>> the c-span podcast listen tol of c-span's podcasts that feature nonfiction books in one place. discover new authors and ideas each. we make it convenient for you to listen to multiple episodes with critically acclaimed authors discussing history, biography, current events and cultures, with programs like q&a and book notes. listen to our podcast feed today. you can find it on the free c-span now mobile video app or where you get your podcts, and on our website missy's and.org/podcasts. -- c-span.org/podcasts.

■♪ >> sebastian thrun, called the fifth most creative person in business and foreign policy magazine calls you globathinker. what exactly do you do for a living? sebastian: that is vastly exaggerated but i have tpleasura university for more than a decade and i have created a few companies in silicon and i also worked at google founding google x. >> what are you studying and >> i look at people with a different lens. i believe the most interesting

things have not been invented yet. i believe in the last 150 years have been■m transformative, light switches, airplanes, cell phones. but this is just the beginning. i really care about what technologies can be invented in condition. >> so there are a lot of unknown unknowns. sebastian: that is true but we see trends towards innovation and connecting people and smart machines and trends need directions. >> you have been quoted that machineving is the core of what i do. what does that mean? sebastian: machine learning is the technical word for artificial ielligence.

that computers can program themselves. consider how you raise a child. when you raise them, you do not give them a rule for every situation in life, you train and teach thse word -- their own rules based on feedback. until now, a software engineer was writing down step-by-step what a computer ought to do. we have a set of technologies that computers can teach themselves, find their own rules based on data. >> maybe we should start at the nn terms for us in a way we can understand. let's begin with computer science. how do you identify that? sebastian: 50 years ago

computers got to a point where more than just a handful of people could program them and a computer program is like a recipe. you put in the salt, water, eggs.computer science became the discipline of computer programming seems like it is bizarre that there would be a discipline, but picture this. a modn phone is 15 million lines of code, a big recipe. the computer scientist learns inciples to make the >> algorithms. sebastian: it is a funny word. it is a funny word for the kitchen recipe. whenever you writey9 things in u

have made an algorithm. >> deep learning. sebastian:■çgrad student i was working on the same ai machine learning as people today because the brainse were able to build given the the size of a cockroach and it is not that impressive. by having fasterster networks, e possible to move from the a person on that thing, bigger data, more machines, is deep learning. how do you identify singularity? sebastian: it is the idea that things accelerate so much so

far, you will not be able to see what happens on the other side. i would argue in the middle of a singularity today. we see so much change and innovation say where it is leading us. but it is expensive eventually accelerating -- expenaccelerati. look at humanity. 300,000 years old as people, give or take. stuff that was not imagined, the discovery of dna, open heart surgery, they have been in the last 100 years. that is exponential acceleration . all of a sudden we become more creative, investigative, acceleration is accelerating

today, getting faster. it is what i would call the term of the year, artificial intelligence. what is the definition of that? sebastian: in 1956 by a researcr at sanford with the inspiration to give computers the same level of intelligence as people have. we as people intelligent, we can do amazing things. so i help make a computer that can do the same thing. 1956, almost 70 years ago, there was the belief that what made intelligence so was logical reasoning, the ability

to do an eight class math lesson ter. what was undervalued at the time was instinct. but despi that, the core of ai is how to makeple. >> it seems in a sense that it has been all of a sudden artificial intelligence has become part of our lives. is that correct, or does it just seem that way? sebastian:of a big football gamd someone scores a touchdown and we all talk about it. it has been accelerating the last 15 months. bu is, this was with us long before. take google. when you google■!google simultas

through hundreds of billions of to find those 10 pages most informative to you. it does not look and feel like ai, but it is intelligence, massive data crunching. ai has been with us in many aspects. ll phone, the fact that relatively good language comes out the other side is the result of very interesting official intelligence. it is part of the fabric, but not the fabric where it is taking■m, endangering or threatening jobs the way it is today, but it has been with us for quite a while. >>■( a little boy more about sebastian thrun -- a little bit more about sebastian thrun founder or cofounder of google

x, google brain, waymo, kitty density, online learning, presto ai, and sage ai laendeavors. what do those two operations do? sebastian: questa is five years old at this point. it can we build ai to make we focus on contact centers, the people you talk to when you do not like reservation. he found in that word of -- world of online that a good experienced person is four times as good as a newhe asked e

