringing it up. and hopefully we, you know, design it in a way that we can stop the robot from, if it's going to hit something, we can just like hold it and it'll stop. It won't like, you know, crush your hand or something. And those are all intervention data.
7:18yeah and we can learn a lot from our simulation systems too where we can check for collisions and supervise that those are bad actions. Yeah. Yeah. I mean, so Optimus, we want over time to for it to be you know, an android, the kind of Android that you'd seen in in sci-fi movies like Star Trek, the next generation, like data, but obviously we could program the the robot to be less robot-like and more friendly and and you know, it can obviously learn to emulate humans and and feel very natural. So as as AI in general improves, we can add that to the robot and you know it should be obviously able to do simple instructions or even intuitit what it
8:03is that you want. so you could give it a high level instruction and then it can break that down into a series of actions and and take those actions. Hi. yeah, it's exciting to think that with the Optimus, you will think that you can achieve orders of magnitude of improvement in economic output.
8:27that's really exciting. and when Tesla started, the mission was to accelerate the advent of renewable energy or sustainable transport. so with the optimist do you still see that mission being the mission statement of Tesla or is it going to be updated with you know mission to accelerate the advent of I don't know infinite abundance or limit limitless economy.
8:54Yeah, it I mean it is not strictly speaking Optimus is not strictly speaking directly in line with accelerating sustainable energy. It it you know to the degree that it is more efficient at getting things done than than a person. it it does I guess help with you know sustainable energy but it I think the mission effectively does does somewhat broaden with the advent of Optimus to u you know I don't know making the future awesome so you know I think you look at Optimus and I don't know about you but like I'm excited to see what Optimus will become and you know this is like you know if if you could I mean we can tell like any given technology If are you do you want to see what it's
9:44like in a year, two years, three years, four years, 5 years, 10? I'd say for sure you definitely want to see what what's happened with Optimus. whereas, you know, a bunch of other technologies are, you know, sort of plateaued. won't name names here but you know so I think Optimus is going to be incredible in like five years 10 years like mind-blowing and I'm really interested to see that happen and I hope you are too.
10:19Oh thank I have a quick question here. I'm Justin and I was wondering like are you planning to extend like conversational capabilities for the robot? And my second follow-up question to that is what's like the end goal? What's the end goal with Optimus? yeah, Optimus would definitely have conversational capabilities. so
10:46like you'd be able to talk to it and have a conversation and it would feel quite natural. So, from an end goal standpoint, I'm I'm think it's going to keep keep evolving and I'm not sure where where where it ends up, but some someplace interesting for sure. you know, we always have to be careful about the, you know, don't go down the Terminator path. that's a you know I I thought for might maybe we should start off with a video of like the Terminator starting off with this you know skull crushing but that might be you know people might take that too seriously. So you know we we we do want Optimus to be safe. So we are designing in safeguards where you can locally stop
11:33the robot and you know with with like basically a a localized control ROM that you can't update over the internet which I think that's quite important essential frankly so like a localized stop button remote or remote control something like that that that cannot be changed. but I mean it's definitely going to be interesting. It won't be boring.
12:08So okay. Yeah. I see you today you have very attractive product of dojo and its applications. So I'm wondering what's the future for the dojo platform? we like provide like infra infrastructure and service like AWS or you will like sell the chip like the Nvidia. So basically what what's the future because the I I say you use 7 nanometer so the develop cost like easily over $10 million US dollars how do you make the business like business wise yeah I mean dojo is is a is a very big computer and actually will be use a lot of power and need a lot of cooling so I think it's probably going to make more sense to have dojo operate in like a Amazon web services manner than to try
12:58to sell it to someone else. so the the most that would be the most efficient way to operate Dojo is just have it be a a a service that you can use that's available online and that where you can train your models way faster and for less money and as the world transitions to software 2.0
13:25and that's on the bingo card. someone I know has to now drink five tequilas. so let's see software 2.0 will use a lot of neural net training. So you know it kind of makes sense that over time as there's more more neural net stuff the people will want to use an an the fastest lowest cost neural net training system.
13:58So I think there's a lot of opportunity in that direction. Hi my name is Ali Jahan. Thank you for this event. It's very inspirational. My question is I'm wondering what is your vision for human robots that understand our emotions and art and can contribute to our creativity. Well, I think there's there's you're already seeing robots that at least are able to generate very interesting art with like like Deli and Deli 2.
