Stephen Wolfram's Take on Artificial Intelligence & The Future of Humanity

some top questions I mean one of them is about the future of the human condition there's a big question it's I spent some part of my life figuring out how to make machines automate stuff that's pretty obvious that we can automate many of the things that we humans have been proud of doing for a long time and the question is then so what's the what's the future of the human condition in that kind of situation so more particularly you know I see technology is about taking sort of human goals and making them be able to be automatically executed by machines and the human goals that we've had in the past have been like you know move this object from here to there use a forklift truck to do it rather than our own hands now the things that we can do automatically are more intellectual kinds of things there are things that have traditionally been the professions work so to speak these are things that we are going to be able to do by machine and the question and so sort of the flow is the machine is able to execute things but something or someone has to define what should its goals be what is it trying to execute and people sort of talk about well you know what's the future of the intelligent machines and you know are them intelligent machines ninety you know take over and decide what to do for themselves and things like this the people think one has to realize is that while sort of figuring out given a goal how to execute it it's something that can meaningful be automated the actual inventing of the goal it's not something that in some sense has a a path to automation that is you know when we say what what makes how do we figure out goals for ourselves how do we you know how our goals defined they tend to be defined for a given human by their own personal history of their cultural environment the history of our civilization things like that the holes are something that is sort of a uniquely human kind of thing it's something that almost doesn't make any sense we ask you know what's the goal of our machine well yeah we might have given it a goal when we built the machine actually the thing that makes this more kind of poignant for me as I spend a lot of time studying basic science about computation and I realized something from that that's a longer story that basically the the question would be if we think about sort of intelligence and things that might have goals things that might have purposes what kinds of things can have intelligence or purpose or something of that kind right now we know one great example of things with intelligence and purpose and that's us and our brains and our our own human intelligence and so on question is what what else is like that and the answer which I had sort of at first assumed was that well there are these systems in nature and they do what they do but sort of human intelligence is is far beyond anything that sort of just exists naturally in the world it's something that's the results of all of this elaborate process of evolution and you know all this kind of thing it's a it's a thing that we can sort of that really stands apart from the rest of what exists in the universe what I what I realized as a result of a whole bunch of science that I did was that that isn't the case but that even when we look at well a funny version of this my children always give me a hard time for this particular quote is you know one says the weather has a mind of its own well you know that's a sort of an animistic type statement and it seems like it has no place in kind of modern scientific thinking but actually that statement is not as silly as it at first seems that's what what to me what what that's representing is if we think about a brain what does a brain doing the brain is taking certain input it's computing things it's causing certain actions to happen it's effective at generating certain output we can think about all sorts of systems as effectively doing computations whether it's a brain whether it's you know a cloud responding to a different thermal environment that it finds itself in these kinds of things and then we can ask ourselves you know are our brains doing vastly more sophisticated computation than happens in these fluids in the app or whatever else and I ever first assumed that the answer that was yes you know we're we're carefully evolved we're doing much more sophisticated stuff than any of the systems in nature but actually turns out that's not the case it turns out that there's this sort of very broad equivalence between the kinds of computations that different kinds of systems do and so that that realization so it makes the the question of human condition a little bit more poignant because we're we might say well look we've you know that's one thing we've got this we've got we're special we've got all this intelligence and all this all these things which which nothing else can have but that's not true you know there are all these different systems in nature that are pretty much equivalent in terms of their sort of computational or for that matter intellectual kinds of capabilities but then the question that you know what makes us different from all these things well what makes us different is the particulars of our history and our which gives us our kind of notions of purpose and goals and so on so this is sort of a long way of saying when we have the the Box on the desk that you know thinks as well as any brain does it the the thing it doesn't have intrinsically is the kind of the goals and purposes that we have because those are really defined by our particular sauer particular biology our particular psychology our particular kind of cultural history and so on so you know I think that the the thing we have to think about as we as we sort of think about sort of the future of these things is there's goals that's what humans contribute