hello freedom to our weekly data talk a show where we talk with data science leaders from around the world today we're talking to dr. Arthur I'm Miller he's the emeritus professor of history and philosophy of science at the University of College London his latest book is called the artist in the machine his book covers how a I power computers are creating art literature and music that way that that may well surpass the creation of humans dr. Miller thank you so much for being on our show today oh thank you for asking me so before we get started I just got a comment on your awesome library in the background I'm using and for those listening to the podcast dr. Miller has what looks like 300 books on bookshelves or more around this room that look it looks beautiful all these different colors how did you arrange your books back there uh by subject and that's it okay floor and back of me it looks really really cool looks like a tapestry all the way looks really really pretty totally you get to the point where you start like piling on top of each other they are so sober get started can you share a little bit about your background sure well I've always been fascinated about creativity what it is and what happens at that moment of inspiration and my interest goes way back to when I was a boy growing up in the Bronx it's a Bronx Tale my growing I grew up in a rather dysfunctional family and so I ended up doing very badly on IQ tests when I was seven or eight years old and that plagued me throughout my public school career I was in classes with not very good teachers and disinterested students was like a Darwinian survival of the fittest information but there was an upside to that the upside was that the work for me was trivial so I had a lot of time for other pursuits and I spent a lot of time in the nearest New York Public Library which in those days was a magisterial building packed full of books and 33 and 1/3 records or vinyls as they call now as well and I distinctly remember a life-changing moment when I was say reading a book and I looked up and there was a record cover with a beautiful pencil sketch I was always interested not always interested sketching the man deep in thought so I decided to take the album home to really copy as a picture I did not the slightest oh I do who the person was it was the composer it turns out to be Tchaikovsky's fifth symphony and absolutely blew my mind and I decided to where did this music comfort I decided to go back in time and then I reach block and that really caught me on on creativity but if in those days if you were smart or thought you were smart you went into physics which I did and I enjoyed the intellectual challenge but my heart really wasn't in it I was really interested in those one of what is the nature of questions like what is the nature of the mind what is the nature of creativity and so I decided to look into switch from elementary particle physics research into history and philosophy of physics and what I did was to read the original German language text in relativity theory and quantum theory books by a text by Einstein Verne Heisenberg Niels Bohr the Great's of the great minds of the 20th century and what jumped out was the importance of visual images in creative scientific thought I decided to look further into that and this led me at the cognitive science namely the questions such as do these images affect thinking and if so how are they created and how are they stored in the mind and to look into this question meant you should look into the similarity between the brain and a digital computer the brain as an information processing system and and so and along the way incidentally I looked in the notions of beauty and aesthetics in scientific research came to the fore and these notions were quite similar to those in art and so and then I came to realize that at that moment of creativity moment of inspiration boundaries dissolved between art and scientists and both think along the same conceptual lines so I look like the next few years I wrote a lot on human creativity and touched on machine creativity and in this book I turn to both source of creativity with some more emphasis on machine machine creativity but this is not one of those you know what is the nature of whiffle awful books it's what is the nature of machine creativity in certain circumstances that is to say AI created art literature and music I thought I if I may one of the other very good questions was resistance to creativity in machines and that's for me after all hunk of machines be created and let me just mention a few glimmers of creativity one early government glimmer of creativity occurred in the 1919 97 chess match between the then chess champion Garry Kasparov and deep blue and the famous 44th move the first great at first game deep blue made a move that absolutely floored Kasparov Kasparov complained that IBM was cheating there's an even being in the loop when they convinced him otherwise now what happened was that the machine which had been programmed with extensive databases in opening games middle game and end games the machine couldn't decide reached a glitch and in order to make the next move it jumped the system and created something new and that's amazing of creativity the and Kasparov said well yeah I sense the weird kind of intelligence then about about a decade later I had this IBM successor to deep blue as IBM Watson which became the Grand Champion at at the quiz game of Jeopardy and again a massive machine with massive databases all the with all of Wikipedia was in it for example tons of articles and books but the the main criticism against both