14:44and I think we'll we'll start seeing AI that can actually generate even movies that have a that have coherence like interesting movies and tell jokes. So, it's it's quite remarkable how fast AI is advancing. at at many companies besides Tesla. We're headed for a very interesting future.
15:08And yeah, so any guys want to comment on that? Yeah, I guess the Optimus robot can come up with physical art, not just digital art. You can, you know, you can ask for some dance moves in text or voice and then you can like produce those in the future. So it's a lot lot of like physical art, not just digital art.
15:28Yeah. Oh yeah. Yeah, computers can absolutely make physical art. Yeah, 100%. Yeah, like dance, play soccer or what have you. I mean, it needs to get more agile, but over time, for sure. Thanks so much for the presentation for the Tesla autopilot slides. I noticed that the models that you were using were heavily motivated by language models and I was wondering what the history of that was and how much of an improvement it gave. I thought that that was a really interesting curious choice to use language models for the lane transitioning.
16:01So there are sort of two aspects for why we transition to language modeling. So the first talk talk loud and close. Okay, it's not coming through very clearly. Okay, got it. Yeah. So the language models help us in two ways. The first way is that it lets us predict lanes that we couldn't have otherwise. As a shook mentioned earlier, basically when we predicted lanes in sort of a dense 3D fashion, you can only model certain kinds of lanes, but we want to get those criss-crossing connections inside of intersections.
16:29It's just not possible to do that without making it a graph prediction. If you try to do this with dense segmentation, it just doesn't work. also, the lane prediction is a multimodal problem. Sometimes you just don't have sufficient visual information to know precisely how things look on the other side of the intersection. So you need a method that can generalize and produce you know coherent predictions. you don't want to be predicting two lanes and three lanes at the same time. You want to commit to one and a generative model like these language models provides that.
17:01Hi oh hi my name is Giovanni. yeah thanks for the presentation. It's really nice. I have a question for FSD team. So for the neural networks, how do you test like how do you do unit tests, software unit tests on that? Like do you have like a a bunch or I don't know thousands or yes cases where the neuronet network that after you train it, you have to pass it before you release it to as a product, right? yeah, what's your software unit testing strategies for this basically? Yeah, glad you asked. There's like a series of tests that we have defined starting from you know unit test for the software itself but then for the neural network models we have VIP sets defined where
17:51you know you can define if you just have a large test set that's not enough what we find we need like sophisticated VIP sets for different failure modes and then we curate them and grow them over the time of the product. Over the years we have like like hundreds of thousands of examples where we have been failing in the past that we have curated and so we for any new model we test against the entire history of these failures and then keep adding to this test set. On top of this we have shadow modes where we ship these models in silent to the car and we get data back on where they are failing or succeeding. and there's extensive QA program. It's very hard to ship a
18:31regression. There's like nine levels of filters before it hits customers. But then we have really good infra to make this all efficient. And I'm one of the QA testers. So I I QA the car QA tester. Yeah. I I So I'm constantly in the car just being QAing like whatever the latest alpha build is that doesn't totally crash.
18:55Finds a lot of bugs. Yeah. hi. great event. I have a question about foundational models for autonomous driving. I have als seen that big models that really can when you scale up with data and model parameter right from GP3 to palm it can actually now do reasoning. Do you see that is essential scaling up foundational models with data and size and then at least you can get a teacher model right that potentially can solve all the problems and then you distill to a student model is is that how you see foundational models relevant for autonomous driving that's quite similar to our auto labeling model so we we don't just have models that run in the car we train
19:42models that are entirely offline that are like extremely large that can't run in real time on the car. so we just run those offline on the servers producing really good labels that can then train the online networks. so that's one form of distillation of these teacher student models. in terms of foundation models we are building some really really large data sets that you know are multiple pabytes and we are seeing that some of these tasks work really well when we have these large data sets like the kinematics like I mentioned video in all the kinematics out of all the objects and up to the fourth derivative and people thought we couldn't do detection with cameras I
20:20mean detection depth velocity acceleration and imagine how precise these have to be for these higher order derivatives to be accurate and these all comes from these kind of large data data sets and large models. So we're seeing the equivalent of foundation models in our own way for geometry and kinematics and things like those.