that's what our civilization contributes there's execution of those goals that's what we can increasingly automate we've been automating it for thousands of years we will succeed in having very good automation of those goals and you know I spent some significant part of my life building technology to essentially go from a human concept of a goal to something that actually gets done in the world there are many questions that come from it so an example of one that I'm thinking about a lot right now it's okay we've got these great AIS and they're able to execute goals but so how we tell them what to do so one answer to that is well just talk to them I think so you talk to what mouth or Siri or whatever you know it's we are understanding the natural language their human utterances and we're doing something based on those utterances and it works pretty well when you're sort of making a when you're holding up your phone and you're asking one question it's a pretty successful way to communicate to use natural language when you want to say something longer and more complicated it doesn't work very well I just have this experience I've been interested in teaching programming to the world and to kids and so on so I was just writing this book and I was writing exercises something like very bizarre thing for me to do done exercises myself I think in any textbook but anyway I was writing exercises and these exercises are typically will inform you know write a piece of code to do X ok and at the beginning of the book when they've done the you know the exercise are really simple it's pretty easy to write the English to say I write a piece of code to make a list of numbers from 1 to 10 or something by the end of the book it was getting bizarrely frustrating because I was thinking this is the exercise want to write I know what the code is supposed to be now how on earth am I going to write a piece of English text that represents that code and what I increasingly realized as some of this text was starting to sound like you know the language that you would find in a patent or something like this you know some very ornate precise you know kind of stylized English and so the realization from that is oh the thing I've spent a lot of part of my life doing which is building computer languages it's actually not such a bad idea because in a computer language you you do get to represent you know kinds of more sophisticated concepts in a clean way which can be progressively built up and way that isn't possible in natural language so one of the things that I'm interested in there is kind of this question of ok so how do we how we communicate goals de-ice how do we talk to the AI so to speak and that's and my my basic conclusion is that it's sort of a mixture human natural language is good up to a point and human natural language has evolved to describe what we actually typically encounter in the world things that exist from nature things that we have chosen to build in the world these are things which in natural languages have evolved to describe but there's there's a lot that well there's a lot that actually exists out there in the world which in the natural language has not doesn't have descriptions yet even though our eye systems might effectively find those descriptions we don't have ways to say those ourselves but also we have a when it comes to describing more sophisticated things the kinds of things that people actually sort of build big programs to do that's we just you know we we don't have a good way to describe those things in human natural language but we can build languages that that do describe that so one of the things I've doing that many different things that have tested in so an example of a question that um I've been and curious about recently but I'm going to get back to there the main the main thread in a second but one question I've been interested in is what does the world looked like when most people can write code so we have a transition maybe five years ago or something when you know from a time when only the scribes so to speak in a small set of the population could were literate and put right you know natural language today a small fraction of the population can write code most of the code they can write is really for computers only it's not code where any human is expected you know you don't you don't understand things by reading code if you know but there will come a time when when as a result of things I've tried to do particularly where the code is high enough level that it is a sort of minimal description of what you're trying to do and it's something where for example when you have a contract you write contracts you know they're written in English you try and write Maine English as precise as possible you know there will be a time when most contracts are written in code where there's a precise representation that you know it might be for a cases where it's a computer says can I use this API to do this well that's some service level agreement that's going on there so isn't a human contract to human it's something that's written in piece of code that is understandable to humans but also executable by the machines so that you don't you know this question of whether you know does this can I do this according to this contract it's an automatic question and that's something well you know that's a one tiny example of how the world starts to change when sort of most people can write and read code and I think the interesting sort of language point is today we have computer languages which for the most part are intended for computers only they're not really intended for humans to read and understand they're intended for to tell computers in detail what to do and we have natural language which is intended for human to human communication and I've been trying to build this knowledge based language where it's intended for communication between