build ai to w and then assist nw people doing their job better faster? he had ai extract from the experienced people and impute it into the brains and the answer turned out to be yes. a recent study from m.i.t. and stanford showed there is a 13% improvement if you have a coach on your side that is ai and learned from wise people and has assisted you in being better. ai becomes theto bring knowledge from smart good people to everybody else. >> what doesdñ-%%y ai in generao better than humans and what do

humans do better than ai? sebastian: ai can crunch through data sets faster than people can. when there is a regularity in dr hand. in 1997, and ai computer beat the world chess champion. ■xbecause it can look like one million more chess games than a person could. ■ain 2018, and ai system beat te world leading goal player. and ai system was able to study

and come to a solution that the the sitting strongest player. google can go through hundreds of billions of until you find the one that matches your theory the best. data.is also the limit of it needs so much more data than people ever look at. if you test ai with the single same data a person has, there is no chance. it needs lots and lots of data. >> data is information about you and i and yes. society and public data in the codes we write and there is personalized data that often has

general utility. in waymo of test drivers and drive cars to public streets and experience situations driver would normally experience and feed that data into a big ai computer to generate safe driving controls. >> back to the list of some of the companies you founded or cofounded, how were these of funded, and are they profit-making? one of the great ways here in silicon valley and worldwide that when you have a good technical idea, you can often find an investor who is willing to banl the development of the idea. any good idea begins with some sort of technology that can't solve an important problem and

before there is a business, you have to be able to build a warchest■c it took more than 10 years of training self driving cars be safe be opened up to the public. so that investment led to amazon, google, alphabet, all the big companies we know today. >> are they making money? sebastian: some are. you density -- udacity is profitable at this point and cash flow positive. others like brand-new sage is technology. >> how is it funded?

are you self funding? sebastia?" there is ventral -- venture capital. it is finding young new ideas and talent and helping them build the team and technology in tech in the technology market. pilly inilic valley you raise a few million dollars, it sounds like a lot of money but it is not as much when you ■employ 20 software engineers to do it for you. then you test in the market and see if a customer is willing to pay for it. >> cory doctorow, the author of technology critic, wrote a que. today, morgan ai is a $6 trillin opportunity. they are not alone. the ceos of endeavor, microft ys

illustrated, caa, are all out every hour, declaring the futu is ai. but even if you add all of this up, double it, and add a lln-dollar confidence interval, is still does not add up to what bank of america analysts called a of the 1990's. for one thing, the most exciting part of the internet in the 1990's was that it had 5 low barriers to entry and it was not dominated by large companies. it had them running scared. the ai bubbleyb contrast is being inflated by massive incumbents whose excitement boils down to this will let the biggest companies get much, much bigger and the rest of you can go f yourselves. some revolution. that is what do you think about that? sebastian: i did not say that,

first of all. therebelief ai will be transformative in a trillion software products. would go back and look at the effect of natural fertilization had on agriculture in the last 50 years. years and years ago, almost everyone farmed and a farmer would make food for his family. it has massively changed the world. if you can make people more efficient in their daily work, it will have the same effect. so i would be careful to take a strong position on the size of

the impact, and the sis it is tt make all the money? i do not it has always been the case as new technology comes along, new companies come out. they eventually become big but it is not like the existing big companies. openai for example is a startup company day when they pushed chatgpt they were less than 300 peopleopenai popu. a lot of us could open source teams have been open source models but there are companies giving away every element of it. meta has done it.

there is no -- a company in abu dhabi has built powerful oils. we are seeing a building of evolution that seems more democratized than other technologies before. again, i think we are in a singularity and it is easy to speculate and make contrastive statements. but the potential exists. >> back in may, samtestified in. here is a little of what he had to say. >> i aee that wh we get very powerful systems the landscape will change. i am more optimistic that we are incredibly finding other things to do with tools and that will keep happening.