20:40You want to add anything John? Yeah, I'll keep it brief. Basically, whenever we train on a larger data set, we see big Okay. Basically, whenever we train on a larger data set, we see big improvements in our model performance. And basically whenever we initialize our networks with you know some pre-training step from some other auxiliary task we basically see improvements the self-supervised or supervised with large data sets both help a lot.
21:08Hi so at the beginning Elon said that Tesla was potentially interested in building artificial general intelligence systems. Given the potentially transformative impact of a technology like that, it seems prudent to invest in technical AGI safety expertise specifically. I know Tesla does a lot of technical narrow AI safety research. I was curious if Tesla was intending to try to build expertise in technical artificial general intelligence safety specifically.
21:39Well, if I mean if if we start looking like we're going to be making a significant contribution to artificial general intelligence, then then we'll for sure invest in in safety. I'm a big believer in AI safety. I think there should be an AI sort of regulatory authority at at the government level.
21:58just as there is a regulatory authority for anything that affects public safety. So we have regulatory authority for aircraft and cars and sort of food and drugs and because they affect public safety and AI also affects public safety. So I think and this is not really something that government I think understands yet but I think I think there should be a referee that is ensuring or doing trying to ensure public safety for AGI.
22:30and if you think of like well like what are the elements that are necessary to to create AGI like the accessible data set is extremely important and if you've got a large number of of cars and humanoid robots processing you know pabytes of of video data and audio data from the real world just like humans that that's that might be the biggest data set. It probably is the biggest data set. because in addition to that, you can obviously incrementally scan the the internet. but what the internet can't quite do is is have millions or hundreds of millions of cameras in the real world and and with with like I said with audio and and and and other sensors as
23:19well. So, so I think we we probably will have the most amount of data and we probably the most amount of of training power. Therefore, probably we will make a contribution to AGI. Hey, I noticed the semi was back there, but we haven't talked about it too much. I was just wondering for the semi-truck, what are the changes you're thinking about from a sensing perspective? I imagine there's very different requirements obviously than just a car if and if you don't think that's true, why is that true?
23:57no, I think basically you can drive a car. I mean, think about like what what drives any vehicle. It's a biological neural net with with eyes cameras essentially. So if and and really what what is your your your primary sensors are two cameras on a slow gimbal, a very slow gimbal, that's that's your head. so if if you know if a biological neural net with with two cameras on a slow gimbal can drive a semi-truck then if you've got like eight cameras with continuous 360°ree vision operating at a higher frame rate and much higher reaction rate then I think it is obvious that you should be able to drive a semi or any any vehicle much better than human.
24:47Hi my name is Akshai. Thank you for the event. assuming you know Optimus would be used for different use cases and would evolve at different pace for these use cases. would it be possible to sort of develop and deploy different software and hardware components independently and deploy them you know in the in Optimus so that the overall you know feature development is faster for Optimus.
25:18I'm trying to see the question. Okay. All right. We did not comprehend unfortunately our neural net did not comprehend the question. Yeah. So, well, next question. Hi, I want to switch the gear to the autopilot. So when you guys plan to roll out the FSD beta to countries other than US and Canada and also my next question is what's the biggest bottleneck or the technological barrier you think in the current autopilot stack and how you envision to solve that to make the autopilot is considerably better than human in terms of like performance matrix like safety assurance and the human confidence and I think you also mentioned for V FSDB1 you guys
26:06going to combine highway and the city as a single stack and some architectural big improvement. Can you maybe expound a bit on that? Thank you. well that's a whole bunch of questions. Well, we we I we're hopeful to be able to I I think from a technical standpoint FSD beta should be it should be possible to roll roll out FSD beta worldwide by the end of this year.
26:33but we you know for for a lot of countries we need regulatory approval and so we are somewhat gated by the regulatory approval in other countries. but I you know but I think from technical standpoint it will be ready to go to to a worldwide beta by the end of this year. And there's quite a big improvement that we're expecting to release next month that will be especially good at assessing the velocity of of fastmoving cross traffic and and a bunch of other things. So anyone elaborate objects?
27:13Yeah, I guess so. There used to be a lot of differences between production autopilot and the full self-driving beta, but those differences have been getting smaller and smaller over time. I think just a few months ago, we now use the same vision only object detection stack in both FSD and in the production autopilot on all vehicles.
