it is still the case that a human teacher helps to motivate a lot it's very hard and one of the hardest things about learning a language by yourself it is possible to learn a language by yourself with the application but it's very hard to stay motivated but on some level i think you're providing the kind of service that no teacher can provide and i think because thing about your technology is that it marries assignments with instantaneous progress evaluation no all teacher could provide that kind of precise feedback to his students right yeah there are many things that we actually think that teachers and doing a combined are better teachers you know one of the pros with teachers especially in a classroom is there's one teacher and there's maybe thirty students or twenty students and they cannot provide feedback to everybody and this is one of the great things about the lingo that it provides immediate feedback to each for each person it's personal i so so that's great but you know a lot of times you may have a question that a teacher is better at answering or also keeping you motivated is a big big things are i think the combination of the two is most most useful i know

that when you were embarking on creating you read a bunch of books on b. theory of language acquisition to figure out for example whether the cheek blue rules before adjectives held by first and you know ultimately ended up running those tests on your own users i wonder how does the language learning reality differ from the language learning theory it differs a lot with that's one of the things we have found before before i started doing go i read a lot of books about how to best teach a language and i realised that a lot of them contradict each other usually it's because they're not based on much data they're just based on somebody else one person's method or something so what we did is we launch to a lingo and then because we have we have hundreds of millions of students we can watch them learn and and because we can watch them learn we can find better ways to teach that a lot of times are not what what people are using in classrooms for example when teaching inning. english

a lot of classrooms in many countries in the world teach the pronouns personal prominent pronouns like he she you they also teach the one that's it that one is much harder for a lot of people so one of the things we started doing with doing was when we teach the personal pronouns we don't teach that when we teach or much later and that's we we found a way to do that based on the data for use as we found that it was better to start doing that now one of the major obstacles that people learning a foreign language have ways social anxiety is the fear of practicing a foreign language and they hear of looking incompetent i know that you have followed your own very elegant i think i'm very easy simple easy solution for that chad but it's but as easy as it may seem don't you think that those conversational skills come at the expense of social and cultural skills that you also acquire when you learn a language from another individual yeah i think learning a language i think ultimately you learn a language to talk to another individual that's the that's the end goal but one of

the things we have found and it's not just us other people have found this. one of the hardest things about learning another language is when you when you practice with another human there's a lot of anxiety in fact the u.s. army you know the u.s. army tries to teach people languages in particular lately they're very interested in teaching people arabic for obvious reasons they used to they used to try to pick some soldiers to teach them arabic and they have found that for some of them it is possible to be to marry very easily whereas for others it was very difficult and they didn't know what the difference was so they developed a test to figure out which ones are better at learning a language and they found before they thought that maybe intelligence had to do with it the biggest factor that had to do whether you're good at learning a language or not is whether you are ok sounding stupid absolutely but when you are delegating that practicing capacity to chad box that here that executive doesn't disappear in fact it may become even stronger you may be. very comfortable talking

to chad yeah so what we have found is that what people do so we develop the chapbook for doing go and the idea is that you get to talk to the computer and not to another human so you don't feel anxious what we have found is that that is a very good thing to do at the beginning and then that gives you enough confidence to go and try to talk to somebody else it's like training wheels with bicycles it's just it just gets you started and we have found that that actually helps later when you go talk to a real human you feel you feel more secure of yourself and what about the cultural aspect of language learning is there any way to incorporate that into your program yeah we're trying to do that we haven't done it as much i mean i think that that's one of the things that where we want duolingo to get better at teaching you the culture of also the countries when you travel i mean there's there's a lot there and the english would be bitching american culture or the british culture they are trying to you get to you get it would go into so that would be the idea and similarly when people are learning spanish there it's you know there is like twenty five different countries that speak spanish you would get to choose and