of these machines was that look they're not creative they're they're doing what the program tells them to do this in the face of that great number for 44th move well David Ferrucci who was head of the design team for Watson was fond of replying to criticisms like that with kind of submarine swim the implication was yes not like a fish better uh some years later well in early 2016 I was talking with a and they put in any person who said yeah well we we've cracked the game of chess but gold take another ten years well just a little while later the game of Go was cracked by alphago a machine created at deepmind in london alphago the innards of alphago is called an artificial neural network it said it's a architecture that's loosely analogous to the wiring in our brain well the innards of deep blue and Watson are massive databases like that in fact they're like our laptops only on steroids now whereas deep blue and alpha deep blue and Watson were programmed to play chess and jeopardy alphago learn chess by by studying 30 million board positions of games played by go masters and then played against itself millions of times and it beat all comers machine and human being the big match was when it put in in 2016 when a clay to go master or at least at all and that was about as has people about as excited as that is over the match between Kasparov and deep blue in the famous alphago made many extraordinary moves but the one everyone remembers is move 37 in game 2 which absolutely floored said oh and floored the alphago team as well it turns out that alphago calculated that the odds of a human being making that move were 1 in 10,000 it was a killer move and Lee attributed to the move was beautiful attributing beauty to the machine and from that moment on called the machine her you mean it's sort of human human outlook of the the next machine was alphago zero which learned how to play go and by the way you could have already sequences of creativity alphago alphago zero learn how to play go by just being fed the rules of go and then it played against itself millions of times and it to beat everybody and then came alpha zero which was the epitome alpha zero not only plays go but chess and Japanese and it learned chess by just being fed in the rules and then playing against itself and having no therefore having no contact with the way humans play chess and some of its games are rather surreal it emphasizes attack of maneuverability over materiality that is to say it it sacrifices like mad and people in a play need alpha zero saying that it wins by style and imagination further creativity and also alphago is a harbinger of things to come and that reason is differently from human beings in ways that maybe come incomprehensible to us it's absolutely amazing about a computer that's learning on its own to play a game and able to compete against these these other computers that have huge datasets with millions of other games and being able to then beat these other game these other computers simply from learning on its own you're just feeding it the rules and and now it's making all these amazing moves and like you said being creative and its approach mmm I thought was interesting in your book you talk about this is one of the quotes and I think it was in regards to one of the alphago matches or at one of the yeah one of the go matches where the person said that it was almost like an alien intelligence that the move was not quite human not quite computer yeah that was very dice it does this person would not to say that we can now speak to a machine machine the machine we it's not a matter of language we couldn't no we can communicate with the go machine with with go language yeah these that's a good point machines are like alien beings the we don't have to go to Mars to encounter alien intelligence Yeah right before us alien intelligence with the potential but unlimited creativity from which we can't learn from and also there are these various aren't algorithms in art that can elucidate but algorithms at AI which can be elucidated by means of art and which can also give you submitted an indication of and into some of the deep problems one has with AI such as what goes on in the so-called hidden layers and one of them - deep dream maybe you've seen it it produces these monster these beasts and things that we have that that we could not imagine these images are not in the machines training set but his way the machine sees the world and gives you an indication of how differently machine sees the world you can take a JPEG of yourself and it would identify you through facial facial recognition but if one pixel were removed it would see you entirely differently hmm that has all sorts of ramifications of course with surveillance software and things like the smuggling messages and things of that sort when you first started when he started thinking about computers being creative and since you've been studying creativity for such a long period of time in so many different areas what were your thoughts initially around a machine being creative I thought that was all fine and good there's any problem with that okay creative a lot of people do machines could be quick machines can help our creativity right now is going on there's a lot of what's called intelligent augmentation where people are working with machines and for example if a scientist works with a machine the machine has this huge database it could scan scan the web and pick up lots of and lots of information and help the scientists in that way and of course you know we are emerging with them next we'll come chips in the brain where we will also be able