humans and machines in a way where humans can read it and machines can understand it too and where we're kind of incorporating a lot of this sort of existing knowledge of the world into the language in the same way that in human natural language we are constantly sort of incorporating knowledge of the world into the language because it helps us in communicating things but so you know what one branch that I'm really interested in right now is this question of what if the world looked like when most people can read and write code another you know coming back to the sort of main question of kind of okay so what's the future of the humans in the world where well once we can describe what we want to do things can get done kind of automatically so to speak you know what do the humans do and um there are you know I've been I've been kind of interested in one of my hobby projects is trying to understand the evolution of human purposes over time so today you know we've got all kinds of purposes we we like you know we sit and have a big discussion about purposes which presumably has some purpose we you know we do all the different things that we do in the world if you look back you know a thousand years people's purposes were really different I mean it's like how do i you know have my food how do I prevent you know how do I keep myself you know safe all these of things which in the you know modern Western world for the most part you know those purposes have kind of you know you don't spend a lot fraction of your life thinking about those purposes so you sort of evolved to different kinds of purposes and the from the point of view of a thousand years ago some of the purposes people have today some of the things people do today will seem utterly bizarre like like one that that I'm was I think of what kind of treadmill every day right now imagine from a thousand years ago say well somebody's going to spend you know an hour walking on a treadmill like what a crazy thing to do why would one never do that well I think then you know as we look sort of to them one of the things that amuses me in today's world is the fraction of people who play video games that take them back to the Middle Ages so to speak so we're kind of you know we think about so the question is you know what happens in the future what do people do in the future particularly what do people do in a time when a lot of lot of purposes that we have today are generated by scarcity of one kind or another we have you know there are scarce resources in the world you know people want to get more of something so on there is guess time in our lives and so on you know eventually those forms of scarcity will disappear I think the most dramatic discontinuity will surely be when we achieve effective human immortality which whether it's achieved by biology or digitally is not clear interesting question but that is something which i think is pretty inevitably will be achieved and an awful lot of current human purposes have to do with well I'm only going to live a certain time so I'd better get a bunch of things done so the question that I am curious about is what does it look like at a time when when so things can be executed automatically if you have a if you have a purpose in the exterior automatically you don't have the kinds of drivers for purpose that we have today what what does it look like and two people end up you know there are some bizarre hypotheses one might have one hypothesis is what people will look back to a time when there was scarcity when people could say well what do people choose to do at that time justice for a very long part of history and even to some extent today people look back to you know to antiquity do you know the religions created long in the past and so on and say well you know when those things were created and you know people were really have had the important issues going on let's look at how they resolve them at that time and one of them or one of my more bizarre hypotheses is is today is sort of the first time in history of which most a large fraction what goes on the world is being recorded in some way or another and so there will be in the future you know this this is the first time in which that's been brutally happening and so one of the you know one of the things that could happen the future one survey the current set of purposes aren't really issues anymore people would say well at a time when people really did have you know scarcities of various kinds what did they choose to do let's go study that time as carefully as possible and then every detail of what we do in our time which ends up getting recorded ends up becoming sort of further for well that's what it really means to be a human with purposes let's go do what they did in 2015 or whatever I think that's a slightly you know I think that's a slightly extreme version although you know but when we look at the the large span of history and going back to the kinds of places people who look to you know purposes from a few thousand years ago it's not quite as crazy as it might have first seen but you know I think one of the issues is one of the things I I would like to have a a great answer to is okay so what do the you know derivatives of humans of the future what do they end up choosing to do with themselves and it's some you know one of the one of the potential bad outcomes is well they're just playing video games all the time you know would be that the future of civilization is everybody's playing video games you know they're playing World of Warcraft of the future so to speak the sort of the history of AI it's kind of a it's a funny history and it's sort of an evolving word in its use in in in technical language so to speak in these years AI is very popular in people some idea of what it means so we can talk about AI and people have some actually what we're talking about I watched this evolution over the