we're starting to do that but we haven't quite done it i mean for example slang terms very very much between different countries and a lot of people really want to learn those so we're going to add that's not really there yet on the lingo but we're going to do that so you know you already mentioned that you think that the best the ways to have you know the best of the two worlds to come by teachers and the application in yourself benefited from having great teachers in life. change your ph d. under the guidance of professor many a little who later became you know co-founder or co-invented could invent a brother so i just wonder from your own personal experience which learning tasks you think about a delegated to algorithms and we each still require the meeting of minds a human to human interaction you know that's a great question i don't know the full answer to that i do know that computers are really good i'm giving you repetition so whenever you need to learn something by repetition computers are very good because they don't get bored whereas a teacher may get bored. computers are also really good when one to one interaction

is required because if a computer can adapt to yourself where is the future usually has to teach many people at the same time what i think computers are not yet good at is keeping you motivated it's amazing how motivating another humanness just the presence of another human gets you to do all kinds of things so that's one thing that i think computers are not good at and the other thing is answering general questions so if you know that the computer may be teaching me but if i am confused and i just have a general question that that does not fit what the computer is trying to teach me. it's stuck where as a teacher you know and can actually answer that now speaking about professor blum the product i feel collaboration with here was capture and recapture application to make you whole to google a few years ago to a rookie like myself like very very think will need ideas that one has to wonder why did it take so long to develop them a why they cost google so much money what was the hardest thing to figure out about

that yeah so the capture of if you haven't seen this is the story of characters that you have to type whenever you're trying to get out an e-mail account or something in retrospect that seems like oh no brainer yeah like a no brainer and the idea by the way the reason that's there is to make sure that a computer is not getting the account that it's a human that is actually getting that and not on a computer it seems like a no brainer but it you know nobody has thought of it and it took us months to think things over it and a lot of times some of the best ideas seem like a no brainer i don't know what the hardest the hardest part was really coming up with with a task that was easy for humans but hard for computers. and it well and also that it was easy for computers to grade so the computer needs to be able to know whether you enter the right answer but it fired off a sickly should not be able to answer itself now one or that often pops up in your speech uses the word. efficiency and i think what you mean by efficiency is a sense. deriving value from things that we didn't think that had value before for

example like you know the time you spent typing that rick up towards and i'm sure many people would be interested in how you train yourself to spot goes up the changes do you start with looking for that wasted energy or do you start with a project where that wasted educate be applied i start with a waste of energy so i start with things that people are anyways doing so in the case of a capture it started out as a security mechanism where people had to type things and then surely we realized that a lot of hours were being wasted with that if you if you think about it it's about two hundred million times a day somebody types one of these guys just and each one tastes about ten seconds and if you multiply two hundred million by ten seconds humanity as a whole as we wasting five hundred thousand hours every day typing. so that all that time the first step was ok there's a lot of time there how can we use it and that's when i came up with the idea that you could be using that type to help

a time to help digitize books. so this is this is the recapture project and that was what google bought the idea that books are being digitized in order to digitize a book first you take a photograph of every page then on the computer he says the cipher all of the words and then the unfortunately the cannot read all of the words but the words that i cannot read where you're using human started typing it after the help was read those words and i think that's a very ironic to some extend idea because you are using value added coude sourcing to fide against another form of malevolent crowdsourcing which spanned really is and maybe that's again another key question but do you think crowdsourcing is a technology is it more amenable to. positive pro-social applications or is it more likely to be has more opportunities to be used in negative ways i mean everything knowledge you're going to use also they're not going to leave the pens on what you use it for but i would say most of the applications that i have

seen of outsourcing have been very positive i mean you can think of wikipedia as a crowd sourcing effort there have been a lot of applications about trying to find missing people or in their case and recapture and helping to digitize books i think most of the things have been pretty positive but life with everything i think it can be used for negatively but now both capture recapture and i think initially until lingle was based on the new economic model it's essentially a swap model which limits or eliminates the need for money as a transaction facilitator at least for the for the users of the platform not speaking about the inventor of the of the platform is that how sustainable is that gay thing that. has any potential as far as the e-commerce is concerned is it possible to eliminate money altogether as some people in this country once i i don't know i don't think it's actually possible to eliminate money altogether but definitely on the internet you're seeing i mean this is not