to scan the web and have we'll have all that information so machines are working with us soon they'll work alone and then they'll work with each other I just want to mention you asked about definitions of creativity yeah yeah a lot of those definitions as I point out have to do either with the end product some with a process but a lot of these definitions are very subjective they consider the creative objects out as having novelty and value but that's what's respect to the person I wouldn't expect at the time there were some great stories about that where Einstein in 1905 discovered relativity theory but nobody really knew what he was doing and it wasn't until six years later that it was people realized that he had done something new about space and time and it was Picasso's his great painting like Demoiselle davon yeoman which he did in 1907 and he kind of ravaged by critics for years ravaged by critics as a nightmare another big dog bye-bye Mikasa it wasn't sold until 1925 and then it was resold to to MoMA cuisine from Modern Art in 1938 but for thirty nine thousand dollars it still wasn't thought to be a big deal so but I have a definition of creativity that I think is more manageable and also applies more broadly as creativity is the production of new objects from or the production of new now creation of new knowledge from already existing knowledge and that's accomplished through problem solving problem solving is is the key here there are there are various then and there are four stages involved also in this process one stage is conscious thought where we sit at our desks and consciously think about a problem and then you may run up against a stone wall what you should do in that case is take a break but it's not really a total break because the intense and passionate desire to solve the problem keeps it alive in your unconscious where it can be worked on without barriers that exist in consciousness and what you have essentially is parallel at times intersecting lines of thought that feed off one another that you can realize connections between disciplines that at first sight are unconnected and then you have the illumination some are another bubbles up and then you get back to conscious work then it's a safe verification which is also extremely important and there are various characteristics that have emerged from my work on creative individuals such as perseverance inspiration collaboration competition these are all the cynic wanna know of process and then there are a couple of major characteristics that cannot be there are they these are problem discovery and the connection of apparently unconnected bill made problem discovery is especially important for example 99.99% of scientists artists musicians work in a particular style a particular domain working on problems something once in a while someone comes along and says you're working on the wrong problem this is what you should be doing and everything that comes up that's what happens I sign in 1905 and Picasso in 1907 when he he created cubism Bronstein it was the the two domains that he found to him were connected was the laws of space and time and the laws of how heat flows what could be different to most people but Picasso it was it was as art with science mathematics and technology that's what enabled him to create one of my team's they discover cubism and I contend as I discussed in great deal to detail in my book that machines can have that definition of creativity can apply to machines the characteristics and apply to machines before stages can apply to machines and those two major characteristics the problem discovery and bringing together domains that don't seem to be connected and also one important program an important part of creativity is unpredictability you know bang it happens it's not bang I mean a lot of going on in your head you know you may meet somebody you you can't remember that person that person's name and then you forget it you're walking down the street and suddenly it pops into your mind well your your unconscious your conscious unconscious has been has been working on that's called the tip of the tongue phenomenon in psychology now machines you may say how can machines do anything predict unpredictable since they're constructed of components made up of physics yeah that's also and deterministic but you put all of those parts together the conglomerate can produce chaotic behavior and that's where some machines have in a sense their unpredictability is built into them yeah I think those are those are some really good definitions I love how you mapped out in the first section of your book around these various definitions of creativity and I think my what you just said about some of the factors being perseverance inspiration collaboration competition unpredictability to me these are all very human right human things and it was hard for me when I was reading the book to see like well how can a computer be inspired how can a computer be persevere and then you gave up you gave so many different case studies like the the game examples with with chess would go where think appears making these decisions that were unpredictable certainly persevering by doing these these computers were doing thousands and thousands I don't even know millions of games potentially so showing signs of all these things that we would say are human the perseverance the inspiration and then these unpredictable moves that those are definitely signals of creativity and I guess one of the questions you asked in the book that I like that was challenging to me was you talked about how you said can we appreciate art from a computer