Rodney Brooks is a roboticist, former head of CSAIL at MIT, and co-founder of iRobot, Rethink Robotics, and Robust.AI. Please support this podcast by checking out our sponsors:
– Paperspace: https://gradient.run/lex to get $15 credit
– GiveDirectly: https://givedirectly.org/lex to get gift matched up to $300
– BiOptimizers: http://www.magbreakthrough.com/lex to get 10% off
– Four Sigmatic: https://foursigmatic.com/lex and use code LexPod to get up to 60% off
– SimpliSafe: https://simplisafe.com/lex and use code LEX to get a free security camera
EPISODE LINKS:
Rodney’s Twitter: https://twitter.com/rodneyabrooks
Rodney’s Blog: http://rodneybrooks.com/blog/
PODCAST INFO:
Podcast website: https://lexfridman.com/podcast
Apple Podcasts: https://apple.co/2lwqZIr
Spotify: https://spoti.fi/2nEwCF8
RSS: https://lexfridman.com/feed/podcast/
YouTube Full Episodes: https://youtube.com/lexfridman
YouTube Clips: https://youtube.com/lexclips
SUPPORT & CONNECT:
– Check out the sponsors above, it’s the best way to support this podcast
– Support on Patreon: https://www.patreon.com/lexfridman
– Twitter: https://twitter.com/lexfridman
– Instagram: https://www.instagram.com/lexfridman
– LinkedIn: https://www.linkedin.com/in/lexfridman
– Facebook: https://www.facebook.com/lexfridman
– Medium: https://medium.com/@lexfridman
OUTLINE:
Here’s the timestamps for the episode. On some podcast players you should be able to click the timestamp to jump to that time.
(00:00) – Introduction
(07:33) – First robots
(28:58) – Brains and computers
(1:01:47) – Self-driving cars
(1:21:57) – Believing in the impossible
(1:32:47) – Predictions
(1:43:49) – iRobot
(2:11:11) – Sharing an office with AI experts
(2:23:21) – Advice for young people
(2:27:07) – Meaning of life
This is an unofficial transcript meant for reference. Accuracy is not guaranteed.
The following is a conversation with rodney brooks one of the greatest robot assists in history. He led the computer size and artificial intelligence laboratory at mit, then cofounded
robot, which is one of the most successful robotics companies. Ever then, he cofounded rethink robotics that created some amazing collaborative robots like baxter and swear. Finally, he cofounded robust daddy. I, whose mission is to teach robots common sense, which is a lot harder than it sounds the support. Despite guess, we share our spotters in the description
as a side. No I'm say that rodney is someone. I've looked up to for many years in my now over two decades journey and robotics, because one he's a ledge
great engineer of real world systems and to he's not afraid to stay controversial opinion
the challenge. The way we see
a world of course. While I agree with him on
one of his critical views of ai? I don't agree with some others and he is fully supportive of such disagreement. Nobody ever built anything great by being fully agreeable, there's always respect and love behind our interactions and when a conversation is recorded like it was for this podcast, I think a little bit of disagreement is fun as usual. I'll do a few minutes of as now no ads in the middle. I try to make these interesting. So hopefully you don't skip, but if you do, please do check out the sponsor links in the description. It is the best way to support this podcast. I use their stuff and enjoy it. Maybe you will too
they show- is brought to you by paper space gradient. These guys are amazing. It's a platform that lets you build, train and deploy machine, learning, models of any size and complexity. I love how powerful and intuitive it is unlikely going to use paper
space for a couple of machine learning. Experiments I'm doing as part of an upcoming video
study. I, of course I highly recommend use. It does, of course, by Jeremy howard, who is
as legit as it gets in the machine learning space as an educator as a programmer. I highly recommend his stuff and plus he's just a good and a brilliant human being. You can host notebooks on there. You could swap out the computer instance any time, so you can start out on a small scale, gp your cpu instance and then swap out once the compute needs increase to give Grady the try, visit, gradient dot, run, slash lex,
and use the sign up link there, you'll get fifteen dollars and free credit, which you can use the power or next machine learning application as gradient that run slash lex. I hope you use it. I hope you enjoy
and I hope to make a bunch of machine learning videos in the near future. This shows also brought you buy, give directly and not prime
the less you send money directly to people living in extreme poverty, get directly. Donors includes previous gusts of this very podcast like jack, dorsey, elon, musk, metallic buren, when the castle and Peter singer,
It may seem, like others more optimal ways to do it, but that's not actually the case. Hundreds of independence,
These have shown that direct giving can have passed
The impact on health, nutrition, income, education and more studies
giving cash unconditionally can more than double.
incomes increase.
School, enrollment and entrepreneurship decrease, skipped, meals, illness and depression and cut domestic violence by one third, it does not decrease hours worked or increase spending on temptations like tobacco and alcohol
there's a spill over effect to our every dollar, given amounted to dollar sixty cents in the local economy in the last decade give directly
has delivered foreigner million dollars to over nine hundred thousand recipients across nine countries. Visit give directly dot. Org splash legs in your first gift will be matched up to three hundred dollars. Let's give directly dot org such lex.
The next sponsor is by optimism that have a new magnesium supplement when I fast or I'm doing, kyoto. Carnivore saudi
potassium magnesium are essential. Magnesium, I think, is the most trick. You want to get rid of those. That's why use magnesium breakthrough from by optimized most?
what does contain only one or two forms of magnesium like glass and eight or citrate,
when, in reality, there are at least seven that your body needs and benefits from
I recently did instagram live with Andrew huber, men or women
about magnesium- and there we are-
about magnesium offline as well. He he educated me
about it. Quite a bit again is its tributaries.
right and as a lot of benefits. If you get a right and by optimizing is just
he's right here right, it's kind of
Credible when you get this part right, how easy it is to do the kyoto or the fasting, I might go to mag break through the council s legs for a special discount. That's mag breakthrough dot com, slash lex. This shows also sponsored by forcing matic the maker of delicious mushroom coffee and plant based protein. Does the coffee tastes like mushrooms, you ask? No, it is not its
delicious, and also just the ritual of drinking hot coffee. There has this aromatic.
Flavour and obviously smell to it in the morning. That's just something that I got
had always from my mind, get extremely focused in the morning. Is the caffeine, the aroma, the hot coffee,
The cap with the steam coming off of it
sign on raised again zone and the first to three hours
to make sure I allocated deep work and deep thinking really focusing on the difficult tasks.
Mine is sharp and jumping from one element of a task to another and getting stuff done, and you I get up to forty percent off and free shipping, a mushroom coffee, bundles. If you go to forcing matic dot com, slash lex, that's for submitting the calm. Slash lacks the shows also brought by simply safe home security company, designed to be simple and effective. It takes only thirty minutes to set up in customized system free needs on simply saved our com, slash lex. I have set up in my place and love it. That's the whole thing the camera
the monitoring, the response from the set up to the day to day operation. Super simple, like I've, said maids as before. I just like on people on design solution.
Then implemented fairly well and in the home security companies base. Simply safe is on the best. You can go on the internet and the ask around on read it and elsewhere, like what is the best home security company in a lot of people, say simple, safe and you go to simply say that cause us legs to customize your system and get a free security. Camera plus a sixty day. Risk free trial again that simply saved our com, slash lex there,
Is it works treatment past, and here is my conversations with rodney brooks what is the most amazing or
beautiful robot that you ve ever had the chance to work with an old timer domo, which was made by one of my grad student and it singer,
it now sits in Daniela ruthless office track of resale, and it was just a beautiful robot. An errand was really clever. He didn't give me,
budget ahead of time. He didn't tell me what he was gonna. Do it started spending money,
things have happened to morrow he Jeff Weber, who those mechanical engineers who arrogance
did he bring with him when he became a grad student built this beautiful gorgeous robot done by which the upper torso humanoid to two arms, where we think there are three things?
handsome and them face eyeballs all not
not the eyeballs, but everything else, serious elastic actuators. You can interact with a cable driven.
All the motives are inside and it's just gorgeous
the eyeballs are actually turner. Oh you rivals or actual with cameras and in a soul, had a visual attention mechanism, while looking
people came in with him they face and talking with them. Why was it amazing? The beauty of it
He said what you said, but was the most ambiguity. What is the most beautiful? It's mechanically gorgeous as
as everything errand builds, has always been mechanically gorgeous. It's just exquisite in the detail. Are we talking about mechanic?
like literally the amount of actuators. The actual aid is the cables he
I'm a dies as different parts, different colours- and it's just looks like a work of art or about face- is that you find the face. Beautiful and robots
when you make her a robot, it's making a promise for how well it were able to
right, so I am always encourage my students not to over promise in them even with it's essence like the the thing it presents, it should not over promise yeah so that the joke I make, which I think you will get is if you robot looks like albert einstein. It should be as smart as albert einstein so that the only thing in in a dummy's face is the eyeballs and
Cause, that's all it can do. It can look at you and pay attention, and- and so there is no f- it's not like one of those japanese robots. That looks exactly like a person at all but see the thing is as humans and
two, don't just use eyes for as attention all mechanisms. They also used to communicate as communication.
Doc unlucky, you looking thing and look back at you and that doesn't know where we're going.
It'd be looking at that thing, yeah or intent you know, and and and both Baxter and sawyer, at rethink robotics. They had a screen with know graphic eyes, so it wasn't actually where the cameras were pointing, but it the the eyes would look
in the direction of about the move it out, and so people in the factory nearby were not surprised by its motions. Just gave that intent away before we talk about back, to which I think is a beautiful robot. Let's go back to the beginning. When did you first?
fall in love with robotics, we're talking about beauty and love to open. The conversations is great. I've got these. I was born in the end of nineteen. Fifty four m. I grew up in Adelaide south australia and I have these two books that I dated
nineteen. Sixty one son guessing my mother, found them in a store and sixty two sixty three,
How and why wonder books how my wonder book of electricity,
and a hound. Why won the book of giant brains and robots and the?
I learned how to build circuits in one eye was able mine, simple circuits, and I'm I'm I'm ready
Let the binary system and thoroughly drawings, mostly of robots and
and then I tried to build him for the rest of my childhood weight. Sixty one. You said this was when the with two books: I've still got them.
At home, that what what is the robot mean in that context, the date they were some of the robots are they had where arms began to?
move nuclear material around, but they had pictures
Welding, robots, tat, looked like humans under the sea, welding staff underwater, so they went real robots about in a what people are thinking about for robots.
What were you thinking about? Were you thinking while humanoids? Were you think about arms with fingers? We, you think about faces a car, not actually. To be honest, I realized my limitation on building mechanical stuff, so I just built the brains for mostly I out of.
different technologies? As I got older, I built asia, a learning system which was chemical based and I had this ice cube tray. Each well was a cell and by applying voted to electric
it would build up a copper bridge so over time it would. It would learn as a simple network and I could teach it stuff, and that was mostly thinks we're driven by my budget and nails as electrodes
Emma ice cream, Emma ice cube tray was was about my budget.
stage later I managed by train justice and I could build gates and put flops and stuff so the one when you first row, basel was an escape. Try, yeah and it was very cerebral- doesn't want to add that as well
Just then a decade or so before in nineteen fifty Alan turing wrote the paper that formerly the turing test and he opened the paper with the question. Can machines think so? Let me ask you this question: can machines think can your ice cube tray one day think.
and certainly machines can think, because I believe, your machine and I'm a machine. I believe we both think and I think it's a threat and I think any other philosophical position is solid
a little ludicrous. What does that mean? If, if it's not something that we do
and we I mean machines, so yes, machines can, but do we have a clue how to build such machines? That's a very different question: are we capable of building such machines? You're? Always mamma? We think with my love to do anything, but maybe we're not maybe it'll but does not stop
item after bill. That's like us, the girl computer, that alan turing was thinking about. Do you think there is something
and the mentally or significantly different between it.