a new thing for the last five to ten years we have seen a lot of examples of people basically trading non-monetary value i mean a search engine like google for example gives you the entire thing for free and what you're exchanging is you're exchanging the data that you're giving them and also every now and then you click on an ad. so there's definitely a lot of non-monetary exchanges in the internet a lot of the economy of the internet is non-monetary exchanges ok well mr fernand we have to take a very short break now but we will be back in just a few moments and. in two thousand and sixteen the panama papers show the world with a tax haven the secrets two trillion united states dollars passed through most. in the amount of time that we in the panama papers exposure that's what it shows

a lot of money it really is. journalism it's a fact of journalism looking at things that people want to keep secret and asking why would they want to keep these things secret. millions of fonseka documents were examine. all the people we basically have tried to get an advantage out of this sort of. split. and probably other politician which was the other politician the media would point to find targets such as the kings of morocco in saudi arabia the president of the. several prime ministers. and russian president vladimir putin of course. that tied so i have sued so many newspapers for defamation some things don't

just happen by chance he was very striking there were no more american symbols specially a lot of people from the brics countries specially brazil russia and china their special project reveals what was missed in the media coverage. of the panama chronicles. welcome back to worlds apart on your and founder of dual lingo. the ultimate form of crowdsourcing is democratic governance at least in theory and i think there are major issues in both developed and developing countries where how power and

governance are practiced and applied do you think technology can or will ever change the way our societies are not only organized by how they are governed. i think it will is very hard to say exactly how that'll happen but i think it will and i think it's starting to happen we're starting to see i mean we're starting to see a lot of the very nascent efforts but we're starting to see people for example there are programs where people can report. problems in their neighborhood all with with computers and eventually they get to the government so we're starting to see some of that. i don't know what the ultimate thing is but i'm sure if acknowledge i mean technology is already part of most of things in the world i'm pretty sure it'll it'll get to the government i don't know if you would agree with me but i think at the root of bad governance is actually institutionalized inefficiency because there are many powerful corporate actors around the world who have a personal interest in keeping the system inefficient and they essentially institutionalize those inefficiencies. if you were asked to approach the

algorithmically what would you start that's a great question i mean i am i myself was born in a country that has very very terrible government i'm from guatemala the government and what i'm of this entirely incompetent i would say in a country like one of my probably would start by starting from scratch just started the government again from scratch when he trying to start from scratch you still have to start somewhere you know it's like starting a company a lot of times i think it is you start with. you start with the main services that you want to provide in the case of a government i don't know them and services are probably health and education and you start with those and you get really good at that and then grow from there are you also pro need a really good tax collection service as you need to monetize you know as i was preparing for this interview i came across a very interesting poll about people's attitudes towards replacing judges with computers to minimize corruption and the influence of bias and it turns out that the stronger the country support system is the lower is the percentage of people

who are in favor of that in russia or i see why the malo the majority of people would rather be judged by computer is than judges and again the question is technological not political gain a thing you can ride an algorithm for something like the application of law i think so yeah it's already it's already the case that there are computer programs that are giving advice in many legal issues better than lawyers i mean that's already happening this is not for all things but there are many cases where there are for example i.b.m. watson is being used to give legal advice then i saw a study that was starting to become more more accurate than legal then human lawyers advice for certain types of law so over time i think that i think that will happen over time well i think it all boils down to the same question you're had to face bit captioned recapture whether some tasks. how to do what have to be completed by by the humans or better to be delegated to the computers and when it

comes to the society can you really ride an algorithm for power of governance because philosophically i think that's. very interesting i think yes i think i mean i i think it's only a matter of time i don't know how long but i think it's only a matter of time that computers will be able to do everything that humans can now speaking about russia i mentioned before and i know that your community here has roughly six million years or so which frankly speaking is not that much but i heard that russia was the first country within the do a little family to deliberate a community generated language course how did it come about when we started doing ghost the courses that we had we thought spanish and we talk german those were generated by us inside that building a company. but at some point because we got we got asked for so many different languages we decided that it was better to allow different communities around the world to create courses the first one was in fact the course to learn english for