my worst fear is that wet harm . if this technology goes wrong, it can go quite wrong and we want tbe that and work with the government to prevent that from happening, but we tried to be very clear about what the downside case is and >> sebastian thrun? sebastian: the thing i admire llingness to really reach out across the aisle between silicon valley and washington, d.c. to the policymakers and decision-makers and leaders and really build a team with them to figure out what the implications are. i think the statement correct that this will impact all of us, not just technologists, it is for everybody and everybody deserves a seat at thumanity has that if they go wrong, they can

go horribly wrong. warfare, nuclear weapons. we as a society so far have done a pretty good job to protect ourselvessasters that could be d with these technologies. most of the technologies also have side, like biotech in medicine. ai should be the same, a source for the good of people a we should be very aware of the dangers. i am generally an optimist, as i'm sure you have noticed by now, i do see immediate danger i have written about, deepfakes. they are now at a place ereyou a person that is indistinguishable from reality. that will be tricky because bad

actors can use to pretend to pretend crimes that did not exist or all kinds of stuff people believe in falsely. so us to keep an understanding of what is right and wrong and true when not true will be a challenge going forward. >> we cheated a little bit here at c-span and we went on some ai chatbot's asked them to ask you questions. we went to openai and got some questions. we had to go one step further and say, how about some critical questions for sebastian thrun. 45 seconds later, we had four sheets of per wit specific questions for you. this is one from googles barda aited a critical question for sebastian thrun.

the first thing this machine said to us, 45 seconds from typing it in to the piece of paper in my hand, it said, no ia fascinating and accomplished individual wita alth of knowledge and experience. by asking thoughtful questions, you can get valuable insights into his perspectives on ai, education in the future of technology. here is one of the questions. sebastian thrun often expresses optimism about theenai to benef. however, critics argue that automation could lead to job displacement and exacerbate econ inequality. how does he address these concerns and advocate for policies that mitigate the negative impacts of ai? it is al question and i think you for

adding the word critical. [laughter] ■val#it has been a supple questn for me in the following sense. i believe that ai will, it is not a zero-sum game. it is a technology that will make lives better for many peop examples from my own life. one is waymo. waymo cars take■ passengers here in san francisco, 100% safe, this technology coullead to fewer traffic deaths. we lose about 1.2 million people ait is a significant number of people, it used to be the leading cause of death for young people in the.s.so it will reale everybody, not just a small

number of people. with that example, the second example is medical diagnostics. we trainedi cancer with your ipr android phone and we found that a well-positioned can find skin cancer as accurately as the best human doctors. it mne a few years ago and it shows that ai can help the medical field diagnostics to a much broader set of patients. in north is the cancer number one in terms of numbers and it can go from completely benign toeadly in less than a year. they are brittle and deadly. -- brutal and of ai, if done we,

will save us time and give us time back and more time to spend weekends with our family. if you do not believe that, look at the last 50 years. words like vacation on weekends it did not exist and now they are common language. is easy. we do have to worry about distribution and make sure the technology really reaches everybody, and in the united states we have a free market economy and the people who■gto e ownership and that ownership has proven to be useful, and i say this is a former german who grew up on the border of east germany and we saw what the effect was

on society. having a dialogue important -- is important. everybody benefits is very important and i have deep faith that technology in the e his access and really helps everybody, but that is my optimism. >> you referred earlier to sam as sam. is silicon valley a small village for those of you who work in it? sebastian: a bit. and it is interesting because a lot of people did not grow up sebastian: many people are finding a home here. there is optimism about technology here that really exists.

the willingnessmany things we te failed but every successful startup, there are 10,000 companies that fail. ■10& the successful ones and not the misery of failure but that allows you to get up on your feet again, trying to find something that makes the world a better place is very unique and i have not seen it anywhere else in the world. host: failure it is frequently said is an accomisen me, focusis missing half the story it is not failure that makes us special, it is our abilityo iterate quickly. it is the wilngss to stick out your neck, take a rk th

the purpose of learning smart learning and fast recovery is as important as fast experimentation. the end effect is secret of silicon valley. incredibletetion speed. it is launch, fail, learn, relaunch. i've seen it over and over again. whoever minimizes the duration of each iteration wins. %y could not add anything to that. host: what is a typical day like for you? sebastian: a typical day -- changing diapers right now. the joy of two small children. a lot of work that i personally do is empowering my teams that i work with. i do this by setting a vision of something that is hard to