Computer between our ears the biological compete
or that humans use and that the computer that he was thinking about from a from a sort of high level, philosophical yeah I, but I believe it the that's very wrong impact on halfway through a I think, it'll be about a four.
the navy page book. Titled wasn't title is not even wrong and
I maybe I'll tell you a bit about yes boys. So there's two to three thrust stood: one is the history of computation what we call com?
station goes all the way back to earth have some manuscripts in latin from sixteen fourteen and sixteen twenty by napier and and kepler through the babbage and and lovelace and then turing's nineteen thirty six paper is, would eat away with what we think of as the invention of of of modern computation and that papal by the way did not say
alt to you know, invent computation. It set out to negatively answer one of hilbert's three latest set of problems. He called it after that effective way of of of of getting answers and and hilbert Hilbert really worked with rewriting rules, as did am, I
a church who are also, at the same time a month earlier, ensuring disproved helmets. One of these three hypotheses, the other two may already been disproved by godel, so during set out to disprove that it's always easier to disprove these things than to prove that there is an answer and- and so he needed,
and it really came from his is the professor- was an underground. It at cambridge had said who turned it into? Is there a mechanical process? So he'd want to have a show? A mechanical process that could calculate numbers, because that was a mechanical process that people used to generate tables. They were called computers, the people at the time,
and they follow the set of rules where they had paper and they would write numbers down and based on the member states, keep riding other members and they would produce numbers for these tables engineering tables
that the more nor innovations iterations they did, the more significant digits came out, and so cheering in that paper set out to define what sort of machine could do that mechanical machine.
Where it can produce an arbitrary number of digits in the same way a human computer did and- and he came up with a very simple set of constraints- where there wasn't an infinite supply of paper- that isn't, the tape of the turing machine and each turing machine had a set of and came with, a set of instructions that, as a person, could do with pencil and paper, write down things on the tape and erase them and put new things there. And he was able to show that that system was not able to do something and hilbert.
I cite three disprove, but he had the showed this was this. The system was good enough to do whatever it can be done by couldn't do this every year and there he said, and he says in the paper I
have any real arguments for this, but based on intuition. So that's how he defined computation and then, if you look over the next from
nineteen? Thirty, six up until a really around nineteen? Seventy five you see people struggling with. Is this really what computation is?
marvin minsky very well on an ai, but also a fantastic mathematician in his book. Finite and infinite machines from the mid sixties, which is a beautiful, beautiful mathematical book and says, says at the start of the book. Well, what is computation
Having said this, there is some sort of think it's that it doesn't really matter whether stuffed made of wood or plastic. It's just you know, but relatively cheap stuff can do this,
stuff and so seems like computation and donald knuth in is the first volume of his in out of computer programming.
in around nineteen sixty eight there as well, what's computation this stuff like tshering, says that a person could do each step without too much trouble. And so one of his examples of what it'd be too much trouble was a step which required knowing whether fermat's last theorem was true or not, because it was not known at the time and it's too much trouble for a person to lose a step and and how craft an omen
sort of set a similar thing labour that year and by nineteen seventy five in a crop. The moment book they saying well below
I really know what computation that has been intuition says. This is sort of about right, and this is what it is at computation. It says sort of agreed upon thing which happens to be really easy to implement in silicon, and then we had moore's law which took off and it's been credibly powerful tool. I certainly wouldn't argue with that. The version we have a computation credibly powerful. Can we just does it take a pause? The goal we're talking about is this
infinite tape, for some simple rules on how to write and on that tape, and that's that's what we're gonna think about this is computation and its model that the union's, how humans do that, and I think it's a cheering says in the thirty six paper, one of the critical fact
is that a human has a women than amount of memory. So that's what we're gonna put onto outcome, a mechanical computers- so am, I am like master chow.
you're here is an animal is not given by the universe it was. This is what we're gonna call computation here.
It has this really? You know it had this really good implementation, which has completely changed out technological work. That's computation! Second part of the book I or argument in the book. I have this two by two matrix with science. In the top row engineering, the bottom row left column
is intelligence. Right column is life. So in the bottom row the engineering, this artificial intelligence and out of life in the top road was near a science and abbe agenda,
how helped us living matter turn and how does non living matter become living bat four disciplines, these four disciplines all came into the current form in the period nineteen. Forty five to nineteen sixty five, as suggested there was neuroscience
four, but it was, it wasn't affective neuroscience. It was the other of these ganglia and those electrical charges, but no one knows what to do with it and, furthermore, the out door, a lot of players who common across them have identified commonplace, except for artificial intelligence. Have your genesis I don't have, but for any other pair I can point to people who working and a whole bunch of them by the way were the reason
lab for electronics at mit, were Warren Mcculloch held, held, held forth and and fact that mccullough the pits let them in mature anna wrote. The first paper on functional neuroscience called what the frog's eye toes the frog's brain, where, instead of just being a bunch of nerves, they sort of showed,
different. I'm a comical components were doing and telling other anatomical components and generating behaviour and they brought it. Would you put them as basically
the fathers, the or the one in their early pioneers in what are now
artificial, neuron networks. Each year, the wind, Mcculloch and pits pit was a much under him in nineteen, forty three had written a paper by bertrand russell on a calculus forty ideas, eminent in
neural systems where they had tried to without any real proof. They had tried to give a formalism for neurons, basically in terms of logic and gates or gets nuggets with with no real evidence of that was what was going on, but they they talked about what was picked up by minsk for his nineteen fifty three dissertation, which was a wasn't it.
Isn't that what we call it today, it was picked up by John VON Neumann when he was designing the ed back computer in nineteen. Forty five, he talked about it's components. Being neurons based on anime references he's only got three references: anonymous. The mcculloch pitts paper.
So all these people, and then the people and the artificial life people which was John VON Neumann, originally as an overlap between
well they're all going around the same time, and three of these four disciplines turned to computation as their primary metaphor. So I I've got a couple of chapters in the book. One is titled weight, computers up people cause, that's where our computers came from. Here am a you know, from people who were computing stuff and then have another chapter. Wait people are computers which is about computational neuroscience. Yes, so there's this whole circle here and that computation is it, and you know I have talked to two people about mil. Maybe it's not computation that goes on in the head course. It is near okay. Well,
When even masks rocket goes up, is it computing? Is that how get and all but by computing? But we ve got this idea. If you want to build an eye system, you write a computer program yet innocent who so the word computation
very quickly start doing a lot of work there.
Was not initially intended to do is to think in the same way. If you talk about the universe is essentially performing, a computational
I won't from our citizens and in the computation even turn rockets in the computation.
By the way, when you say computation and our conversation, you tend to think of him,
imitation narrowly in the way touring thought of computation. I it's it's. It's gotten very okay, it you know: squishy yeah, squishy or gravitas, and but
computation and away touring thinks about it and the way most people think about it actually fits very well with
Thinking like a hunter gatherer. There are places and can be stuff in places than the stuff in places can change
stay there until someone changes it, and it's this metaphor of placing container,
which you know, is a combination of our place cells in out hippocampus emma cortex, but this is this is how we use metaphors.
when you think about when we get outside of a metaphor range, we have to invent tools which we can sort of switch on to you. So calculus is an example of a tool. It can do stuff that our raw reason,
can't do it we got conventions of when you can use it on, but sometimes you know people tried it all the time we always try to get this.
Oh metaphors, for things, which is why quantum mechanics it's been such a problem for a hundred years, because
the particle. No it's away, it's gotta be something we understand, and I say no it somewhere.
Mathematical logic that's different from most, but we want that metaphor. We know I suspected that
in one hundred years or two hundred years from now, neither quantum mechanics nor nor dark matter will be talked about in the same terms, in a in the same way, the protestants theory eventually went away because it just wasn't an adequate explanatory metaphor. That metaphor was the stuff: there is
thus, in the burning the burning is in the matter?
The burning was outside. The matter was the auction, so our desire
metaphor and combined with their limited cognitive
capabilities, gets us into trouble. That's my argument in this book now
people say well. What is it then, and I say well, I wish I knew that Fred's book about that, but I know it gives some ideas, but so so this is the three things. Computation is sort of a particular thing we use and oh can I tell you one beautiful thing
won't be asking of him so he's you know. I used an example of a thing, that's different from computation, you hit a drum and it vibrates, and there are some some out stationary points on the drums
the waves going up and down the stationary points. Now you could compute them to arbitrary precision, but the drum just knows them run doesn't have to compete with
was the very first computer program ever written by a lovelace to compute bernoulli numbers,
newly numbers are exactly what you need to find those stable points in the drum surface while anyway- and there was a bug in the program, half the arguments to divide were reversed. In one place
and still worked when I did she never got to run, but they never built the analytical engine. She wrote the program without without you know, so, as a computation, computation is sort of a thing
become dominant as a metaphor, but or is it the right metaphor and all three of these four fields adopted computation and,
You know that a lot of its world around roundworm Mcculloch and his all his students- and he found a lot of people
and the
and and our human metaphors are limitations to human thinking and play into this the three themes of the book. So I have a little to say about computation, so I see you're saying that there is a gap between the computer or the the the machine that performs computation and this machine.
In that appears to have consciousness and intelligence yeah. We are that piece of meat in your hair piece of me, and maybe it's not just the meat in your head. It's the rest of you too. I mean you, have you have seven year old system and it got I'd. Ident also believe not believe, but we're not dance
around things without now, but I tend to believe other humans are important like so we almost like I just
I don't think we would ever have achieved a level of intelligence. We have with our other humans that I'm not saying so confidently, but I have an intuition that some of the intelligence is in the interaction yeah and I think you know, I think it's seems to be very likely again. We, you know the speculation, but we are species and probably are.
I mean the and those to some extent, because you can spine the old bones where they seem to be counting on them by putting notches that when near them in near neanderthals had done. We are able to put on some of our stuff outside our body into the world and then other people can share it, and then we get these tools that become shared tools and so the whole coupling that would not occur, and you know the single
deep learning network which was fed all of welsh or something yeah. The than your network can step outside of itself book. Is there
there. Some mom can explore this dark room.
A little bit and try to get something. What what is it
magic words. The magic come from in the human brain decrease the mind. What was your sense scientists?
had tried to understand it and tried to build it. What are the directions it if followed might be? Productive is a creative, interactive robots. Is it creating large
Deep neural networks that do like self supervised learning and just like wolf will will discover that when you mix things
enough? Some interesting things will emerge. Is it to physics and chemistry biology?
like artificial life angle lego, sneak up in this house for quadrant major theme.
Is there anything you at your most?
get to bed all your money to finance that I wouldn't
Ok, so every intelligence. We know women
in an animal intelligence, dog intelligence, octopus intelligence, which is very different sort of architecture from promised and all the intelligence we know and perceive the world in some way and then have action in the world
what they are able to perceive objects in a way which is actually pretty damn for not phenomenal and surprising. We tend to think you know that the that that the the the box over here between us, which is a soundbox, I think is, is a blue box but a blueness and is something that we construct with with them or color constancy. It's not a it's, not a it's. Not the blueness is not a direct function of the photons were receiving it's actually context in, which is why you can turn them in, or you may be seen the examples
someone turns a stop sign into a some other sort assigned by just put a couple marks on them in the deep learning system gets a wrong them of this, but the stop signs read it in a way that was a thing is to have. A sort of time is red must not intrinsic in just the photons exactly
a construction of an understanding of the whole world and the relationship between objects, the count color constancy, but era tendency in order that we can archived paper
quickly- is due to show a lot of data and give the labels and hope he figures it out. But it's not figuring it out. In the same way, we do. We have a very complex perceptual understanding of the world dogs have a very different perceptual. Understanding based on smell may go smell smell a post. They can tell
How many different dogs have visited that in the last ten hours and how long ago, there's all sorts of stuff that we just don't perceive about the world and taking a single snapshot is not perceiving about the world has not seen the the registration between us and the object and registration is a a philosophical concept. Brian Kemp will smith talks about a lot very difficult, squirmy thing to understand, but I think
none about systems do that we ve always taught them in a high about the symbol, grounding problem, how symbols that we talk about a grounded in the world and when
deep learning, came along and start of labeling images. Peoples.
the grounding problem has been solved now the late
The problem was solved with some percentage. Accuracy was different from the grounding problem.