russian speakers and that was generated entirely by community members the way that came about i mean the we left when we opened up this program called the dual lingual language incubator which is where the new courses are created the russian the russian team was the first to apply if you don't organize team war or just the community if users there they they are now an organized team but they didn't know each other they had different people applied and now the you know now they know each other and there's a clear there's a clear leader of the team and they organize it so initially it was a grassroots operation initially it was aggressive operation of people who applied who didn't even know each other they apply to us and we say we found about five different ones that were some of them lived in russia some of them lived in the united states but they were native russian speakers. and we put them together as a team and they have worked very well together well it's interesting you say that because in this kind. where we often amount they bill lack of sort of citizens' activity and the passivity on the part of the fraggle are bogus but it seems to me

that what you're saying is that there is some extra exceptions to that yeah yeah and we're very happy i'm in the russian team house still to this day this was about four years ago that they created the course still to this day they're working on improving it so they've lasted a long time now i know that the one you just launched a lingo you relied on the crowdsourcing to source translations for various google pages and or other websites and i know that the international association of professional translators and interpreters was very unhappy with this and other kinds of efforts because they they were afraid. to lose their own livelihood do you have any sympathy for their argument because i know you stop sourcing because they should but do you think something like this will replace that professions yeah i mean i there are many tasks that there's a spectrum of tasks that are about how hard they are you know to do by computers

some are already entirely done by computers they weren't done before for example i don't know one hundred years ago the first computers were actually. people who were hired to do you know additions or multiplication and they needed that i mean there's a whole story about how you know the space agencies had entire rooms full of people doing the dishes and multiplications that's gone that is now done entirely by computers unfortunately i think translation is not yet there but it is one of the easier things to the right computers that will be gone in the next in the next twenty years most translators will probably be out of a job i mean i'm sorry for them but this will just happen this is inevitable i just want to out of curiosity why did you then have to drop the service so we did not drop it because of their arguments we dropped it because for us we were when we when we started doing well the main thing we wanted to do was to language. the

translation was a side business to try to monetize the teaching of the language one of the challenges for free and we the idea was that if people were learning a language some of them would be doing translation for us. at some point we realized that this is a lot because that's where the money came in a lot of our effort was being spent on translation and we didn't want to do that we actually wanted to teach so we stopped doing that and we decided to come up with other business on a list that were more related to teaching rather than translation that's interesting now another profession that feel that pressure from tech interpreter is like yourself is that of the driver i know that you used to be driving or perhaps still driving a tesla. how far do you think we sell driving. cars very close i mean i think the thing that will take the longest is probably adoption you know people have to buy their cars get rid of the old one so there's a whole cycle of that which may take ten to fifteen years but the technology is very very close so i think that will happen now tesla c.e.o.

elon musk recently generated a lot of publicity it with his warnings about artificial intelligence a lot of c.e.o.'s weighed in on that i also want to ask you whether you consider his warnings perhaps a form of alarmist marketing or did indeed share his concerns i think there are a number of concerns i mean i think you may be a little a little alarmist but i there are a number of things i mean one of the biggest ones is over time more and more people are going to be left without a job and that's a big problem in a lot of our society they get meaning out of life from their job that's not going to happen anymore you know in the future and we need to figure out as a society how should we deal with that and i am you know i'm not too concerned about countries say like the united states because a lot of the artificial intelligence produced there so they're deriving value i mean for example a company like google or a company like facebook is they're writing a lot of value for this which if they pay taxes to the government the government is going to have enough money to you know pay to everybody but poorer countries where

they're not developing this technology they're going to be in trouble because a lot of times the economies for example an economy like my country where one of my lists got amala. our economy is based on cheap labor that's what our economy is based on and if cheap labor starts going away and we are also not deriving any value out of artificial intelligence there's going to be a major problem so i do think i do think we need to think about that another concern i heard many analyst war with regards to artificial intelligence is that. in a way it would be similar to. i government structural or some sort of a social order because sooner or later you will have to make ethical decisions about how to prioritize human lives and again is that the kind of for example if you're using a driverless car and it gets into an accident you have to prioritize let's say the life of which i borrow or the life over the life of the pedestrians is that another capture moment is that something that a computer can deliver