and then providing a safe space this vision. an analog i would give is imagine you find yourselfmountad ever been climbed. go back 100 years, to climb mount everest. how would you do this? you cannot pick up a book that says here's how you climb the mountain. you've got to take some chances and as you climb, you take your best guess.hopefully you will nb the same false summit again. there are days when you cannot see the peak, you have to believe it is there. yet to go on faith, go and go and go. and to

you have to have endurance. one foot next to the next foot, no shortcuts, you just have to do it. to■[ me the imaginary mountain s the same as starting a successful company. you find a great team of people, set the vision. say here is the mountain. a good mountainaying the world from traffic exigent's with self-driving car's and then you realize with all of your arrogance to pick this mountain, you have no clue how to get there. you've got to be a first grader again. a growth mindset, willing to make erythe only rule i have isn make every mistake, but make every mistake wants, not twice. host: you've mentioned teams. what do you look for when hiring?

sebastian: a certain skill set that they usually need for people, its imptant that skills are more savvy. in the early stages writing software is important but the most important interview question is something you are really proud of in the last 10 or 12 months. tellwrong and you would be amazt how few people can answer this question. i do something wrong everdareale day. when people have no answer to the questionyou did wrong in th, it signals that the person is not as self effective as i want them to be.

i believe all of us make mistakes. every person have met makes mistakes, gets things wrong. but to learn you have to have an attitude to say let me reflect on my mistakes. reflect, you should be able to produce an answer to that question. it is amazing how many people tell me of a mistake that their boss made and they had to fix. the reason they had to fix the mistake, that was a persoi would not hire. host: kind of like machine learning. sebastian: our drive in humanity is based on learning from mistakes. in aviation every aspect of an airplane reflects a prior crash where this thing was missing and

somebody died. i'm not making this up, there are books made about this. in society when we sit together science, it's about an experiment and we report the findings openly to learn from ththat is what drives humanity. it is the biggest power that nature has ever invented. host: wt tosmith, microsoft prem his book in 2019. ■"this will require that those o create technology, not only from disciplines such as computer and data science but from social and nasciences and humanities. if we are to ensure artificial intelligence makes decisions based on the best that humanity has to offer, its development must resulfr a multidisciplinary process. as we thinkbo the fure need to

make sure every computer and data scientist is exposed to the liberal arts. st who majors in liberal arts will need a dose of computer and data science. sebastian: happy. i have been fighting together hd computer science tools because of the core we are dealing with societal things. and it iour nsmake us drift apae to talk to each other. this is the time that all of us have to talk. and humanities and social science are extremely important because at their corethey care n technology and people. use of■/ technology, the book, e

broadcast. now film aredriven by technolog. so is art, performing arts, things that go hand-in-hand with technology. technology has influence these things. we need a deephumanities social, history, to shape the future. orie was winning the dartmouth grand challenge, he won a $2 million grand prize. in 2016i had the opportunity to ride in a self-driving car in pittsburgh connected to carnegie mellon university. 'óit fel moving from

horse and buggy era to the automated era for the first time. it was a little disconcerting to be vehicle. are these ready for prime time? sebastian: dangerous to talk to a person with a memory. 2005 when many listeners were still chi we had a race organized by the u.s. government. they have an office called defense research which is behind things such as the internet and other inventions. qthey had prize money for someoe who could build a self-driving car effectively from las angele. it was 140 miles. the car had no driverñ4■= and is

making life or death decisions for six hours. it was an incredible race attended by who got together. it was not thepo■ñ level that we at today. fast-forward to 2009, the founding of a project that became waymo and today you can summon one in san francisco. it is 100% safe. ther is no driver inside, but the steering wheel turns and keeps you safe. as safe as you would be with any driver. host:le with that now? sebastian: 100%. i have been confident for a decade.