So you are there, you go,
sounds more like in the woods called. The marks paradox that the highlights this color intuitive notion. That reasoning is easy.
but perception and mobility are hard yeah, we shared an office owen, will
I was working on television and he was working on his first about robot forward. Those conversations I have a great
He's still can maybe
my brain just to believe this kind of notion that perception is really hard?
they can make sense of why we humans have this point tuition, but what's hardened,
well. Let me let me give a give us an another another story. If you go back to the original,
you have teams working on a I am from the late fifties into the sixties.
You gotta there I lab at mit.
well, who was it that was doing. That was a bunch of really smart kids who got into mit and they were intelligent. So what's intelligence about well, stop there.
That playing chess doing, integrals, that was that was hard stuff, but in a baby could see stuff. I wasn't
I wasn't meant tat- would allow me in anyone could do last on intelligence, and so, if you know this, there was this intuition, but the hard stuff is the things they were good at and the easy stuff was
stuff that everyone could do yeah, maybe I'm over playing a little bit, and I think that an element of that
Yeah I mean the ira. How much truth there is too on the chest, for example, has was the longest time seen as the highest level of intellect until we got computers with,
rather than people, and then we really it'll be. If you go back the nineties, you see. Do you know the stories in the press around when when them passport was beaten by the blue of this is the end of all sorts?
things. Computers are gonna battle, do anything from now on and we still have exactly the same stories with alpha zero. The go playing program yeah, but still to me reasoning, is a special thing and perhaps nowhere
I think we were really bad at reasoning. We just use these analogies based on our humble gather intuitions, but why
That's not don't you think they believe construct. Metaphor is really.
Awful thing always worried. It is its constructive, the metaphor and registering back the ambling concentric rings.
like this. Now, what we're doing with with vision to and were telling stories
struggling good models of the world yeah yeah, but I, but I think we jump between what we are capable of and how it
doing it right. Yeah was a little confusion that when I'm sure I'm telling each other stories as exactly that trying to do at each other. Now I I just think a I'm, not exactly some trying to pull apart this more of a paradox. I don't feel-
the paradox that little, what did evolution one of evolution spend time on it spent his time on getting us to.
see that move in the world that was
six hundred million years, as multi cell creatures doing that, and then it was. You know relatively recent that we that we, you know were able to, hunt or or gather or even even animals, hunting, that's much more reason and then and then that anything that we, you know with speech
the language, those things are a couple of hundred thousand years, probably if, if if, if that long and in agriculture, ten thousand years, you know all that stuff
was built on top of those earlier things which took a long time to develop the few then look at them
engineering of these things, so building it into robots was the hardest part of robotics do think as
the deck, easy work on robots in the context of what we're talking about vision, mega perception, the actual cost of the bar mechanics
movement. I'm kind of drawing parallels here between humans and machines. Always like a would, you think, is the hardest part of robotics, I sort of think all of them. There are no easy paths to do well and we
we sort of go reductionist and we reduce it to one. If only we had all the the location of all the points in three, the things would be great
if only we had labels on the on the images. You know things would be great, but you know, as as we see, that's not good enough of some deeper understanding. But if you, if I came to you and I could
of one category of problems and robotics instantly
give you are the greatest pleasure.
I mean. Is it? You know you look a robots that manipulate objects. What was hard about that, you know, is it
perception? Is it the the reasoning about the world that common sense reasoning? Is it the actual bill
being a robot, thus able to interact with the world? Is it like human aspects of a robot interacted with humans and that their game,
the theory of how they work. Well, together. Let's talk about manipulation, procession cause. I had this really blinding moment. You know I'm a grandfather, sir grandfathers had blinding mounts. Yes, just the three or four miles from here and last year, my sixteen month old grandson was in his new house
first time, my first time in this house and had never been able to get to a window before with his head, similar windows, and he goes up to this.
though, with a handle on but he's never seen before, and these
one hand pushing the way
on the other hand, turning the handle opened the window? He knew it two different hands, two different things. He knew how to how to put together
and these sixteen months old and there you are watching in awe,
in an environment, environment, he'd, never seen before mechanism. Now, how did he do that? I guess that's a good question. How do we do that? That's why it's like a guy like you could see the the
the leap of genius from using one hand, to perform a death to combining doing for fun and manipulation is really difficult to has both unnecessary to complete their
action in completely different and it never seen a window open the curve,
inferred somehow handle open something
there may have been a lot of slightly different failure cases that you didn't
see yet not with a window but with other objects of turning and twisting handles
well the the you know. There's a there's, a great counter to you know: reinforced reinforcement, learning will just give you know the the robot or you give the robot plenty of time to try everything. Yes, se can I
a little side story here, though, I'm in them
mind in london as this for four years ago, where you know, there's a big google building,
and then you go inside and they go through the small security. And then you get the deep mind where the other Google employees cargo and I'm in a I'm in a conference room by a conference room with some of the people, and they tell me about the reinforcement, learning experiment with robots which are just trying stuff out and then my robots.
their their is that we sold them and and they really like him, cuz soils are compliant and consent forces, so they don't break when it's bashing and the wolves they they stop him because the stuff and enter so you just let the robot do stuff and eventually figure stuff.
By the way sort without my robot manipulations, a robot arms son that so as tony is a robot yeah, I'm just yet,
swear, so he's a robot. I'm that my company rethink robotics built. I give her cards
cause the warranty mine. I am that you know
in the next room. These robots are just bashing around to try and use reinforcement, learning to learn how to act. And what can I go see them? Oh, no
Now my robots tat. Yes, if it has a less anyway. The point is you know this idea that you just let reinforcement, learning figure everything out is so count of how a kid does stuff so again story. My my grandson. I gave him this. The sub box that had lost,
Different lock mechanisms he didn't random in army was eighteen months, are hidden randomly try to touch every surface or push everything he found. He could see what where the mechanism was
and he started exploring the mechanism. But each of these different lock mechanisms it and there was reinforcement, no doubt- have some sort going on there, but he applied a pre filter which cut down the search space dramatic
What is the I, I I wonder to what level we're able to introspect what's going on, because, what's also possible is yes, something like reinforcement, learn
having gone on in the mind in the space of imagination. So, like you have a good model of the world, you predicting and you may be running those too
The thousands of like loops but you're like as a human you're, just looking yourself trying to tell a story of what happened in a mite,
It seems simple, but maybe there's a lot of computation goin on whatever,
but there is also a mechanism- has been built, happens. You're not just ran with search that improves dramatically yeah the that that pruning or pruning step, but it doesnt it's possible that that's the isa
I think that the to a neural network inside a reinforcement learning algorithm is it possible it. Yeah
and so it's possible I, but but I you know I I I I will I'll be incredibly surprised if that happens. I'll also be incredibly surprised that you know, after all, the decades that I've been doing this where every every few years, someone thinks now. We've got
but now we've got to you know four or five years ago I was saying: I don't think we got it yet, and everyone was saying yo, you don't understand how powerful AI is. I had people tell me you don't understand how powerful it is, and I you know I I I sort of had a a track record of what the world of boundless think. While this is no different from
or will we have bigger computers? We have bigger computers in the nineties and we could do more shit stuff back but against. Let me them. They pushed back their aunt em journalists optimistic and try to find the beauty, and things
I think, there's a lot of surprising and beautiful things that neural net
works. This new generation of deep learning revolution, as revealed to me, is continually been very surprising. The kind of things is able to do,
higher that over saying like this, we ve solved intelligence that another big lee. But is there something surprise
It's beautiful to you about neural networks that were she, you said back and said I I did not expect this. Oh, I think I think the performance their performance-
on image in? It was shocking for computer vision. Those early days was just very vague, while ok, that doesnt mean
They are solving everything
in computer vision we need solve, or in
vision for robots cut about alpha zero and self play mechanisms and reinforcement, learning isn't that that was all in in donald mickey's, nineteen sixty one paper, everything that was there, which introduced reinforcement, learning and not but come on so now, you're talking about the actual techniques.
But isn't this surprising to you? The level is able to achieve no human supervision of chess play like to me. There's a big,
big difference, maybe may blue men, maybe that's what I'm saying is.
Overblown our view of ourselves is in a way that just as easy it's that yeah I mean I I I came across this nineteen forty six report that I am saying this as a kid in one of those books that my mother gave me. Actually
nineteen forty six report, which pitted out someone with an abacus against an electronic calculator and he beat the electronic calculator. You know so they're at that point was well humans, still better than machines at calculating. Are you surprised today
a machine, can you know, do billion floating point operations, the second than you know your your puzzling for four minutes through one. So I answered
mean that I don't know, but I am certainly surprised, there's something to me different about learning
the system that enables the learning now see. Now you go into one of the deadly sins, but because of using terms
worthy, broadly yeah, I mean with so many different forms of learning and so many to function
I learned my way around the city I learned to play chess. I learned latin. I learned to ride a bicycle. All of those have you know a very different capabilities, yeah and if someone you know has a were in a you know. In the old days, people would write a paper about learning something MAO. The corporate press office puts out a press
weeks about how company ex has has is leading the world because they have a system that can, however, is a thing gets. What is learning
when I refer learning many things, but I suitcase word is a suitcase word blames, let's loosely there's a dumb system and over time it becomes smart. Well, it becomes less time
at the thing than its doing. Yes, my is your minuses. Unloaded word. Yes, what was done with a thing? It gets better performance and the some measure and some set of conditions at that thing and
most of these learning algorithms. Those learning systems fail when you changed
conditions just a little bit in a way that humans don't so right. I was that deep mind. The alpha go had just come out and I said
What would have happened if you'd give another twenty one by twenty one born instead of a nineteen by nineteen bodies failed totally, but are you
play. It would actually, you know well
We had a play and funny enough to look at the mines work since then they are present
they're presenting lot, algorithms. That would do that
would do well at the at the bigger board to their slowly expanding this generation. To me, there's a call
element there. It is very surprising to me that the even in a constraint game of chess or go that through self play by system playing itself that can it it can achieves
Super human level performance through learning alone, so look at so so easily.
Don't fundamentally did in search of that. You didn't you didn't like it when I referred to them,
mickey nineteen sixty one paper there in a second part of it came a year later.
They had self play on electronic computer at tik, tok toe, ok, it's not as but it learnt to play, tiktok go through solid. Lay that does not
I would to play optimally. What I'm saying is I
Ok, I have a little bit of a biased by I'd. I'd I'd find ideas.