27:31there's still a few differences. The primary one being the way that we predict lanes right now. so we upgraded the modeling of lanes so that it could handle these more complex geometries like I mentioned in the talk. in production autopilot, we still use a simpler lane model, but we're extending our current FSD beta models to work in all sort of highway scenarios as well.
27:51yeah and the the the version of FSD beta that I drive actually does have the integrated stack. So the it it uses the FSD stack both in city streets and highway and it works quite well for me. we but we need to validate it in all kinds of weather like heavy rain, snow, dust and and
28:12just make sure it's working as better than the production stack in you know across a wide range of environments. but we're pretty close to that. I mean I I think it's I don't know maybe it'll definitely be before the end of the year and and may maybe November. Yeah, in our personal drives the FSD stack on highway drives already way better than the production stack we have. And we do expect to also include the parking lot stack as a part of the FSD stack before the end of this year.
28:44So that will basically bring us to you sit in the car in the parking lot and drive till the end of the parking lot at a parking spot before the end of this year. Yeah. And and in terms of the like the fundamental the fundamental metric to optimize against is how many miles per between inter necessary intervention. So just massively
29:06improving the how many miles the car can drive on in full autonomy before an intervention is required that is safety critical. so yeah that's that's the fundamental metric that we're measuring every week and we're making radical improvements on that. Oh hi thank you. Hi thank you so much for the presentation. Very inspiring.
29:33my name is Daisy. I actually have a non-technical question for you. I'm curious if you are back to your 20s, what are some of the things you wish you knew back then? What are some advice you would give to your younger self?
29:58Well, I'm trying to figure out something useful to say. Yeah. Yeah. Join Tesla would be one thing. yeah, I think just generally try to expose yourself to as many smart people as possible. I don't know, read a lot of books. you know, I do that did do that though. so I think there's there's some merit to just also like not being like necessarily too intense and and like enjoying the moment a bit more. I would say to 20 or 20 something me just a you know stop and smell the roses occasionally would probably be a good idea. you know it's like when we were developing the the the Falcon 1 rocket and on on the quadrilene atole and we had this beautiful little island that
31:03we're developing the rocket on and not once during that entire time did I even have a drink on the beach. I'm like, I should have had a drink on the beach. That would have been fine. Thank you very much. I think you have excited all of the robotics people with with Optimus. this feels very much like 10 years ago in driving, but as driving has proved to be harder than it actually looked 10 years ago. What do we know now that we didn't 10 years ago that would make for example AGI on a humanoid come faster?
31:40Well, I mean it it it seems to me that AGI is advancing very quickly. hardly a week goes by without some significant announcement. And yeah, I mean at this point like AI seems to be able to win at almost any rule-based game. it see it's able to create extremely impressive art. engage in conversations that are are
32:13very sophisticated, you know, write essays and these these just keep improving. and there's there's so much more so so many more talented people working on AI and the hardware is getting better. I think it's it's a a AI is on a super like a strong exponential curve of of improvement independent of what we do at at Tesla. and obviously we'll benefit somewhat from from that exponential curve of of improvement with AI.
32:47like Tesla just also happens to be very good at actuators at motors you know motors, gearboxes, controllers, power electronics, batteries, sensors and you know really like I say that you know the the biggest difference between the robot on four wheels and the robot with arms and legs is is getting the actuators right. Actu it's an actuators and sensors problem. and
33:14obviously the you know how you control those actuators and sensors but it's yeah actuators and sensors and how you control the actuators. It's I don't know we have to have like the ingredients ne necessary to create a compelling robot and we're doing it. So hi Elon you are actually bringing the humanity to the next level. Literally Tesla and you are bringing the humanity to the next level. So you said Optimus Prime Optimus will be used in next Tesla factory. My question is will a new Tesla factory will be fully run by Optimus program and and when can general public order a humanoid?
34:01Yeah, I I think it'll it'll you know we're going to start Optimus with very simple tests in the factory. you know like maybe just like loading a part like you saw in the video, loading a part you know carrying a part from one place to another or loading a part into a one one of our more conventional robot cells to you know that that welds a body together. So we'll start you know just trying to how do we make it useful at all and then and then gradually expand the number of situations where it's useful. and I think that that the number of situations where Optimus is useful will will grow exponentially.