course of probably was it must be basically 40 years now and it's gone from being well in the sixties before I was aware of what was going on everybody thought let me even go further back back when computers were first being developed in the 1940s and 1950s the typical title of a book about computers are in order an article about computers in newspaper was giant electronic brains the idea was that just as things like bulldozers have automated and steam engines and so on automated mechanical work so computers will automate intellectual work there'll be a a giant art tranq brain that promise turned out to be harder than people expected people didn't know what was involved in making brain like activity and it turned out was very easy now they're even like kung they're amusing movies from 1950s computers as a eyes got into sort of science fiction ish portrayals from long ago that's one cute wonderful desk set which is about basically an IBM computer being installed in some company and making everybody not have a job to do and so on it's kind of cute because the computer gets asked a bunch of reference library questions and it's movie and done so as we were building Wolfram Alpha one of the questions was can we do all of the reference library questions from the desk set movie back in 53 or whatever it was and actually we could do them all finally in 2009 turn the thing that that happened was so there was first a great deal of optimism that we could automate intellectual work in the same way as we've been able to automate mechanical work and a lot of government money got spent on that in the early 1960s and so on and it basically just didn't work and and things happened like there was this particular approach to well normal networks have been discussed particularly by McCulloch and Pitt's in 1943 and they've kind of come up with this model for for how how brains conceptually formally might work and they made the observation that they're sort of brain like model would correspond to being able to do kinds of computations like Turing machines and they had the idea of they knew about the universal Turing machine idea from Alan Turing from 1936 and so from that it came emerged well we can make these brain like neural networks that will be able to be general computers in fact that thinking was the way that Turin's work on universal computation flowed into the practical work that was done by the Antioch folk and by men and people like that on practical computers it didn't come the record from Turing machines it came through this sort of sight road of neural networks but then people didn't they set up simple neural networks and the simple neural of books didn't do terribly interesting things there's a guy called Frank Rosenblatt who invented these things called perceptrons which were kind of one layer they're all networks and then there's sort of terrible thing that happened in the 60s to neural networks was Marvin Minsky and Seymour Papert wrote this book called perceptrons where they basically proved that perceptrons couldn't do anything interesting which is correct that proof was absolutely correct they can only make sort of linear distinctions between things the problem was that people and this is a typical sort of academic trait or something that people said well these guys have written a proof that these neural networks not do anything interesting therefore no neural networks can do anything interesting so let's forget about neuron that works and so that that happened for a while meanwhile there was sort of an old there were there were these couple of different approaches to AI one base stones that are really understanding at a formal level sort of symbolically how does the world work the other based on kind of doing statistics and kind of probabilistic kinds of things and there was sort of a a well are we going to be able to do symbolic AI and one of the sort of test cases of that is can we teach a computer to do to do something like integrals teach a computer to do calculus that was sort of a a test case from the late 1960s AI and then there were things like machine translation that people thought would be good example of what computers could do things like this anyway the basic bottom line was by the I guess the early seventies that started kind of crashed and there was a phase where there would be sensible exploit systems which were the next round of AI which came up in the late seventies early eighties which were sort of teach the Machine from a human have a machine learn the rules that an expert users to figure out what to do and so on kind of petered out fact my my first company ended up somewhat against my wishes going into somewhat into that direction in the end may kiss the the that was some that was the next phase and then kind of AI kind of became this crazy sort of phone nobody really does that it's a fake thing that doesn't you know there's nothing interesting that but for quite a long time there's been a this question of AI I myself have been interested in some how do you make an AI like thing since I was a kid basically which is depressingly long time ago now term I was interested in particular and how do you sort of take the knowledge that as humans accumulate or have accumulated in our civilization how do you automate kind of answering questions on the basis of this knowledge and so on and I thought about this first actually around 1980 and I sort of thought about how do you do that sort of symbolically by actually building a systematic system that can break down questions and turn them into symbolic things and answer them or and I kind of concluded well to really do this well we have to have sort of a brain like thing that involves sort of fuzzy questions fuzzy answers these kinds of things and I thought you know building a brain is kind of hard I've worked on it a bit I you know worked on dawn that looks even at that time