Before, but only when they actually realise the promise that another level be like. For example,
the basis than you must go down with rockets where Iraq is for
on time, but do reusable cheaper august is very impressive and same way. I ok yeah. I would have not predicted for store where I was
this. This started and fell in love with AI. The game of go was seen to be impossible to solve. Okay, so I thought, maybe you know I maybe it'd be possible to
I have big leaps in a moore's law style of way in computation, it'll be able to solve it, but I would never have guessed that you can learn your way. However, I mean
the narrow sense of learning. Learn your weight to to beat the best people in the world at the game of go without human supervision, not studying again of exports
okay. So so that's just the size that using a different learning technique. Yes, arthur samuel in the early sixties, and he was the first person to use machine learning. Got rut had a program that could beat the world champion at chequers now, yes, so, and that at the time was considered amazing by the way
arthur samuel had some fantastic advantages. Yet do you want to hear at least two things one? He was at the the nineteen fifty six ai conference a new author later in life. He was at stanford when I was gretchen there. He wore a tie and a jacket everyday. The rest of us didn't and his delightful man the life from him, and it turns out claude shannon in a nineteen
if the scientific american article outlined the on chess playing outline the learning mechanism that author samuel used and and they had met in nineteen? Fifty six? I assume there was some communication, but I don't know that for sure, but arthur samuel has been and of the
it came, true engineer and getting reliability vacuum tubes and then had overseen the first transistor eyes, computers that I b m and in those days before you shipped the computer, you ran it for a week to seek to get early failures. So he had this whole farm of computers running
random code for hours and hours a week for each computer. He had a whole bunch of them. So he ran his chest learning program with self play on an I b m's production
I had more come computation available to him than anyone else in the world, and then he was able to produce a chest playing programme up a check is playing program that could beat the world champion. So, that's amazing that the question is what I mean surprised. I don't just mean it's nice, though, that accomplishment is. There is a stepping towards something that feels ah
more intelligent than before, and that question is that in your view of the world, because they want me than them
you mean I'm wrong, and I know that some of the question is: if we keep taking steps like that, how far that takes us, I will go to build a better recommend
systems are going to build a robot, nor will we solve intelligence. So
no I'm putting my bad arm, but still missing a whole lot.
I am, and why would I say that well in these games, they're all in one hundred percent information games but again, but the each of these systems have very short description of the current state, which is different from registering and perception in the world which gets mcnamara of X. Paradox,
Firstly, not saying that the chest somehow harder than our perception or out
in any kind of, even even any kind of robotics in the physical world. I definitely think is is way harder than the game of chess. So I'm I was always much more impressed by like
the workers of the human mind is incredible. Dehumanizing credible, as I believe that
It may be, I want to be a psychiatrist for the longest time. I always thought that's way more credible in the game of chess. I think the games chess is a. I love the olympics. It's it's just another example.
You must picking a task and then agreeing that a million humans will dedicate their whole life to that task, and that's the cool thing
that the human mind is able to focus on one task and compete against each other and achieve
weirdly incredible levels of performance. That's the aspect of chester super cool, not that chest in itself is really difficult. It is like the from us,
There is not in itself to me that interesting. The fact that thousands of people have been struggling to solve that particular promise is faster.
What can I tell you, my diseased in this way sure which actually is closer to what you're saying so as a child? You know I was building various. I call them the computers. They want. General purpose. Computer ice cube tray
I ask you: trade was on, but about our machines, and when I like to build was machines that could beat adults at a game they couldn't by adults, couldn't beat my machine
yes, that will you were like ass, fast, powerful, like this, that the way to rebel
by the way I'm did what was the first time you build something that had performed. You do remember like
Well, I knew how work those probably nine years old and I built a thing that the you take turns and taking it matches from a pile and
is the one who takes the last one on the one it doesn't take. The last one wins I forget, and those parties to build
out of wires. Males and little coils
plugging the number and, if you light bulbs and the one
though I was proud of. I was twelve when I built her a thing.
well done telephone switchboard switches that could always win,
at tic- and that was a much had a second design again was just- was no active components. That was just three positions, which is empty ex zero
and them and nine of them and light by one which which move it wanted next among the ema gun, move that either
magic in their creation. I was there, I didn't do it. I didn't see magic and robots. That's like! I also think that intelligent
is the a little but bit overrated. I think we can have deep connections with robots very soon and I will come back to connections robot sure, but but I do want to say I I don't. I think people too many people make the mistake of seeing that magic.
thinking. Well, it will just continue, but each each one of those is a hard fought battle for the next step, the next step. Yes, maybe the open question here is- and this is why
I'm playing devils advocate by often do when I read your blog posts in my mind, because I have like this eternal optimism. Is it's not clear to me,
do what, obviously the journalist you're like a given to the hype. But it's not obvious to me
these steps away we are from from, are truly transformational
understanding of what it means to build intelligent systems or how to build intelligent systems
I am also where the whole history of artificial intelligence, which is why europe deep grounding of this is is there has been an optimism for decades.
The optimism just like reading old optimism is absurd, because people were like this, as they were saying, things are trivial for decades. Since the sixties, they're saying everything is true, computer vision is trivial but,
I think my mind is working crisply enough to where I mean we can dig into it. If you want, I'm, I'm really surprised by whew. The things demands that I don't think, there's something that you get close to solving intelligence, but I'm not sure it's not tender at ten years away
What I am referring to is interesting to see when the engineering take that idea, the scale
and this and the idea was no fault people, ok, as it really is,
You me and in federal forget the press. Forget all those things just as a scientist as a robust. As you know, it was a surprise to you that
at scale. So we're talking about very large know will get okay. Let's pick one, that's the most surprising to you. Please don't yell at me! G p, t three:
it's ok! I gotta bring with about ok, they alpha zero. I half ago africa's there are four zero and then after fold, one into so
Our annual judge any of these kind of have this core of that not forget, usefulness or application, and so on, which you could argue for alpha fold like, as the scientist was doors surprising to you that it worked as well as it did. Okay. So, if we're going to make the distinction between surprise and usefulness and and and I'll have to explain what I would say alpha fold.
and one of the problems at the moment with alpha folders. You know it gets a lot of them right, which is a surprise to me because they're a really complex thing, but you don't know which ones that gets right, which then
it's about a problem that I've come out with a resuming the structure the protein. It gets a lot of those right. It's it's surprising. Room right has been really hard from that's a surprise how many it gets right. So far the usefulness is limited because you don't know which ones are right or not an thou. There come out with a thing in the last few weeks, which is trying to get a useful tool out of it and they may well do it. In that sense, the least off a folded.
Because you're awful too is different because now is producing data that are actually in a potential
revolutionizing competition biology like they will actually help a lot of people but you'd say prevent potentially revolutionizing. We don't know yet, but yeah, that's true yeah but they're. You know, but I I got your I mean this is okay, so you know what this is: gonna be so fun, so let's go right into it. Speaking of robots that operate in the real world
Talk about self driving cars- oh okay, because you do you have built robotics companies. Do it you're one of the greatest roboticists in history and thus not in space of a just in the space of ideas would also probably talk about that, but in the actual building and execution of businesses that make robots that
useful for people and the actual work in the real world and make money. You also sometimes a critical of MR la musk or less more specifically focused on this particular technology, which is on a pilot inside teslas.
What are your thoughts about us autopilot or more generally, vision, based machine learning, approach to semi autonomous driving that these are rob
They are being used in the real world by hundreds of thousands of people and will, if you
go there. I can go there, but let's not too much, which there is that say there on par safety, wise ass humans, currently
meaning human alone versus human plus rebecca so first, let me say I really like the car I came here in here today, which is twenty twenty one model, mercedes e4. Fifty I am impressed by the machine vision sooner other things
pressed by what it can do, I'm really impressed with the many aspects of it and am is able to
finland, as it does that it does the lane staff it up. It is it's working.
On the side of me telling me about nearby cars applies.
The saw you when I when I when I'm well I'm going in close something the park. I get this beautiful gorgeous top down view of the world. I am impressed up. The was due of how you
registered metrical. Well, you those like multiple cameras in this already produced the three sixty view country sixty view or you'll synthesized thoughts above the car I'm. It is unbelievable that I got
The current january, it's the longest, I've ever owned a car without digging it float it's better than me it's up on me and it together better. So I'm not saying technologies are bad or not useful, but here's my point. Yes, it's just it's the replay of the same movie. Okay, so maybe you've seen me asked this question before, but damn when
when, when did the first car go over fifty five miles an hour for over.
over ten miles on a
look freeway without trafficker, I'm driving completely autonomously. When did that happen,
does it seem you, and eighty is something that was a long time ago was acting nineteen eighty seven in in munich, munich at the bundesbank, so they they had a running and ninety seven. When do you think I'm gonna said he's gonna do this? When do you think will have the first car
DR coast to coast in the us hands off the wheel, hands off the wheel feet off the pedals coast to coast. As far as I know, a few people have claim to do it. Nineteen, ninety five that was comedy I didn't know,
That was the good yeah the they didn't claim. I did. They claim a hundred percent in a not hundred percent, not one hundred, but but
a few marketing people who have claimed one percent. That my point is that you know I I what I see happening again is someone sees a demo
and they over generalize and say we must be almost there, but we've been we've been working on it for thirty five years as as demos, but this is going to take us back to the same conversation
There are four zero. Are you not ok I'll see what I am, because I thought you get when I first our story interact with
the with the mobile ai implementation, tesla autopilot that have driven a lot of car. You know I have been in
there were driving car since the beginning? I thought there was no way before
sat and use mobile, I I thought there just knowing computer vision,
about there's no way I could work as well as working so
My model of the limits,
Computer vision was
I was way more limited than the actual limitation of out so out so example, one example surprised that while there was there was a credible the seconds. The second surprise came, one
and tat the throw away mobile. I am starting from scratch. I thought there's no way they can catch up to mobilise. I thought would mobile. I was doing
it's kind of incredible like the mod work and the annotation on mobile. I was stopped by a mom's treasure and and used a lot of traditional. You know hard fought computer vision techniques, but they're also did a lot of good sort of I like non resource stuff like actual awake. Just
good like where you do to make a successful product great gale. They all that stuff ass. I was very surprised when they first scratch were able to catch up to that that very impressive, and I thought the lot of engineers those involved. This is that was impressive
and the recent progress, especially under what would the environment under capacity of the what they were, what they're doing with the data engine, which is converting into the driving task into these multiple tasks and then do
miss edge case discovery when they're pulling back like the level of engineering made me rethink was possible. I dont I still
Add to that intensity, but I always thought is very difficult to solve autonomous driving with all the sensors with all the computation. I just thought is a very difficult problem, but I have been continuously surprised how much you can engineer. First of all, the data acquisition problem, because I thought you know just because I worked with a lot of car companies and they're there so a little a little bit old school to where I didn't think they could do this at scale like a w s style
data collection. So when test was able to do that, I started to think I guess what are the limits of this? I still believe that the driver, like sensing
and the interaction with the driver and like studying the human factor, psychology problem is essential. It's it's always going to be there and it's always going to be there even with fully autonomous driving. But I have been surprised what is the limit, especially of vision, based alone, how far that can take us so that my levels of surprise now? Okay, I can you, can you explain in the same way, you said
alpha zero. That's the homework problem. The scaled large is chess like who cares go with here's the actual people using an actual car and driving. Many of them drive more than half their miles using the system, so yeah they're doing well with with pure vision,
your vision and dimly and now no radar witches. I suspect that
can't go all the way, and one reason is what, without without new com,
is that have a dynamic range closer to the human eye, because even I have credible, dynamic range and we make use of that dynamic range in
it sir. I'm an order of magnitude, some crazy number, like that. The cameras don't have that, which is why you see that the the
I had cases where the sun on a white thing and the blinds are in a way it wouldn't blind a purse. I think there's a bunch of things to think about before you say this is so good. It's just going to work, ok,
and I ll come out from multiple angles:
a lot of time- a again
I have thought about these things that I know you.
writing a lotta, great blog post about it for a while. Before does the head
a pilot right. The you ve been thinking by autonomous driving for awhile from every angle. So so a few things you know in the? U s, I think that the
death rate from motor vehicle accidents is about thirty five thousand a year, which is an outrageous number not outraged.