34:43like really really fast. in terms of when people can order one, I don't know. I I think it's not that far away. well, I think you mean when can people receive one? so I don't know. like I'd say probably within 3 years and not more than 5 years with within 3 to 5 years you could probably receive an optimist.
35:14I feel the best way to make the progress for AGI is to involve as many smart people across the world as possible. And given the size and resource of Tesla compared to robot companies and given the state of human ordered research at the moment, would it make sense for the kind of Tesla to sort of open source some of the simulation hardware parts? I think Tesla can still be the dominant platformer where it can be something like Android OS or like a iOS stuff for the entire humanoid research. Would that be something that rather than keeping the optimus to just Tesla researchers or the factory itself can you open it and let the whole world explore human research?
36:00I think we have to be careful about Optimus being potentially used in ways that are bad. because that is one of the possible things to do. So I think we we you know we provide Optimus where you can provide instructions to Optimus but where those instructions are you know governed by some laws of robotics. that you cannot overcome.
36:31so you know not doing harm to others and you would have I think probably quite a few safety related things with with Optimus. Yeah. So all right we'll we'll just take maybe a few more questions and then and then and then thank you all for coming. questions one deep and one broad. On
36:54the deep for Optimus, what's the current and what's the ideal controller bandwidth? And then in the broader question, there's this big advertisement for the depth and breadth of the company. What is it uniquely about Tesla that enables that? Anyone want to tackle the bandwidth question?
37:14Yeah. So the technical bandwidth of the close to your mouth and loud. Okay. For the bandwidth question, you have to understand or or figure out what is the task that you want it to do and what is the fre if you took a frequency transform of that task. What is it that you want your limbs to do and that's where you get your bandwidth from. It's not a number that you can specifically just say you need to understand your use case and that's from that's where the bandwidth comes from. can what was the broad question? I don't quite remember the breadth and depth thing. I can answer the breath and depth. Yeah.
37:48we'll just we can I mean seriously on the band question I think we probably will just end up increasing the bandwidth or re you know which translates to the effective dexterity and reaction time of the of the robot. like you could it's safe to say that it's not one hertz and it's maybe you don't need to go all the way to 100 htz but I don't know maybe 10 25 I don't know but o over time I think the the bandwidth will will increase quite a bit or or translate it to dexterity and latency you you'd want to minimize that over time yeah minim minimize latency maximize dexterity I in terms of breadth and depth I I guess you know we've got we're a pretty big company at this point
38:46so we've got a lot of different areas of expertise that we necessarily had to develop in order to make autonomous or in order to make electric cars and then in order to make autonomous electric cars we've we've just I mean Tesla is like a whole series of startups basically and so far they've
39:05almost all been quite successful so we must we must be doing something right. and I you know I consider one of my core responsibilities running the company is to have an environment where great engineers can flourish and and I think in in a lot of companies I don't know maybe most companies if if somebody's a really talented driven engineer they they're unable to actually their talents are
39:34are are suppressed effect at a lot of companies and it's you know and and and some of the companies that that engineering talent is suppressed in a way that is maybe not obviously bad but but where it's just so comfortable and you're paid so much money and you but you your the output you actually have to produce is so low that it's like a honey trap you know so like there's a few honey trap places in Silicon Valley where they're not necessarily don't seem like bad places for engineers but you have to say like a good engineer went in and what did they get out and the output of of that engineering talent seems very low.
40:17even though they seem to be enjoying themselves. that's why I call it the a few honey trap companies in Silicon Valley. Tesla is not a honey trap. We're demanding and it's like you're going to get a lot of done. and it's going to be really cool. and it's you know not going to be easy but if you are a super talented
40:39engineer your talents will be used I think to greater degree than anywhere else you know SpaceX also that way. So hi Lan I have two questions so both to the autopilot team. So the thing is like I have been following your progress for the past few years. So today you have made changes on like the lane detection like you said that like previously you are doing instance semantic segmentation now you guys have built transform models for like building the lanes. So what are another some some other common challenges which you guys are facing right now like which you are solving in future as a curious engineer so that like we as a researcher can work
41:23on those start working on those and the second question is like I'm really curious about the data engine like you guys have like told a case like where the car is stopped so how are you finding cases which is very much similar to that from the data which you have like so little bit more on the data engine would be great so that's it okay I I'll answer the first question using occupancy network as an example. So what you saw in the presentation did not exist a year ago.