and I think that was a question that got me because you were making you were making the case that look machines are showing these signs of creativity look how look at all these different use cases of game gaming art literature music these are all signals of machines being creative being unpredictable doing these really cool things then you ask that question can we appreciate art from the computer and I mean I think that was my stopping point because I was like okay well maybe that's a glitch maybe that you know I can kind of like give answers for reasons why computers were doing nothing that's not really creativity and I think my problem was that that question I think I'm stumbling with a pre 'seeing art from a computer well let me let me talk around that yeah yeah let me go back to a alphago zero alphago rather than that that 37th move if a if a human being had made that move I mean the crowds washing that that match would it but roses won and roaring but a machine made and what often happens that when a machine does something like that the bat you know the the goal lines are moved it's nobody wants to appreciate it one has to appreciate machine machine created products for what they are let me talk about a game I play when I get when I give lectures I call it the Box game and that's the Johan Sebastian Bach I play two selections one by Bach and one by an algorithm is the audience can you identify which one is from Bach Hughes he goes about 50/50 it never is Wow never totally blocked I mean professional musicians can get it because there are some turns in the music that produced by algorithm you know it's different in Bach but you can see how people's faces that you know that the pushback the what is this you just is this is this some some some sort of trick what you wonder how did you how did block do this yeah and you know those of us who have some association with Queens with the piano can I think I track an appreciating box creativity in that if you look at a piece of a composition look at a chord and press keys and an outcome something which would blow your mind anyway yeah how did how did he do it I mean my birthright when Bach was composing it is mine he crossed the line into mathematics and crusted duck again because this is in this in and they make mathematical property to his music but you asked the question can machines producer melody like that and to me that's that's the wrong question because what we want is machines to do something totally unexpected totally different to produce melodies that will someday blow our mind and so a Bacchus but not by Bach and I agree with those researchers who disagree with placing too much emphasis on those plot or not thought games like you play with Bach's music or alright poetry and some lecturers just feast on that much as much much because the products of may I should not be judged on whether they can be distinguished from products of humans because what's the point what you want they ought to do is to produce something that you can't even imagine that you may not be able to understand that you may think is nonsense and at the end of the day may be better than what we can't do so can machines be creative yes absolutely why not we've only recently grudgingly agreed that animals can be created so why not silicon silicon life-forms why only the push backs our looks yeah and I like the example of the the Bach example you gave because part of the appreciation is really understanding the music and what's happening there and because I don't have a musical background like my my appreciation for Bach is gonna be at one level but people who have like studied music and like the way you said a kind of an understanding of the maybe the or the music is going the movements and the variety and the creativity they're like people who who understand music like they're gonna see a level of genius that I won't see right it's similar with mathematics I'm not strong with mathematics but people who are really good at mathematics may look at certain equations or algorithm to see beauty right because the beauty of aesthetics in yeah they'll still see feet in and for me not having that background I'm like oh cool right but there are at a different level wrong so I think sometimes that's where similar with go I don't understand the game go and there's all this amazing movements moves being done and go and so people who know the game go are like oh my gosh look what the computer did at this move this was so unpredictable this was a genius move it changed the entire course of the game and for me looking at I'm going I'm gonna trust you on that because you know the game but I think like I can definitely relate to your these cases around you know the music and art visualization because I'm a very much a visual person so those those things really resonate with me but I think you're right part of like can we appreciate the art that machines create part of its gonna be well how much do we know about that type of art because can we see the beauty in it it'll be different I mean their art will be there but right now they are pushing artists into into other areas so I mean just as machines are affecting literature of course of literature affecting me the course of art and music as well pushing there changing changing the landscape and soon we you know maybe there'll be a gradual transition probably will be a gradual transition from what we're used to hearing to this totally new aesthetic I also think it's interesting around machines creating literature and you shared some examples in your book are there any favorite examples around machines developing text that is