The covert deaths. But you know there is no rationality and that's part of the thing people have said engineers say to me: well, if we cut down the number of deaths by ten percent by having autumn,
Striving that's gonna, be great, ever love it, and my prediction is that if autonomous vehicles kill more than ten,
People a year, though we screaming and horror, even though thirty five thousand people a year
killed by human drivers. It's not rational! It's a different set of expectations.
and that will probably continue so there's that aspect of the other aspect of it is that
When we introduce new technology, we
and change the rules of the game. So when we introduced cause first into our daily lives, we completely rebuilt our cities and we.
changed all the laws. Yeah jay walking was not an offense that was pushed by the car companies so that people would stay off the road so that wouldn't be deaths from pedestrians getting hit. We completely changed the structure.
Cities and had these foul smelling things you know everywhere around us anywhere. Now you see pushback it'll cities like Barcelona is really trying to exclude, cast, etc, and so I think that to get to self driving we will and large adoption is not going to be just take. The current situation take out the driver and put the
car doing the same stuff, because the end cases too many is an infant question. How many are fully autonomous, some trains systems? Do we have in the? U s and he d called them
fully autonomous. I dunno, because there usually is a driver but they're kind of autonomous right now. Well, let's just let's get rid of the driver. Okay, I dunno admit it. It's either fifteen or sixteen most most of them are in airports and there's a few that go about five to go about five kilometers out of airports, yeah and ah wendy
when one is the first fully autonomous trains system for mass transit expected to operate fully autonomously driver, I'm giving you a city and expect the operating twenty. Seventeen in honolulu
it's the labour, but they will get their bought by the way was originally gonna be autonomous. Here
The I mean they're, all very close to fully autonomous
yeah, but at the beginning the closest thing- and I hope I have often gone on fully autonomous train in japan- one that goes out to that.
Wake island in the middle of Tokyo bay? I forget the name of the and and what do you? What do you see when you look at that? What do you see when you go to a fully autonomous train in an airport? It's not like regular trains that are at every station, there's a double set of doors, and so that the door of the train and the door off the
The m after the platform is really visible in the japanese one because it goes out in animals buildings. The whole track is built
that people can't climb onto it. Yeah, so does the engineering but then makes the system safe and makes them acceptable? I think we'll see similar sorts of things happen in the. U s, what a surprise me
I thought wrongly, that we would have special
buzz lanes on one or one in the bay area that the left most lane
that it would be normal for teslas or other cars, the movement that lane and then say. Okay, now autonomous and have that dedicated lame. I was expecting move
to that in five years ago, I was expecting, would have a lot more movement towards that. We haven't and it may be, because tesla's been over promising. I saying that even calling the system fully self driving- and I think they may have been gotten there quicker by collaborating to change the infrastructure,
one of the problems with along whole tracking being autonomous. I think it makes sense on freeways at night for the trucks to go autonomously
but then is that how to get onto off of the freeway? What sort of infrastructure do need for that?
do we need to have the human in there to do that, or can you get rid of even though I think was ways to get there, but it's an infrastructure argument, because the law
tale of cases is very long and the acceptance of it will not be at the same level as human drivers, though,
I'm with you still now was due for a long time, but I am surprised how well humming cases
machine learning and vision based methods can cover the sort this around trying to get get out his them. I think there's something fundamentally different with vision
These methods and test autopilot in any company, thus trying to do the same
I well I'm not I'm not going to argue with you, because I I went with speculating yes, but I am you know my gut feeling tells me it's going to be, and things will
will speed up when there is engineering of the environment, because anyway happened with every other technology. I am. I am a bit about you, but I'm a bit cynical that infrastructure, which relies,
government to help out in these cases is, if you just look at infrastructure in all domains,
it just government always drags behind and infrastructure there, so many
Well in this country in the future, so yes in the in this country,
Of course, this many many countries that are actually much worse than infrastructure earlier does not only the much worse than with someone like high speed rail, the other countries that are much better. I guess my question is like, which is at the core of what I was trying to think through here and ask. Is the car heart is the drowning problem as it currently stands
you mentioned that we don't want to just take the human eye to procure the human was doing, but if we were to try to do that, what how hard is that problem, because I used to think is way harder
guy I used to think it's with vision alone. It it'll be three decades for dec.
right, I don't know the answer to this thing, I'm about post, but I do notice the long away to eighty here and the bear him which largely has concrete surface rum blacktop surface the white lines that are painted there now have black boundaries around them.
And my lane drift system in my car would not work without us black boundaries.
I dont know whether they thought of doing it to help the lane drift, whether it is an instance of infrastructure following the technology, but it
but my car would not performers well without a change in the way they paint. Unfortunately, really good, laying keeping is not as value
like its orders, the magnitude more valuable to have a fully autonomous system like, but but for me when keeping is really
helpful way or the other, but you would
Pay ten times like the promised. There's not financial like it doesn't make sense to do too are revamp the infrastructure to make link keeping easier
It does make sense to revamp infrastructure. I see what you have a large fleet of autonomous vehicles. Now
change what it means to own cars. You change the nature of transportation mean, but that that, for that you need autonomous vehicles. Let me ask about way more than
then about your chances do to ride in a way more self driving
car in there. I, if you'd call them some writing, but
Well I mean I, I wrote in one before they called waymo. We are still at annex that is currently the resume bigly another surprisingly, but I didn't think it would happen, which is they have no driver.
The aim in chandler in John there's honour, and I think that figure doing then asked than his work, but there they are.
very little know, although I do an annual check up on this. So
of late last year, there were aiming for hundreds of rides a week, not thousands, and
There is still no one in the car, but certainly safety people in the loop and it's not clear how and in what the ratio of tasks to safety people is.
I wasn't obviously they're, not gonna percent transparent about there's. No none of them were hundreds of transfer. This everyone transmitted by ad
It's the way there
I want to make the fairly but they're saying, there's no telling operation.
like there. I mean to
If and that sort of fits with with youtube videos, I've seen of people being trapped in the car by a red cone on the on the street, and they do they do have rescue vehicles that come and then a person gets in and drives it yeah, but isn't it incredible to you? It was to me to get in a car
are with no driver and watched the steering wheel turn like first somebody who has been studying these. Certainly the humans.
at time of vehicles for many years, and you ve been doing.
and for way longer that goes incredible to me that this is actually could happen. I don't care if the scales
The car is not a demo. This is not. This is me, is it
as of right now. The only argument I have is that people make interpolations from that interpretations that are in appearance here. It's dumb and you know it's just you know, we've solved that no, we haven't yet and that's my argument. So I'd like to go
it's just you or you keep a list of predictions on your amazing blog post it'd be fun to go to them, but before then, let me ask you about this.
yeah.
You have a harshness to you, sometimes in your criticism
as hype, and so like cars, people extrapolate like you said, and they they kind of
by into the hype and then they they can start to think that
the technology is way better than it is. But many ask em, maybe a difficult question shunned. Do you think if you look at history of progress, don't you think to achieve the quote impasse, the boy? You have to believe that all possible steps absolutely but his his is his. His two great runs great, unbelievable,
Nineteen o three first human power, your human, now in a heavier than their flight. Yet nineteen,
sixty nine we land on the moon at sixty is I'm sixty six years old in my lifetime that span of my life and we have barely getting on flying animal what it was
defeat the length of the first flight or something the landing on the moon, unbelievable, yet plastic. But that requires
by the way, one of the Wright brothers, both the number one
and believe it's impossible like a year before right, so not just possible soon but like ever so so
you know how important is that the believe and be optimistic as well. I guess I'll wait and innocent point it's when it goes crazy when men, when I you know, you said it wanted, what was the word used, my bad harshness harshness? Yes, I just get so frustrated. Yes, when, when people make these leaps and tell me that I'm then I dont understand right. Yeah.
this. Just from I robot, which I was co founder of yeah. I don't know the exact numbers now cause I haven't at ten years since I stepped off the board, but I believe
as well. Over thirty million robots cleaning houses
one coming. There now has lots of other companies. Just was
Crazy idea that we had the believe in
thousand and two we released it yeah. That was why we had we had the you.
I believe that it can be done to get them ask about this, so I robot one of the greatest robotics companies ever in tourism may fetnah.
Building a robot that actually works in the real world, probably the greatest revise company ever you're them
co, founder of it if it
ride me brooks of today thought through the reindeer back then, but would you tell him because I, since that, would you put him on the back and say: will you
who is going to fail, but go it anyway. That's why I'm referring to with the harshness you ve accomplished an incredible thing there, one or several things will talk about what we would like this, I'm trying to get at that line, nor its when my harshness harshness of people who are not doing it, who claim
it's just while this shows that it's just gonna happen. But here here's the thing this show where you have that harshness.
For you on to an no no
the different harshness. Now it's it's different and with delight, I think spacex is amazing.
companies. On the other hand, you know I in one of my bumpers, I said: what's eating what had I said: if space exe
vertical landing rockets. It had been done before and grid fins had been done since the sixties, every soyuz has them and reusable space and dc ex reused those rockets that landed verdict
and the whole insurance industry in place for rocket launches thought. Oh such an infrastructure that was doable, it took a great entrepreneur at great personal expense. He almost drove himself in a bankrupt, doing a great belief.
do it, whereas hyperbole there's a whole bunch more stuff, it's never been thought about, never been demonstrated. So my estimation is
I pulled it was a long walk, long lie further off
by an and if I gotta criticism of of of alarm, it's that he doesn't make distinctions between when the technologies
coming along and ready, and then you go off and and off about other things,
then the little gone compete about trying to do and
Is this where I I I understand what you're saying I tend to draw a different distinction? I I have a similar kind of harshness towards people who are not telling the truth
are basically fabricating staff to make money, or do you believe what he says?
This thing I'd want me that on a very important difficult, because I think that, in order to fly in order to get to them,
when you have to believe, even when
people tell you you're wrong and most likely you're wrong, but sometimes you're right. I mean that does the same thing
I would have thought about it. I think that's an interesting one. I was especially when I was, and am I d and just the entire human factors, and thereby the community
We were very negative towards elon. It was very interesting for me to observe colleagues at mit. I wasn't sure what to make of that, as was very upsetting to me, because I understood without where that's coming
from and I agreed with them in a kind of almost felt the same thing in the beginning. Until I can I open my eyes and and
realize there's a lot of interesting ideas here that might be over hype. If, if you focused yourself,
On the idea that you should call a system for self driving when it's obviously not autonomous, fully autonomous you
going to miss the magic. Oh, you are gonna miss the magic, but at the same time there are people who buy it.
literally pay money for it and take those words as given. So that's why I haven't so the a
words is given is one thing I haven't actually seen people that use autopilot. I believe that that behaviour is really important, that the actual action so like this is to push back
and the very thing that your fresh state about, which is like journalist, general people, buying all the hype and going out.
In the same way, I think there's a lot of hype about the negatives of this too,
the people buying without using people use the way is to work. This. Was this open my eyes actually at the way people use the product is very different in the way they talk.
On this issue with robotics. Did everything everybody s, dreams of how a particular product may be useful to sew on this and then, when it meets reality is a lot of fear of robotics. For example,
the robots tat somehow dangerous in our skies. Things when you actually have
What's in your life, whether in the factory or in the home, maybe I'd better. That's going to be that way.
For the your perceptions of it are going to be way different, and so my just tension was like years in it innovator of like what is it sets a super cruise from cadillac was super interesting too. That's a really.
system theirs, and if we should like the excited by those innovations, don't let can. I did not know what really annoyed me recently thrilling.
For me that the press
friends of mine on facebook at guy these
it is in this space game
I like doing that. The first really pissed me off. I must say I applaud that. I applauded it's. The taking
I'm not least from the people who are doing the things, but the know that I keep having to push back against unrealistic expectations when these things can become real there. I this is interesting because there has been a particular focus for me, as autonomous time was driving yours prediction of one certain must also be hit.