doesn't really make much interesting progress kind of put it aside for a while I kind of have this approach of having these you know difficult projects which I try to think about every some number of years and try and figure out you know is the world it's the ambient technology and the world ready to actually do this now and so I backin now in the mid you know 2002 3 ish timeframe I was like okay I should think about this you know make a sort of computational knowledge system I should think about that question again what does it take do it and I realized that actually the science that I had done pretty much showed that my original belief about how long had to do this was completely wrong in my original belief had been in order to make a serious sort of computational knowledge system you first have to build a brain like thing then you have to feed it knowledge just like we learn things in a standard education and then you'll have sort of a good computational knowledge system but what I realized as a result of a bunch of science that I done was that sort of there wasn't this was talking about earlier learning that there isn't sort of this bright line between what is intelligent and what is merely computational it kind of assumed that there was some magic thing so the you know that the transistor of intelligence or something there was this sort of magic mechanism that allows us to be you know vastly more capable than the other thing but it's merely computational and it turned out what I kind of showed scientifically that's just not the case so you know one of the challenges always in in something like me at least is how do you take this kind of basic science sort of almost philosophical conclusions and actually decide to do something on the basis of do you actually you know take that philosophical dog food and believe in it and for me taking that was okay so I actually build technology if it's possible to do this actually build a technology stack that actually does it and so that's what led to a mouthful for example and what I discovered from that is that yes it really works to be able to take a a large collection of sort of acknowledgement in the world and automatically answer questions on the basis of it using what are essentially merely computational techniques now there's a footnote to that which is kind of an important footnote which is that when one thinks of what is merely computational one often thinks okay one's writing the program how does one write a program well a programmer sits down and they say I want to write a program that does this alright this module I'll write that module I think about you know how am I going to achieve what I'm well I'm trying to achieve with this program memory every sort of step has you know it's I'm taking one step at a time to get to where I'm to go the the what I discovered was that was sort of an alternative way to do engineering which is something much more analogous to what biology does in an evolution and so on which is just to say out there in this computational universe of possible programs as an infinite number of possible programs if you just go out and look in that in that space of possible programs even just look at random at a trillion programs and say what do these programs do one might have thought that programs which are simple enough that one can actually have good coverage of so all possible programs were given kind but might think that one just none of them would do anything interesting they'd all be just simple programs that do simple things who cares but actually what I had found scientifically was that that wasn't the case and that even very simple programs particularly I looked at Southern automata but also Turing machines lots of other kinds of things even very simple examples of those kinds of programs can already do very sophisticated things and one of my conclusions was that's really interesting in terms of understanding how nature works but that's also important in terms of finding technology in effect what we what we normally do when we build a program is we sort of build step by step as piece of technology the other thing we can do is just go out into the computational universe and mine technology out of the computational universe you know typically the challenge is the same challenge that we face in doing physical mining that is we go we find this amazing supply of I don't know let's say iron with magnetic properties or cobalt or something that some gadolinium with some special magnetic properties say okay great it has these wonderful magnetic properties what do we do with this well the question is can we connect that that capability to an actual human purpose something that a goal that we have to something we want a technology in the case of magnetic materials we have plenty of ways to do that what we find is that there are all sorts of wonderful things in nature the question is can we in train them into our technology by finding some useful human purpose that they achieve and in terms of programs sort of the same story there are all kinds of programs out there even very tiny programs that do very complicated things the question is can we in train them for some useful human purpose and this is a thing that we learned have do have you know given a particular purpose given a particular goal just go exhaustively search a trillion programs and find one that does a useful thing for that purpose and sometimes those programs are doing things like making random number generators hash coding systems doing things that have to do with natural language understanding sometimes they're doing more creative things like one thing we did but years ago now was having a music generation system why you just basically press a button will go search a big space of programs it will find a program that according to the same heuristic matches some particular musical style and we'll play you that