meaningful and creative are the each AI person has tested her own favorite text no there is no one favorite one I mean when I first started to write this book I I knew a lot about AI and music and AR and art but not a lot about AI and literature and I said to myself am I gonna have enough to write and what's more enough absolutely fascinating now what I find fascinating is that the text that that machines produced use language in a way that we're not used to and so we're seeing we're seeing sort of an an open-ended transition in language it's what is sense and what is nonsense is something that is being a lot of people I mean I want to computer poetry I mean strike some people as nonsense but it's not that much different from the beat poets or what one reads in avant-garde poetry poetry journals it's it's another way again it's another way of looking at the world it's another way of looking at the linguistic landscape it's an investigation aoi people consider AI poetry as a way of investigating semantic space yeah and I think about I think about also things like urban dictionary because we're language is always evolving like we're always changing words you know we pronounce words like this is about language is that's it's constantly evolving you look at English language you know in 1280 I look at it today entirely differently and it's continuing it's gonna continue to evolve and right yeah I think I said yeah it's right that's right dibban and who you talk to I mean we're changing languages I mean Twitter has changed our language there was an article couple days ago on a time New York Times about how you know teenagers are using Twitter these days it's called the key smash and then they send it as a tweet and then they'll say the other person say well maybe you should remove such a session then the world would become more pronounceable and it'll be used in a certain context so that's you know machines are teaching the way we are looking with the way we are looking at the world and these machines if I could change the subject the bits not that much machines kind of emotions all this will be done with emotions as well machines can be out there that's always a favorite criticism machines are not out there so you can't put this real art you know this is its unarguable that machines could produce our kind of machine be an artist well I contend yes that machines can be out there by scanning the web and when machines have better and read the natural language processing they'll be able to do that better and better too to deal with puns and tropes of puns and tropes of all sorts machines will then be will soon be programmed with instantiated what programs I'm certain next time there will be instantiated with regulatory mechanisms sensors communication pathways and they'll be able to evolve human emotions duplicate human emotions and then go on from there to evolve a set of emotions that's more palatable to a silicon life woman who knows what that will be but one thing that we we need to do and in language in communication is we need to communicate better with assistants like Siri and Alexa we that we need to we need machines with a sense of humor it's considered to be the final frontier because and humor everything is there all facets of intelligence are there there's a lot of research being done fascinating research being done in that area which I discuss at some length in my book as well as scripts for a screenplays and different kinds of poetry yeah I think it's I think it's very exciting I I'm I'm looking forward to seeing how his technology improves and and how we interact with our voice assistance I and I think you're right like it'll be it'll be fun when we start to hear some emotion coming from these voice assistants I think they would be actually kind of funny sensing that and I was actually thinking about love your your chapter dealing with yourselves and dealing with emotions and computers also you can oh yeah because there's always this image of the artist or I'm a writer and I love to create but there's always that that moment of hesitation you're sitting at the keyboard looking at that word doc that's blank and you're just like you're about to write something but there's a little bit of fear holding you back and you're just questioning yourself whether I should go ahead and type that thing or you type that sentence and you're like no no no gotta delete it and I love the idea of an emotional computer that's creative like that that's like hesitating oh there there add their algorithms now that will go on you know you've typed and it will go on from there like you know stuff that's that not even true love database is truly fake news taking from databases of you know millions of words and I think it's not open AI has an algorithm which which they wouldn't release because they say it's just too powerful but I've seen some of the less sophisticated versions of it and yeah you you were you type in a seed sentence and then it will go on from there but do something you know that this looks cogent but it's not really but actually them a lot of Reuters reports and sports and sports columns are because there are templates for that mm-hmm but using artificial neural networks which are not you know powered up with all these algorithms that's that's where I think the real experimental part of a oh I writing occurs this machine has much much more latitude yeah it's also funny to like there's some I felt what the technology is called I think what sits within Google when you're writing an email and it'll prompt words that it thinks you're gonna write next and it has all these predictions right so