Their several things to be said there that I owe owes I thought about the biggest whenever he said. The most obvious does not want to me as a
Then they can not inside the system is obviously the unlikely to hit. Those other two comments I want make one. He legitimately believes it and too much more importantly, I think
that having ambitious deadlines, drive people to do the best work of their life,
when the odds of those deadlines have very little to point
not too many, unless you I'm just saying so, there's a line. There is what you have to have a line because over I extend damage
demoralising yet, but I will
that is an additional thing here, that those words also drive the stock market,
Yeah, and you know we have because of the way that rich people in the past have manipulated the robes through investment we have we have developed.
About what you know what you're allowed to say. Let us- and you know
an area here which is
I I I tend to be,
maybe naive, but I attend to believe that
engineers, innovators, people like that they're, not there. My they don't think like that, like manipulating the price of the stock price, but as possible them, I'm certain it's possible that I'm wrong. I it's a it's a very cynical view of the world, because I don't think
people then run companies the like, especially regional founders,
they are nothing. That's the intent, I'm saying it to eventually ca you. I am yet you for you fall into that kind of behavior.
At an identity, crisis
it was designed falling into that intent is just to you also have to protect investors in this market.
After you have. First of all, you have an amazing blog that people should check out, but you also have this. In their blog a set of predictions such a cool idea
how long ago he started three four years ago, was them generally first, two thousand and eight teen eighteen here am. I made these predictions and I said that
every generally. First, I was gonna check back on how my predictions that setting right thanks for thirty two years always sees that third year.
Eternity is because it's cool that'll be January. First, twenty fifty here I'll be. I will just turn ninety five right. You know, sir, I answered
If people know that your predictions, at least for now, are in the space of artificial intelligence yeah, I didn't say I was
make new predictions. I was just going to measure this set of predictions that I made because yeah it was sort of. I was sort of annoyed that everyone could make pretty,
since they didn't come true and everyone forgot, so I should hold myself to a high standard there for us to just putting years, unlike date. Rangers on things is the good thought exciting
like, unlike reasoning, your thoughts out, and so the topics are artificial intelligence, autonomous vehicles and space as one if it took a just,
go to some extent are maybe from memory I can just mentioned to you some. Let's talk about self driving cars like some predictions that your particularly proud of, or a particularly interesting from flying cars to the other element here is like how widespread the location where the deployment of the autonomous vehicle
as a result, just a few fun ones. Is there something that jumps to mind that you remember from the predictions? Well, I do. I think I did put it in there that there would be a
dedicated self driving lane on one or one by some year, and I think I was over optimistic on that one yeah exit yeah actually do remember that, but you are, I think, you're mentioning like difficulties at different cities
Cambridge Massachusetts, they think, was an exempt yeah by contain bridgeport port, even within cambridge point for a number of years, and
you the roads and manner on
Demi anywhere is a human drivers? Incredibly frustrating when you start to put an end,
We'll drive the wrong way on one way street there. It's just do your prediction was driverless taxi services operating on our streets in cambridge port.
massachusetts in two thousand and thirty five yeah, and that may have been too optimistic.
You think so I got a little more pessimistic since I made these internally on. Some of these things
what.
Can you put a year to a major milestone of deployment of a taxi service in in a few major cities like something where you feel like yeah so with vehicles? So let's, let's take the grid streets of san francisco north of market, okay, okay and relatively benign environment. The streets are wide,
the major problem is some delivery truck stopping everywhere which to make things more complicated taxi system. There were somewhat designated pick up and
well paths, unlike with open, lift wakens where get to any place.
I was all figure out how to get in there.
I was still a few years away. I e not idle.
in that area. So I see you locked the self driving how come these cars multiple multiple once every day, now, crews luke's less often way mon?
the time and different and different ones come under when there's always a driver, there's always a driver.
them. I have noticed that some sometimes the driver does not have the authority to take over without talking to the home office, because they will sit there waiting for a long time and clearly something's going on where the home office is making a decision,
A so the you know, and and so you can really see where they've got their hands on the wheel on up and and it's the instrument.
Revolution time that tells you gives you some clues so what year, do you think what's your intuition? What date range are you currently thinking san francisco would be a timeless taxi service from any point agent, at any point b, without a driver.
are you a I think, tank, if not for years now, thirty years now, certainly not ten years from now on be longer,
Let us have a market way longer and I've unless this re engineering of the roads by the way was the biggest challenge the commission a few, is it? Is it the delivery trucks? Is it
the cases the computer perception. Well, you know right now is a case to restore outside my house a few weeks ago.
P m on a friday night, it was getting dark was before the solstice museum,
Cruise vehicle come down the hill turn right and and stop dead, covering the crosswalk while accepted cause. There was a human just two feet from it. Now I just glanced. I knew what was happening. The human
was the woman was at the door of her car trying to unlock it with one of those things that yeah you know when you don't have a key. Yes, that car thought
she could jump out in front of me any second year as a human. I could tell no she's not going to jump out she's busy trying to unlock her she's lost her keys, she's trying to get in the car, and it it stayed there for until I got bored near
And so that the union driver- and they did not take over but here's the kick enemy
a guy, comes down the hill with a stroller. I assume there's a baby in there and now the crosswalk is blocked by this cruise vehicle. What's he gonna do cleverly, I think he decided not go in front of the car
He went, but he had to go behind it had to get off the cross walk out into the intersection. Were the pushes baby around this car
which was stop there, and no human driver will stop there for that. Like a time,
what about a mile away and that's it-
one of my pet peeves, that safety is being compromised,.
for individuals who didn't sign up for having this happen in their neighborhood. Yet, but now you sent us in an case, but
Yet while in general, not a fan which
anecdotal evidence for stuff like this is one of my biggest problems, with a discussion of autonomous vehicles and in general people that criticize them or support the
he has kids using it. So let me limber, I got you, you know, you know your question is: when is it going to happen in San francisco? I say monsoon, but also going to be one of them, but when
where it is gonna happen is in limited. The mains camp of various sorts data commute
these were the other drivers are, are at and not arbitrary people, and then people who know about these things. They have been warned about them and at the loss of these, where it's always safe to stop dead. Yet you can't do that on the freeway that I think we're going to start to see, and I may not be shaped
like you know, current cause I may be in things like in may. Mobility has those things and various companies heavily see. I wonder if that the compelling experience to me as though, is important. It's not just about automation. It's about creating a product that, like that makes, is not just cheaper but makes your this fund the ride on one of the most one of the least fun things for a car that stops in like weights. There's something deeply frustrating for us humans for the rest of the world to take advantage of us as we wait, but I think about you know not you, as the car
but someone who's in their eighties in an earlier. In order a time a village kids have said, you're not drive and any more, and this gives you the freedom to go to the market, but that's a hugely beneficial thing, but it says
very few orders of magnitude, less impact on the world is not has just a few people and in a small,
me too, using cars as opposed to the entirety of the world. Ah, I like that the the first time that a car equipped with some version of a solution to the trial,
The problem is that was, and I am therefore going on in my one item. Well, I'd find my lifetime as absolutely fifty four
I asked I ask you win when, if you had to decide which person shall I kill now, you put them
brakes on the brake as hard as it can yet
I mean in that position. It is that you know I do think autonomous vehicles or semi autonomous vehicles do need to solve the whole pedestrian problem that
has almost at the trolley problem within it, but it's not yet well so he's
taka battle in one of the articles or blog posts that I wrote his his his and people have told me. I one of my coworkers has told me he does this: he he
watches autonomously, driven vehicles and pedestrians will torture them. Now
once they realise that Europe's putting on foot off the curb makes the carthage?
but they might walk into the road kids teenagers.
we don't meddle with time. They will
by the way? What am I as the whole, another discussion might manage with robotics is hr. I human
Robot interaction. I believe that robot they interact with humans will have to push back like they can't be bullied because there's crazed very
and compelling experience for the humans yeah. Well, you know waymo before it was called waymo discovered that you know they had to do that at a four way intersection. I hit the that the nudge forward to give the cue that they were going to go cause, otherwise the other drivers would just
beat him over time. The! U cofounded! I robot mentioned one of the most successful robotics companies ever. What are you most proud of? That?
company and the approach you took two robotics,
Will I lose something I'm quite proud of there, which may be a surprise, but I was still on the board when this happened was march, two thousand and eleven and and we sent robots to Japan and that we used to shut help shut down the fukushima.
The fukushima daiichi nuclear power plant, which was everything was well I've been there since I was there in two thousand and fourteen to the ama robots. Some of the robots were still where I was. I was proud that we were able.
To do that, why will we able to do that? And- and you know people have said well- you know Japan is so good at robotics
it was because we had had about sixty five
Hundred robots are deployed in Iraq and afghanistan telling opt, but with intelligence dealing with roadside bombs.
So we had, I think it was at that time, nine years of in field experience with the robots in harsh conditions, whereas the japanese robots, which were you know, getting up just get. This goes back to what annoys me.
Getting the height look at that now that humble, robotic and walk the features gear couldn't do thing because they went the ploy
but we had deployed in really harsh conditions for a long time.
and so were able to do this
I very positive in a very bad situation. What about just the same?
More and more people are now one of the things that I robust
It is the roma.
Vacuum cleaner. What about the simple, robot?
that that is the wrong birth concord. Simple, that's deployed in tanzania,
is if, in tens of millions of homes, what do you think about that when I make the joke? I start my wife is a pure mathematician and turned into vacuum. Cleaners,
salesman. So if you gotta be an entrepreneur, be ready for that,
I knew anything. But I was, you know those up, those where he was and I got posed for
not too many years ago, and I was the only one who had a from the nineteen nineties is
No one in the company had its. I went and went through. My email, em
and it reminded me of you, know the the joy of what we were doing and and what what was I doing? What was I doing at the time we were building and building?
the roomba one of the things was. We had croix in an incredibly tight budget because we wanted to to put it on the shelves at two hundred dollars. There was another home cleaning robot at the time. It was the electrolux trilobite which sold for two thousand years until
that was not going to be a consumer product, so we had reason to believe that two hundred dollars was a was a thing that people would buy at that time. But that meant we had. You know, that's that's on the shelf making profit. That means the cost of goods has three men
so I find all these amounts of meat are going to be in them. Taipei, four, a m! I t meeting and I'd stay a few days ago. The shoe and talk these little tiny companies lots a long time coming outside of that. The assembly taiwan semi combat, I once and combat the menu-
action corporation, which let all these little companies be fabulous. They didn't have to have their own fab, so they could innovate and they were they were building. The innovations were built, stripped down. Sixty eight, oh two's, sixty attitudes. What was an apple one, get rid of half the silicon still have it be bible, and I had previously got some of those for some earlier fail
products of of I robot, and I'm that was done in hong kong. Going to all these countries are built in other than that.
The won't gaming current since there were these handheld games that you would play an all, but they cards as we had about a fifty cent budget for computation, so man tracking from place to place
Looking at the ships looking at what they'd removed or are they interrupt the interrupt him when he is too weak for our general purpose, so I was doing deep technical detail and then I found this one from a company called wind bond which had
I cannot forgotten that had this much ram, it had five hundred and twelve bytes of ram and it was in our budget and it had all the capabilities we needed. Yes, so
If you excited- and I was reading all these emails cause, all my family's, so did you think, did you ever think that you guys can be so successful like eventually this company would be so successful? Did you could you possibly have imagined and know we never did think we'd had fourteen failed business models up to two thousand and two. We had two winners same year.
Ah, no, and then you know we, I remember the board. By this time we had some venture capital in
The board went along with this building some robots for aiming at christmas, two thousand to market and
we went three times over what they authorized and got seventy thousand of them and solemn all in that first great released, released on september. Think- and I will also by christmas. So it was that.
so we ve got see by the african union figures not take over the world? Why is this.
so a lot of amazing robotics companies have gone under over the past few decades. Why do you think
so damn hard
to run a successful one is it does about? Its company was a few things. One is expectations of capabilities by the founders rub off
it's the foreigners, not the consumer and the founders yeah expectations. What what can be delivered sure mispricing and what a customer thinks is. A valid price is not rational, necessarily yeah and expectations of customers and.