41:52So we only spent one year on time. We actually ship more than 12 occupancy network and to have a one foundation model actually to represent the entire physical world around everywhere and in all weather condition is actually really really challenging. So only over a year ago, we're kind of like driving a 2D word. If there's a wall and if there's a curb, we kind of represent with the same static edge, which is obviously, you know, not not not ideal, right? There's a big difference between a curb and a wall.
42:23When you drive, you make different choices, right? So after we realize that we have to go to 3D, we have to basically rethink the entire problem and think about how we address that. So this will be like one example of challenges we have we have conquered in the past year. Yeah. To answer the question about how we actually source examples of the tricky stopped cars. There's a few ways to go about this but two examples are one we can trigger for disagreements within our signals. So let's say that parked bit flickers between parked and driving. We'll trigger that back. And the second is we can leverage more of the shadow mode logic. So if the customer ignores the car but we think we
43:05should stop for it, we'll get that data back too. So these are just different like various trigger logic that allows us to get those data campaigns back. Hi. Sorry. Go ahead. thank you for the amazing presentation. Thanks so much. so there are a lot of companies that are focusing on the AGI problem and one of the reasons why it's such a hard problem is because the problem itself is so hard to define. Several companies have several different definitions. They focus on different things. So, what is Tesla? How's Tesla defining the AGI problem and what are you focusing on specifically?
43:42Well, well, we're we're not actually specifically focused on AGI. I'm simply saying that AGI is some is seems likely to be an emergent property of of what we're doing. because we're creating the all these autonomous cars and autonomous humanoids that are actually with an a truly gigantic
44:05data stream that's coming in and and being processed. it's by far the most amount of real world data and and data you can't get by just searching the internet because you have to be out there in the world and interacting with people and interacting with the the roads and and just you know it's earth is a big place and reality is messy and complicated. so, so I think it's sort of like, likely to just, it just seems likely to be an emergent property of, if, if, if, if, if, if, if, if, if, if, if if you've got, you know, tens or hundreds of millions of autonomous vehicles and, and maybe even a comparable number of humanoids, maybe more than that on the humanoid
44:41front. well, that's just the most amount of data. and if that that video is being processed, it just seems likely that, you know, the the cars will will definitely get way better than human drivers and the the humanoid robots will become increasingly indistinguishable from humans perhaps.
45:06and and so then, like you said, you have this emergent property of of AGI. and and arguably the you know humans collectively are sort of a super intelligence as well especially as we improve the the data rate between humans. I mean the thing like that seemed to be way back in the early days of the internet was like the internet was like humanity acquiring a nervous system where now all of a sudden any one element of humanity could know all of the knowledge of of humans by connecting to the internet almost all the knowledge or certainly a huge part of it. Whereas previously we would exchange information by osmosis by by you know by we'd have like in order to transfer
45:53data. So you would have to write a letter someone would have to carry the letter by person to another person and then a whole bunch of things in between and then it was like yeah I mean insanely slow when you think about it. and even if you were in the Library of Congress, you still didn't have access to all the world's information, and you certainly couldn't search it. and and obviously very few people are in the Library of Congress. So, I mean, one of the great sort of equality elements like the the internet is has been the most the biggest equalizer in history in terms of access to information and knowledge.
46:38and and any student of history I think would agree with this because you know you go back a thousand years there were very few books like like and books would be incredibly expensive but only a few people knew how to read and only an even smaller number of people even had a book now now look at it like you you can access any book instantly you can learn anything for basically for free it's pretty incredible so you I I was asked recently, what period of history would I prefer to be at the most? And my answer was right
47:18now. This is the most interesting time in history. And I read a lot of history. So let's Yeah. Let's do our best to keep that going. Yeah. And and to go back to one of the earlier questions I would answer like you can the the thing that's happened over time with respect to the Tesla autopilot is that we've just the the neural nets have gotten have gradually absorbed more and more software and in in the limit of course you could say simply take the videos as seen by the car and compare those to the the steering inputs from the steering wheel and pedals which are very simple inputs. and it in principle
48:03you could train with with nothing in between because that's what humans are doing with a biological neural net. You could train based on video and and and what trains the video is the the moving of the the steering wheel and the pedals with no other software in between in between. We're not there yet, but it's gradually going in that direction.