you're writing hi mom how are you it's like automatically filling in the blank how are you like it already kind of knows probably what you're gonna say um your phone does that too yeah that's right you mean the next word yeah and so it definitely speeds up the process and it can make all these predictions based on the next words you're gonna be typing in or even phrases based on what you've written in the past so it's very a lot of a lot of business letters are pretty much the same and there are algorithms to you know type a whole type 15 different business letters which are essentially the same but they're varied so that people in the same office think they're getting their own letter that's right that's right and we were I was talking to a company that does they uses AI for for email marketing where you go ahead and you put in your data set of here like here are subject lines and here are emails that have worked very well for us and it'll just begin to like cycle through and just test different variations test different words to then improve itself to make it's like paints yeah makes a document for you test exactly exactly and it just kind of runs for you and I think automation can be very beautiful like that okay I I'm like looking for the day cuz I work on the social media side and we develop lots of different pieces of content to promote videos and articles and you know sometimes it's pulling a quote out of an article or tweaking a headline or using the headline but I'm like like you know it's just a matter of time before an AI can just hey Mike here's ten different versions of tweets that we can send out on your behalf let us know which one you like yeah yeah Ryan Oh save me like lots of time this I'm like I'm looking forward to like these these this creative technology like empowering us because it'll save me time and I can work on other things don't we lots of tweets out there too yeah exactly like with job applications I mean that's what computers are for I just smash out hundred of them all around the place yeah and I think you know as you talked about your section on emotion I was thinking about how what I'm also excited about is seeing how a eyes can help with therapy alright somebody because therapy can be very very expensive to go see a therapist and it's very time-consuming and sometimes you can only spend an hour with a therapist because the therapist is to move on but if you had a eyes that were you know eventually trained to be therapists it would I don't think I'll ever replace a human therapist but oils but for those right you can't right sure or they don't have the money like what what a great thing that someone can then talk to about there's going to be a different kind of togetherness as we as we move along I mean marriages may be out the window who knows there sexbots etc so there'll be a totally different kind of togetherness in the world totally different sort of emotion sepia emotions I mean what what is it what is a human being has changed over over the millennia and it still is changing now and will drastically change it will change very quickly in it best in a Darwinian sense because of implants for example silicon implants I mean silicon ships can't communicate with with neurons in our minds you you can your brain can be taken apart piece by piece and silicon chips can be a pleasant how it's amazing yeah that the future is rapidly coming and it's very fascinating to see or how to think about what is possible when these chips this technology can be added to our brain like that that's like so to me is very far-fetched I'm like I know people been talking about it but it's it is amazing I mean we're already using our mobile phones so much and now voice assistance so it's just that next step I to you know consider is a I would say our yeah that's great friends the book is called the artist in the machine it's chock-full of case studies that of research around how AI is being used to create art in various ways there is so much there's so many case studies I found the book fascinating I really really enjoyed it and if you're skeptical about machines being creative or you're curious about the future of machines and art this is a book you need to pick up again it's called the artist in the machine I created a you can just google it the artist of the machine you'll find the book on the MIT website you'll also find it on Arthur's personal website and also on the URL the artist machine dotnet is that right Arthur sure that's the book looks like great the artist in the machine net or again you can just google it or just ask Siri I look at a book gets directly and order it or Sir Arthur thank you so much for being our guest today dr. Boyd take care you do hey everybody thank you so much for checking out this episode of the dating our podcast my name is Michael Gatto and we record this show every single week on Facebook live if you found this show helpful please let us know by rating us on iTunes and if you want to know about past episodes or upcoming ones you can always go to the experience the URL is just experienced comm slash dater talking again it's Experian comm slash get a and that will bring you over to the page where we have a listing of all of our show notes and past episodes we recorded over a hundred so far and we're also looking for your recommendation so we ever have suggestions around topics or guests you'd like to hear from please let us know you can reach out to us on Twitter our twitter handle is Pat Experian data lab or you can just reach me directly at Mike Doug Otto thanks so much for your support and thanks so much for subscribing to our shows