Just the sheer hardness of getting people to adopt a new technology and I've suffered from all three means have had more failures and stresses in terms of companies have suffered from all three so
They think one day there will be a robotic company and by robot
company. I mean where your primary source of income is from robots
There will be a trillion plus dollar company and is so what come? What would that company? Do? I can't in a because I still staring robot companies
I'm not making any such
actions. In my own mind, I'm thinking about a trillion dollar coming and by the way I dont think in the nineties,
was thinking than apple. Would ever be a trillion dollar company, so visa visa very hard to predict what society interrupt, but
you could. I can have a vision in a small way
in vision, in a small way that I see that there be robots in the home at scale like room by, but more
That's trillion dollar right- and I am, I think,
the the real market pull for them because of the demographic inversion. You know who's who's going to do the stuff for the older people and there's too many you know, I'm I'm leading here
Don't be too many of us, but we don't have cable enough robots to to make that economic argument is point. Do I expect that that will happen? Yes, I expect the will happen, but I gotta tell
We introduced a member in two thousand and two and I stayed another.
nine years will always try.
And what the next homer about would begin still today, the primary product of twenty,
it's late. Almost twenty years later, nineteen years later, the prime product is still a roomba, so I robot hasn't found the next one. Do you think is possible for one person the garage to build it versus, like Google,
Launching google self driving car that turns into waymo, you think, is pot like this is almost like what it takes to build a successful robotics company. Do you think it's possible to go from the ground up or is it just too much capital investment yeah? So it's very hard to get there without a lot of capital and we're starting to see you know of the chunks of capital for some
box. Companies notice series bees. So one yesterday for eighty million dollars, I think it was for coherent, but it can take real money.
get into these things, and you may fail one way, I certainly failed rethink robotics hum and we all
whilst one hundred and fifty million dollars in capital, so ok rethink, robotics,
another amazing about his company cofounded? So what was
the vision there was the dream and what what are you must proud of what we think or box? I must proud of the fact that we got em robots out of the cage in factories that was safe,
we say for people and robots to be next to each other, that has a robotic arm, robotics homespun able to pick up stuff and interact with humans. I met the humans, could return them without running code.
and and now that sorta become an expectation for a lot of other little companies and big companies are advertising them into it. As both an interface problem and also a safety problem, yep yep, so I'm most proud of that- I complete I let myself be-
out of what I wanted to do, and you know you've always got you. I can't replay the tape and I can't replay it- maybe maybe a I you know if I'd been stronger on- and I remember the day- I remember the exact meeting and can take me through that meeting, yeah and so I'd said
what I'd set as a target for the company that we're going to build three thousand dollar robots with force feedback, and that was safe for people to be around. While that was my goal,
We built, I was so style in two thousand and eight, and we had prototypes built of plastic plastic gearbox system.
another three thousand dollar lifetime of three thousand dollar.
was saying we're gonna go after not the people already have robot arms in factories. The people who.
Never have a robot come we're going to go after a different market, so we're going to have to meet their expectations and and so we're going to build it out of plastic,
It doesn't have to have a thirty five thousand mile lifetime. It's going to be so cheap that it's opex, not topics, and until we had we had a prototype that worked reasonably well.
But the control engineers, what complaining about peace plan to give up a beautiful little planet for gearbox, but we can
chemical series. Alas, the actual, as we men and women there we can measure forces,
new when we hit something they control
you need to think about this. This is talk ripple, because these plastic is not great gears and this ripple
and trying to force control around this ripple is so and
I'm not gonna name names, but I remember one of my mechanic
engineers and we'll just build a metal gearbox with spoke is an ill. Take six weeks will be done, problem solved
Two years later we got to get the spur gearbox waiting. Now I'm wee wee wee castro,
is that every possible way we could near? But now the price went up to and then
see all the times and why we have to have two arms, not one arm, so at first robot product baxter, now across twenty five thousand dollars,
and the only people who gonna look at that were people who have arms in factories, because that was somewhat cheap of two arms and patrick, but they were
We used two point, one millimeter reproduce every reproducibility of motion and certain velocities, and we I kept think. But that's not what we're giving you you don't need position repeat. The building useful
control like a human, doesn't know. Oh no, but we want. We want that repeat ability. We want that repay the belly as all the other about.
That repeated bellamy. Why don't you have that repeat about so he can keep. I forced controls and grabbed the arm
You can never well, you can move around, but but suppose you can you see that? Yes suppose you want to yes suppose this. This thing are in a precise thing: I've got a fit here in this right and and opposition position control. You say,
you'll une you have fixed should where this is. You know where this is precisely and you just move opening up, and it goes there if forced control, you would do something like
Slidell over. Here, though, we feel that and slide in there and that's how a human gets pretty flat, precision yeah. They used force feedback, yes and get the things to make robin just go straight to it. You couldn't convince.
Compliments our customers, who were in factories and were used to thinking about things a certain way and they wandered one wonders so then we,
ok, we're gonna, build an arm gives do that, then
We ended up building a thirty five thousand dollar about with one arm with m o. What are they called since I sort up,
made by a company whose name I can't remember right now, but it's the name of the game and and but it's it's got talker
Well, isn't it so now those an extra two years of solving the problem of doing the forced with the torque ripple, so we had to do the the thing we had avoided anna for the plastic gearboxes. We ended up having to do that. Robot was now overpriced.
And they in those your intuition from the very beginning kind of it. This is not europe
being a door to solve a lot of browser that you're your revenge. You gonna to solve this problem anyway,
and also I was aiming at a low price to go into different milo price, that that day,
Have really hours and hours should be amazed me. I think we could have done it for five, but you know you said took them.
setting the golden too far
why? Why would you say that company not failed, but one under we had buyers and I'm.
There's. This thing called the committee on foreign investment in the: u s serious and that had previously been invoked twice around, where the government-
Stop foreign money. Coming into: u s: company, based on the fence requirements. We went through due diligence multiple times, we're gonna get a quiet, but every consortium had chinese money in it and all the bankers would save us,
in know, I'm get pacifists and moon vessels go away, and then we had two buys whence about to run money to buyers.
and one used heavy handed legal stuff with the other one said they gonna, take it and pay more dropped out when we were out a cash and then what the assets at one thirty of the price they had offered a week before.
there was a tough week. Do you does it hurt to think about that?
An amazing company that didn't.
yeah like I robot didn't find a way yeah. It was tough
I said, I was never going to start my own company. I was pleased that everyone liked what we did so much that the teams with the theme was hired by three companies within a week. Everyone had a job in one of these free company. Some stayed in their same desks because the camp another company came in and rented the space. So I felt good about people not being out on the street, but sub back
as a screen with a face. What that's the revolutionary idea for a robot manipulation, though for robotic arm or something
What position did you get well first. The screen was also used during code list programming. Were you taught by demonstration that showed you what its understanding of the task was, so it had two roles and some customers have
isn't it, and so we made it so that when the robot was running, it could be shown graphs of what was happening and not sure the other camp. Neither people and some of them surprised me who they were saying.
this. One doesn't work as soon as the old one we like the human lying, so those mixed bag. Where do you think that.
I I m kind of disappointed whenever I talk to to revise this the best robotics,
for the world. They seem to not want to do the eyes had a thing that make they seem to see it. As the machine is opposed to machine, they can also have a human connection. I'm not sure what to do with. It seems like a lost opportunity. I think the trillion dollar company
we'll have to do the human connection very well, no matter what it does get angry ghastly, ridiculous question shore
give a ridiculous. I do think
maybe by way of asking the question, then the first mentioned that you kind of critical of the idea of the touring tests as a test of intelligence. Let me first ask this question: do you think we'll be able to build
ai system, that humans fall in love with and it falls in love with the human like romantic love. What we've had that with humans falling in love with cars, even back in the fifties, was a different love.
Well, I think I think, there's a lifelong partnership or you can kill me
and grow like
think what a long way from that
I think we're a long long way.
Income and blade runner was, you know, had the timescale totally wrong? Ah yeah, but the salt, as it did to me honestly, the most difficult part is the thing you said with the marek's paradox is to create a human form.
direct and pursues the world, but if we just look at a voice like the movie, her we're just like an alexa tat voice, I didn't.
think we're not that far away well for some for some people- maybe not, but I hate in I I.
In our view, means as we think about the future. We always tries to enlist the premise of most science fiction movies. You got the world
This is today a change one thing, but that's not. How am I said the same with itself
in cardiff, we change. One thing in all: you do is everything changes yes grows together. So, surprisingly, might be surprising to your mind, not I think the best movie about this stuff was bicentennial man
and what was happening where it was small him, but what was happening it as the robot was trying to become more human. The humans were adopting
technology, the robot and changing their bodies near so there was a convergence happening in as citizens.
we will not be the same in a way already talking about genetically modified a baby sit over there,
more and more staff happening around that we will.
We will want to modify
I was even more for all sorts of things we put out
the technology in our bodies to improve it. You I've got.
I've got things in my ears so that I can sort of hear you,
always modifying body. So so
it's hard to imagine exactly what it will be like in the future, but on the torrent s side, do you think the forget about love for second, let us talk about just like
ilex surprised actually, as avoided to be what is the interviewer for the likes of whatever that's in two days.
Their idea is the success, looks like a person wanting to talk the nice system for prolonged period atomic twenty minutes.
How far away our we and wise difficult to build any system with which you'd want to have a beer and talk for an hour or two hours like
not for the check the weather or to check music. But just like the talk, his friends
yeah. Well, you know we saw we saw and weizenbaum back in the sixties, with his program eliza and being shocked at how much people would talk to elisa. I I remember we are in the seventies typing stuff to advisers.
but it would come back with you know. I think right now, and this is the thing that
yeah amazon's been trying to improve with alexa. There is no continuity of of of of topic does not account refer to what we talked about yesterday and it's not
same was talking to a person with seems to be an ongoing existence, might changes which
a moment together they last in our memory together, yeah,
but there's none of that and there's no sort of intention of these systems that they have any.
Goal in life. Even if it's to be happy, you know they don't they don't even have a semblance of that. Now. I'm not saying this can't be done. I'm just saying. I think this is why we don't feel that way about.
well that's that's how I am I'm sort of a minimal requirement. If you want the sort of interaction you're talking about it's a minimal requirement, whether it's going to be sufficient, I don't know we haven't seen it yet. We don't know what it feels like. I I tend to be ident to think it. It's not
as difficult to solving intelligence, for example, and I think is the tool bar in near term, but on the turing test. Why don't you think the turing test is a good test of intelligence, our eye, because again the trick during
I read the paper during wasn't saying this is a good test. He was using that as a rhetorical device to argue that if you can't tell the difference between a computer and a person, you must say that the computers thinking because can't tell the difference when it's thinking he he can say something different and what it has become as the sort of weird game of fooling people.