years ago driverless cars were driving to lake tahoe and back, no one knew about it. ride. host: one of your other companies was kitty hawk, this was about flying cars. are tho happen in our lifetime? sebastian: he sold the majority of the assets to and the idea is back to the jetsons, what if your car could fly? what if you had a button that made the car levitate in the air , go in a straight line to where it has to go, and land? thanks to advancements in batteries, it is now feasible to build these vehicles and have them fly about 100 miles on a

single charge, so we could fly 100 miles on a single charge. that is quite interesting because in the future, i firmly believe most transportationbe ao that will lift you up and then get around all the obstacles on the ground. trees, houses, traffic lights. with a safer, faster and greener, car. that is going to happen in 10 or 15 years. host: have you been in a flying car? sebastian: where did you go? sebastian: a training camp where we tested them in las vegas. we tested over a lake because we were not certain it was safe enough and the fear was if wrong we could

take a swim. but nothing went wrong. we were able to make 25,000 test flights of these vehicles successfully. host: if i wanted to fly one of tho4ose, how long d it take me to train? sebastian: there are a bunch of companies that seek certification. we saw the newest with boeing, a pilot plane breaking off. the faa is concerned about safety■p a them requires approv, certification leading companies are three or four years away. host: i want to switch gears with you and talk about military use of technology and ai. jeffrey, the godfather of ai,

believes that todd military robotshe said that in october. what do you think? sebastian: this e. there are a lot of autonomous things out there today called landmines. they are being tsomebody for dey still kill people. it is a massive problem and that should teach autonomous -- should not be legal. are ethics involved. i wish a world existed without warfare. warfare exists but there should be ground rules and crossing the border to completely autonomous weapons is a mistake.

you might not control what they are able to do, we might not be able to stop them or hold them back. able -- not be able to stop landmines today. that's a massive mistake. host: d consider artificial intelligence to be corporate or state secrets? sebastian: technology wants to secrets are incredibly public and that is good for the world. i want to remind everybody ai is not intelligent. it is not a system that takes responsibility for actions. it is a tool like a shovel that lets you dig faster. it should be in the control of people. people will use it for good or bad purposes. they might abuse it and then you should put those people in jail, but we should never seize

control to the machines. host: isn't there a lot of theft , intrigue, espionage in silicon valley? sebastian: there is a lot around the world and there always has been. lucky for us, freedom of information makes it harder to be a criminal. i live in the unitedta i am a proud nationalize citizen and one thing i love about the united states is how you can be transparent about people and business. we have the freedom of information act that lets us get that is unique and helps put limits to what government can do. i anspen force to make it harder for criminals to be criminals. host: growing germany where your parents scientists? were you exposed to it you work

on today? sebastian: no, my parents were my mother was a stay-at-home engineer, my father was a struggling engineer with companies that did not work out. saw a lot of hard work. the thing about germany that rubs me the wrong way is the deep skepticism that germans tend to have. they say this is a great idea. wrong with this, what is wrong with you? there is a deep skepticism in most ofurfind in silicon valley. silicon valley is young enough that people are optimistic, sometimes naive like myself. i love this optimism. host: we talked about theths but recently, pope francis called

for a binding global treaty on artificial benefits but the raw potential for destruction, accountability, do you agree with pope francis? host: i adore the pope for many reasons. is too early to have strong regulations today. the way i look at law and regulations, it proceeds certain events. you see a form of abuse that you do noty put a law in place to make sure it does not happen again. lawmakers are struggling to understand where to draw the line. italy when chatgpt came out for legalized it entirely. eyes printing press entirely.

i want to be thoughtful as to what the abuse is and really think about how we can put a block against a specific abuse and to the present day, technology is so young that we ve seen abuse. host: this was an publishers weekly. the new york times sues openai, microsoft for copyright and judgment over unauthorized use of intellectual the training of artificial intelligence technologies. are there legitimate fair use issues at work? st this is the point where i have to take the fifth. i can fully understand the it is a fact that ai systems have been trained on copyrighted material. it will see if

it ends up along the lines of the internet. l it be more traditional and you have to and pay for copyrighted information. if i had a wish i think there is real value in oducinthe people e reimbursed and get something back in return. if you can capitalize everyone's work, the amount of people producing high-quality content -- i seem to side with journalists. host: the ceo of applied excel dinot take the fifth and here is what he wrote in the wall street journal about media and ai. ai deep diving into deep data, quantifying occurrence of events