a judgment on or is that something to be left to the to decide an individual decision making i think it's going to be done by computers and i'm glad that it's going to be done by computers because a lot of times. a lot of these decisions when they're done by humans they're not fully thought out i mean a lot of times when you are driving as a human and you're faced with one of these decisions you usually do not have the time to think about all the implications yet in fact that there are lots of studies which show that people tend to prioritize that whole lives over the lives of even passengers will in most of the accidents people tend to go to the right side reaching dangerous the passenger oh and gives priority to the driver so right i'm not just i mean a lot of times they just you know a lot of these this is i just think are not well thought out whereas i think if we were to put this in an algorithm at least we would be able to a society come together and say ok in this case what is the best thing that could happen but then the other is it ok to prioritize a society over an individual can consumer in such

a way because when you buy something of value you buy it for yourself when well that's that's exactly i think the question that we're going to have to deal with i mean what is that what is the right priority but what i'm saying is a good thing is that. we'll be able to make that decision as opposed to now where you know the decisions just be made in this in a in a half a second at least if we put it into an algorithm we will come together and come up with a good answer for them it may not be there may not be a perfect answer but there will be one that we at least them accredit can we agree on. now your answer is somewhat similar to the wand that was given by mark zuckerberg facebook c.e.o. who also suggested the theories of artificial intelligence was maybe a little bit exaggerated but he was immediately accused of not being able to come up with an algorithm that would. be growth of the so-called fake news and i want to ask you the last algorithmic question whether you believe it's possible to develop

such an algorithm how would you define fake news in terms of you know writing a program to eliminate them yeah that's very hard to that's a very hard question because. i mean if i knew the answer i would have done it already have you looked into it have you thought about i have not thought much about it i mean i i thought about it as much as a lot of kind of computer scientists have thought about it but not not specifically why would it be so difficult because it seems the fake news is a very commonly used term especially in the american the environment why is it so difficult to come up with a definition a technical definition for it well i mean maybe a definition is just news that is not true but the problem is truth is very hard to define and why so i mean even you know defining it as one thing finding it is just very hard i mean i can just say anything i want and because i have a lot of followers on facebook some people will believe it we are we are currently having that problem with the president of the united states who just this it seems

whatever he wants and some people believe it and it's hard you know are you claiming that the new york times is failing and if you look at the actual graph of the new york times revenue is in fact going up but as a good fraction of the population believes the new york times is failing just because he says that it's very hard this is it's very hard to change your feel for mr zuckerberg after i do i don't know how to how to you know i don't know how to fix it well anyway we have to live in there really really appreciate your time with us please share your comments on our twitter facebook and you to pay just i hope to see you again same place same time. there was the part.

in the heart of the swiss alps this is a place probably more secretive than the pentagon more mysterious than the cia and better guarded than for knox customs are here permanently all the science is controlled by them and they imposed the opening time so it was it it was it took these forms all plus the procedures in place of the strictest in all europe must to pieces by artists like pecan so and modigliani i can't boards unsold inside this warehouse that's where the report comes in it covers up deals which are naturally discreet commercially discreet felt but also discreet because they concern fraud. some of those paintings are linked to dog secrets nobody knows how many of these secrets they kept inside the geneva freeport system you'll never obtain an inventory of all the works in the freeport who knows how many there are three

hundred three thousand three hundred thousand is it a matter of confidentiality only is it the world's black box of the art business. unfortunately it appears that once people. have been involved in that they become scared if you instead of being scared of the perpetrator. not. besides getting what you eat.

the headlines on. wiki leaks julian. the country's foreign minister pushing. the release of an interview with one of. russia on the program we discussed with the. russian attorney whose mention. of being a broker between. trying. to. get. a new u.n. report says iraqi refugees are being forced by their country's security forces to.