So back at the year I live in the late eightys we had. This thing still goes on hold
an olympics. One of the events we have one year was.
the original imitation game as turing talked about because he starts by saying: can you tell whether it's a man or woman? So we did that at the at the lab. We have, you know, you'd gone type and the thing
come back and you had to tell whether it was a man or a woman and the the ah one of the one of the one of the one man came up with a question that he could ask of which was
always a dead, give away of whether the other person was really a man or woman in what he will ask them. Did you have them green plastic toys soldiers as a kid? What do you do with them?.
and a woman, a woman trying to be a man would say I lined them up. We had was we had battles and the man just being announcer. I stomped on when I burned the sir yeah, that's what that's what the turing test the turing test
with computers as harmless. What's that trick question where that that's? Why that threatens that order, it that's what evolved into this we're this. Nevertheless, conversation not formulated. The test is a pretty fast amy challenging dance. That's really hard pressed to make conversation were not possible
test is a is the more intuitive illustration how far away we are from solving intelligence than a computer vision is hard. Computer vision is is harder for me to pull apart, but will
language with conversation? You could see his languages so human, we don't so human wiki would console console. Clearly I see it shit. You mentioned something I was going to go on. Often, okay, and I mean I have
I to ask you do you- are the head of csail Allah for long time, your time you're owner
to me where I can tell you you're, like one of the great said mit. So what was that time like what
and an plus you with the I
friends with, but you knew men ski another other folks thereof,
legendary I've people of which your one so was
time like what what our memories that stand out you from that time from your time at mit from
laugh from the dreams the ira lab represented to the actual
a revolutionary work. Well, let me tell you first of disappointment in myself. You know I've been researching this book and so many of the players you know. Well,
active in the fifties and Sixtys. I knew many of them when I oughta
I didn't ask him all the questions now I wish I had asked I'd sit with them and arab thursday lunches, which we had a faculty lunch, And- and I didn't ask him so many questions that now I wish I had asked you that question cause cause. You wrote that
You wrote that your fortune to known shoulders with many of the great
founded ay, I robotics and could be the size and the world
wide web? He wrote that
big regret. Nowadays that often have questions for those who have passed on, and I didn't think to ask them any of these questions
right, even as I saw them and said hello to them on a daily basis. So may be also another question I wanna ask: if you could talk to them today or question, would you ask or questions? Would you ask I will licklider I will. I will ask him, you know he had the vision for
Humans and computers working together and he really found that top her and he gave the money to mit, which started project mac in nineteen sixty three
and- and I I would have talked to him about what what the successes were, what the failures were, what he saw as progress etc, and I would have asked him more more questions about back cause. Now I could use it in my book.
It has lost its last forever, olano motivational last. Ah I,
I should have asked marvin why he he in Seymour, papert
the council had unusual networks in nineteen sixty eight and they book perceptions because Marvin's phd thesis was on your own networks, and I heard you
since the day he was destroyed the field,
Probably do you think he knew that the effect that book would have
Although the arms and negative theorems near
So here s the well, that's the way of life were still this kind of tragic that he was born.
The proponent and the destroy of neural networks. Yeah,
is there. Other memories stand out
from the thereby.
In the air. I work at mit, wouldn't we ear but you'll be most.
Specific. Well, I mean like a statue magical place, any means the lead, but also heartbreaking. That.
You know what's going on and facebook like did mine and so on so much of the tower, and you know it doesnt stay necessarily.
For long periods of time. In these and in this universe
oh yeah, I'm in the sun,
companies are more guilty them. Others, a pang, fabulous.
our answer to some of the highest produces and then system.
I hear from him again in my lab, give public talks and walked away and it sort of like collecting, collecting hollywood star.
or something, however, make movies anymore. I alone am yeah, that's tragic as their vision,
openness to the university setting. We do research to love
What the innovative ideas and speed like publication of those kinds of things yeah. You know- and you know, there's the publication of that amount from you know. Although these places say they publish or there's pressure and I, but I think for
since in our net man, I think.
Google buying those eight or nine robotics company was bad for the feel. Like those people away, I didn't have to make the countries succeed anymore, locked them
wait for years and then I'm sort of all printed away yeah. So how do you have hope from eighty four m for mit yeah? Why shouldn't I, while I can be harsh and say that
I'm not sure I would say mit is leading the world. May I or even stanford berkeley, I would say
I would say, deep. Mine
goodbye,
okay, I would take a slightly different approach, a different answer and leave or come back to facebook in a minute, but I think those other places are
Following a dream of one of the founders, and I'm not sure that it's well founded the dream,
and I'm not sure that it's going to have the impact that he believes. Is he talking about facebook and google and so on, and I'm about Google, Google. But the thing is those research labs? Aren't there there's the big dream
I a fan five, no matter what the dream is. A big dream is a unified, because what happens
you have a lot of bright mines. Working together
On a dream, what results is a lot of like adjacent ideas? I have so much progress
made yeah, I'm an so I'm not saying they are actually leading. I'm not I'm not saying that the universities are leading yeah, but I don't think that
companies are leaving on general because they, you know that we saw this incredible spike in you know attendees at europe's
as I said in my january first review this year for two thousand and twenty to that,
And twenty will not be remembered as a watershed. Dear,
he learned or I not those nothing surprising happened anyway, I'm like when people are being hit image in it. That was it that was a shape and does the lot
people writing papers, but the papers are fundamentally boring and uninteresting incremental work.
Their particular memories of admins gear or somebody else. There mit the stand out a funny stories
unfortunately foresees another. One is passed away, you ve known
the biggest minds and high yeah. You know they they did. The amazing things and some sometimes there were grumpy
what he was a he was, as he was very grumpy, but that those
I remember him thing. I interview that the key to success or being to keep being productive is too high,
Everything you ve ever done in the past. In a minute
maybe that explains the perception book.
they really wasn t told you exactly by he meaning like just like
I mean. Maybe it's the way to not treat yourself to serious issues always be moving forward.
They d, I mean they back crankiness does
so let me limit limits anywhere. You know what really I'm.
You know the joy memories are about having access to technology before anyone else and seen so so you know, I can't I got Stamford in nineteen. Seventy seven and we had come in. We had terminals that could show live video on them, the jewels digital, sound system. We have them as their xerox, graphics.
know we can print and it wasn't. It wasn't like a typewriter ball, hitting with characters. It could print arbitrary things on mean in one bit in a black or white arbitrary pictures. This was science fiction. Sort of stuff
and at mit the list machines which you were the first
Personal computers and the internet-
one hundred thousand dollars each, and I could you I got there
enough and that there is not one for the day couldn't couldn't stand up
people working
having that like direct glimpse into the future
yeah and you know I've had the email, every basins, nineteen, seventy seven and then the the host field was only eight bits.
In many places, but we I can send email to other people at a few places, so that was that was pretty exciting to be in that world. So different from what the rest of the world new and I get. Let me ask you probably added the sub
in case you have a story. I'm hanging out with don knuth for awhile tomorrow did. Did you ever get a chance to such a different world? The yours a he's, a very kind of theoretical computer science, the puzzle of of computer science, mathematics and you're so much about.
the magic of robotics, like the practical did you mentioned him earlier for like not about computation? Did your worlds cross sector?
in the you know, I I know him now we talk you know, but let me tell you myself, my done so, and you know, besides in an analysis of algorithms, he's well known for writing tech, which was in latex, which is the academic publishing system. So he did that at the ai lab and he would do it. He would work overnight at the outlet and one
one day the one night, the year the mainframe computer went down and am.
we're a guy named Robert Paul was there he only did his phd at the media lab at mit and he was an engineer, and so I am he and I
You know tracked down what were the problem was? It was one of the big refrigerator size of a washing machine sized describes had failed. That's what brought the whole system down. So we got panels pulled off and were pulling in a circuit
doubt and donald committees, whose are really talk. I walked in and is looking dallas. When will it be
fixed because he wanted to get back. The writer has to system will grace the old can move instantly. We've we figured out. It was a a particular chip. Seventy four hundred series chip, which was socketed- we popped it out, we put a replacement in put it back in smoke,
that has been pulling backwards, because I was so nervous that canute Annie s affect everyone. We eventually got up fix the other, the mainframe running again that as you
Why was that again? That will? I must be before october. Seventy mm
because we moved out of building and so some time probably seventy
Sometimes some early, seventy nine yeah, those all those
Is it just as any other people with pass passed through mit, really fascinating?
Is there a lemme ask you put on your big wise men had is advice you can give to young people today, whether in high school college,
For thinking about their career or thinking about life,
to live a life there proud of a successful life. Yes, so many people ask me for advice and have asked for my give I talk to her.
all the time and there is no one way
There was a lot of pressure to produce papers and that will be acceptable and publish.
Maybe I was maybe I come from an age where I would like to be a rebel against that an and stole succeed. Maybe it's hot it today, but I think it's important not to get too caught up,
With what everyone else is doing, and if you, if loved ones like what you wanna life, if you wanna come,
I have a real impact. You have to be ready to fail a lot of times. They have to make a lot of unsafe decisions and the only way to make that way.
is to make keep doing it for a long time and then one of them will be workout cancer that at that will make something successful or not.
Oh yeah, oh yeah! Oh you just may end up, you know not having a having a lousy career, I'm glad it's certainly partake in the risk of that so but it it, but there's no way to to make all safe decisions and actually really contribute. Do you think about your death by your mortality.
I gotta say when covert hit. I did because we in on the early days we didn't know how bad it was going to be, and I that made me work
what kind of her while but had started this
company and now I'm don't full time and mournful time companies of the books on hold. But I do want to finish this book
when you think about it? Are you afraid of it, I'm afraid of dribbling?
Enough in creating an I'm losing the details are ok, yeah yeah, but the fact that the right ends
I've known that for a long time come so.
Never, there's knowing and knowing
yeah and I'm really sorry. It feels it fills a lot closer. So my in, in my my blog with my predictions, my sort of push backing, and that was that
I said I'm gonna do review these every year. For thirty two years puts me in my mid nineties so might apply a whore,
Every everytime you write the blog posts getting closer and closer to her own prediction. That's that's true. Are your death yeah? Will you hope your legacy is?
one of the greatest robotics researchers of all time, and one, I hope is that I actually finish writing this book and that who's the one person who reads it and sees something about changing the way they're thinking and that leads to the next,
big and then they'll be on a pike ass two hundred years, russia once read the book and that changed everything.
What do you think is the meaning of life. This whole thing the existence
the all the hurried things we do on this planet.
Greetings and me we're off well in either were really bad at it, life,
meaning or both air. We get caught up in in the it's easy to get easier to do the staff, its immediate them up to the stuff. It's not immediate answer the big picture or better. Yet do you have
sense of what the big picture is like Y ever look up the stars and asked why the hell
we hear.
You know my my young, my my atheists m tells me it's just random.
But you know I want to understand the the way random in the in the spot. I talk about in this book how order comes from disorder yeah, but the kind of sprung up like most of the whole thing is random, but this little pocket of complexity.
We'll call earth that like why the hell does that happen, and- and what we don't know is how common that put those pockets of complexity are or how often they may not last forever, which is the more exciting, very sad you if we're alone
Our affairs infinite number always stations. I think it's possible for me to believe.
if that were alone. That was just be too horrible to prove to be the sad thing he could be like a great
yard of intelligent civilizations, o everywhere, though, by maybe the most likely outcome, and for us to get exactly
all of this will be forgotten. Now
including all the robots. You build everything forgotten.
Well, I'm average
Everyone has been forgotten history right death.
most people are not remember, the young.
generation or too I mean
Yeah well not just on average, basically very close to one hundred percent of people who have ever lived or forgotten yeah. I mean
in the long arc of I don't know anyone alive who remembers my great grandparents cause. We didn't meet them so still this fun at the Sarah. This life is pretty fun somehow yeah. He definitely.
immense absurdity and at times meaninglessness of it all it's pretty fun,
For me one of the most fun things is robots. I've looked up to work. I've looked up to you for a long time from africa and broad dead
It is an honour that you would
spend your valuable time with me today. Talking as the amazing conversation. Thank you so much for being here. No thanks, but thanks for talking with me, I enjoyed it. Thanks for listening to this conversation with rodney brooks the support. This podcast, please check out our sponsors in the description.
and now let me leave you with the three laws of robotics from isaac asimov, one,
a robot may not injure a human being or through inaction, allow a human being to come to harm two. A robot must obey the orders given to it by human beings, except where such orders would conflict with the first law and three, a robot must protect its own existence as long as such protection does not conflict with the first or the second laws.
Thank you for listening. I hope to see you next time
Transcript generated on 2023-05-07.