and revealing insights that were once obsre with checks for biase discernmen into precise narratif documents the melding of human intuition and machine precis puts journalism on the brink of becoming a faster yet more analytic profession. sebastian: i think that is a good statement. the relationship -- i have seen ai going through massive amounts of data and giving meaningful answers. go to chatgpt and ask it any question, it's amazing. the same with dollythat is a fa. it will shift the job of journalists. the fundamental hypothesis that a journalist seeks deep

information and tries to relay this information to a broader ■■ce wl same but the method will shift. host: you've mentioned a company you formedto quote you on education before we talk about it. the basic problem is rulesor edn and employment no longer apply. this means tradition bound higher education systems must adapt. we need to transition from a one time education society to a lifelong education society. degree at stanford, was it worthwhile? sebastian: i have no degree, i was only a professor and was ranked at the bottom of all computer science departments. i founded udacitcause in the

united states we had the best institutions but they are exclusive and limited. china has 20 million students hungry to get the same access, let's not forget. so when they said let's open up stanford i offered a course. it was the first massive■)course world and 160,000 students signed up within a few weeks. 100 60,000 students was way more than they could fit on campus. i vote really good about it. 160,000 finished andhe top were not at stanford.

the message is therere amazing number of people in the world who are every bit as capable, but they are not part of the system. maybe it■x is demographics, age, affluence e. they are not part of the system. so i wanted to democratize access to the best education. you can. we have a massive footprint. foregypt has spent significant resources to teach people to become freelancers. freelancer, you bring hard currency into egypt. after one year of work we found the average freelancer briabouty into egypt and across egypt to train them. it's kind of amazing to see how

we can change the world and make to education affordable. host: back to our critical question for sebastian and again this is from chatgpt. critics arguegram profit over educational quality. how do you respond to such claims? sebastian: we turned the same problem into georgia tech. they started an online program in computer science and the degree is indistinguishable from the degree on campus. ou the end whether you are online or on-campus, it has the same value to your future life. the quality of education is equivalent.

the only difference is the costs $6,000 versus the on-campus degree which cost $45,000. why is there a big difference? it is much easier to teach online to reach more people than cramming them into a classroom physically. so as a result we have 10,000 active students in the masters program in the united states. it has a big impact on the u.s. economy. economists believe it supplies a percent of scientists in this country. but it shows thatechnology, youe education more accessible and affordable. to those who believe we are putting profit ahead of education, it is hard to argue against it, because at some point education cost■'s money ad a company has to take in money. not just education, but also the

engineering in the management behind it. and we are profitable, so we've been saving a few bucks for ourselves for a it is so much cheaper and there are many existing universities that crgan we charged. call it 2000 u.s. dollars when atypical degree maybe $100,000. so i think rather than pointing the finger and saying you are let's work together and invent new technologies to reach more people. of the many billions today, theg higher education system reaches a small sliver. i would challenge every uneradmit 10 million chinese students?

10 million indonesian or indian studentsi't you become the most important university? i see the possibilities and we' taken a tiny step in the right direction i hope. host: let's close with this video from stephen hawking in 2014 and>> primitive forms of al intelligence we but i think of full artificial intelligence could spell the end of the human race. develop artificial intelligence, it will take off on its own and redesign itself at an ever increasing rate. humans who are limited by slow biological evolution could not

in would be superseded. host: sebastian, final comment? sebastian: i thinklity that thit exist in the distant future. elon musk sent others have warned the same way. is more opportunity to use the technology to make work better, lives better, connect this right now on the shelf, keep my eyes and years open, make sure we do not give too much capability to these limachines, and utilize the good in ai today. host: that was sebastian who opened our program. most things have not been invented yet. we appreciate you spending an hour with us. sebastian: thank you so much. ♪

>> all q hyundai programs are available on our website or as a app.

the supreme court oral arguments.

the case that could potentially restrict access. and then the center for american women on the anticipatwomen in . also, this. veteran■u■ -- increases representation while promoting bipartisanship in congress. c-span's washington journal, known in the conversation live at 7:00 p.m. mr. on c-span, c-span now. >> c-span is your unfiltered view of government. we are by these television company's and more, including cap broadband.