48:27all right. Wait, last question. How you going? I think we got a question at the front here. hello. They're right there. Oh, we'll do two questions. Fine. Over there or here? hi. thanks for such a great presentation. We'll do your question last. Okay, cool. with FSD being used by so many people, do you think what's the comp how do you evaluate the company's risk tolerance in terms of performance statistics? And do you think there needs to be more transparency or regulation from third parties as to how what's good enough and defining like thresholds for performance across some many miles?
49:07sure. Well, the you know I I the the number one design requirement at Tesla is safety. So like and that goes across the board. So in terms of the mechanical safety of the car we have the lowest probability of injury of any cars ever tested by the government for just passive mechanical safety essentially crash structure and airbags and whatnot.
49:36we have the best the highest rating for active safety as well. and I think it's going to get to the point where you you the act of safety is so ridiculously good, it's it's it's it's like just absurdly better than than than a human. and then with respect to autopilot we do publish this
49:57broadly speaking the statistics on miles driven or with cars that have no autonomy or Tesla cars with no autonomy with kind of hardware one, hardware 2, hardware 3. and then the ones that are in FSD beta. and we see steady improvements all along the way. and you know, sometimes there's there's this dichotomy of, you know, should you wait until the car is like, I don't know, three times safer than a person before deploying any technology.
50:29But I think that's that is actually morally wrong. at the point at which you believe that the that adding autonomy reduces injury and death, I think you have a moral obligation to deploy it. even though you're going to get sued and blamed by a lot of people because the people whose lives you saved don't know that their lives are saved and the people the people who who do occasionally die or get injured they definitely know or their estate does that it was you know whatever there was a problem with with with autopilot. that's why you have to look at the at at the numbers in sort of total miles driven, how many accidents occurred, how many accidents were
51:12serious, how many fatalities, and you know, we've got well over 3 million cars on the road. So, this it's that's a lot of miles driven every day. it's not going to be perfect, but what what matters it is that is that it is very clearly safer than not deploying it. yeah. So I think last question
51:36last I think. Yeah. So thanks. What's the last question here? Got to come here. I got it. Okay. yeah, I got it. Okay. hi. So I do not work on hardware. So maybe the hardware team and you guys can enlighten me. why is it required that there be symmetry? in the design of Optimus because humans we have handedness right we are we use some set of muscles more than others over time there is wear and tear right so maybe you'll start to see some joint failures or some actuator failures more over time I understand that this is extremely pre-stage also we as
52:24humans have based so much fantasy and fiction over superhuman capabilities like all of us don't want to walk right over there. We want to extend our arms and like we have all these you know a lot of fantasy fantastical designs. So considering everything else that is going on in terms of batteries and intensity of compute, maybe you can leverage all those aspects into coming up with something well I don't know more interesting in terms of your the robot that you're building and I'm hoping you're able to explore those directions.
53:04Yeah, I mean I think it would be cool to have like you know make inspector gadget real. That would be pretty sweet. so yeah I mean you know right now we we just want to make a basic humanoid what work well and and our goal is fastest path to a useful humanoid robot. I think this is this will ground us in reality literally and ensure that we
53:27are doing something useful like one of the hardest things to do is to be useful to to actually and then and then to have high utility under the curve of like how many people did you help t you know and how much help did you provide to each person on average and then how many people did you help the total utility like trying to actually ship useful product that people like to
53:56a large number of people is so insanely hard. I it's boggles the mind. you know that's why like I said like man there's a hell of a difference between a company that has shipped product and one that has not shipped product. it's again this is night and day. and then even once you ship product can you make the cost the value of the output worth more than the the cost of the input which is again insanely difficult especially with hardware. So but I think over time I think it we cool to do creative things and have like eight arms and whatever and have different versions and maybe you know there'll be some hardware like companies that are
54:36are able to add things to an Optimist like maybe we we you know add a power port or something like that or attach them you can add you know add attachments to your Optimus like you can add them to your phone. there could be a lot of cool things that could be done over time and there could be maybe an ecosystem of small companies that or big companies that make add-ons for Optimus. So, with that
55:02I'd like to thank the team for their hard work. you guys are awesome and
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