sort of an interesting case actually this is a case of kind of automated creativity people say well you've got these machines they'll never be you know if there's one thing that humans are better at it's it's being creative actually the thing that I find most interesting with that is a creation site is that I had kind of assumed that people would say oh you know I need some inspiration you know composers and so on would say I need some inspiration about my composition and then maybe I can dress up that inspiration using a computer but instead I run into people who say it's kind of a nice site that you have I go there to get inspiration that's some kind of you know a little core of a tune which I then dress up as a human to make it be meaningful and put it into what I'm trying to do so it's kind of a case where you know we're seeing that this attribute of sort of originality creativity is something that is is is readily available in its computational universe it's so the same thing as saying you know go out into the physical world and go find these these beautiful places to photograph so to speak in the world they exist already the question of us picking one we care about to look at but I mean back backing up to this question of so what sort of the the arc of AI so one of the things that we discovered was that it really does as a practical engineering matter there's really a lot that you can do by basically discovering programs in the computational universe of possibilities rather than merely building a program step by step what we also spend a lot of time doing is building this thing called language which is this knowledge-based language which tries to sort of incorporate the knowledge of the world right into the language so so kind of the traditional approach of computer languages is to say let's make a little computer language that represents the operations that computers intrinsically know how to do allocating memory you know setting values of variables you know iterating other things changing program counters whatever else is so the slightly higher level version of that but it's fundamentally once once telling computers to do things in their own terms and that's been kind of the tradition of basic programming languages for 50 years my theory about these things is let's try and make a language which handles not so the computers but to the humans and try to make a language where the languages as much as possible just being able to take kind of what the humans think of and convert it into some form that the computers can understand and part of what the humans think of is the humans know of the world they know about you know the existence of Cambridge Massachusetts or they know something about you know that there'll be a sunrise tomorrow type thing and the question is can you encapsulate the knowledge that we've accumulated both in science and in the collection of data in the world into a language which we can use to to communicate with computers and and that's sort of the big achievement of my last 30 years or something has been to be able to do that and one of the things that the significant there is when you're trying to solve the sort of problem of doing computational knowledge having such a language that's the way you need to encode sort of things about the world and things you can you can do in the world but in terms of the sort of the arc of AI so one sort of set of things that would be considered sort of very a iishe is being able to take the knowledge of the world and be able to answer questions on the basis of the knowledge of the one has of the world being able to and so you know there's a whole list of things people with seven six days you know when we can do this we'll know we have AI when we can do a some you know an interval like from a calculus course when we can do this than that well know we have AI or we can sort of do a conversation with a computer and have it seem like a human well you know at this point one of the things that had seemed to be difficult there was well gosh the computers doesn't know enough about the world you know you start asking they you know the computer what day of the week is it I might be able to answer that whose president probably can't answer that these kinds of things and at that point you kind of know you're talking to a computer not to a person at this point but it comes to these sort of Turing test conversational tests of AI people who've tried connecting for example of an alpha to the Turing test boss they lose every time because we really have to do is start asking it sophisticated questions and it can answer them and not even can do that at the time you've asked it a few different you know disparate kinds of questions there'll be no human that that knows all of those things yeah the system can know them so in that sense we've you know we've achieved really good AI at that level now there's another branch which is there's certain kinds of tasks that are sort of very easy for humans that have traditionally been very hard for machines standard one is visual object identification what is this thing I we can know what this is we have some easy description of it but the computer is just hopeless of that well in the last basically year I've completely changed so for example in in march/april sometimes over the spring we brought out a little image identification system website etc and by companies have done somewhat similar things I think I was very somewhat interesting reasons other people's it doesn't deserve to be better but it happens to be somewhat better um it and what it does is you show it something and for about 10,000 kinds of things it will tell you what it is and it does a pretty good job you know you can it's it's fun to try and confuse it it's fun to show it an abstract painting and see what thinks it is but it basically does a pretty good job of saying what it is how does it work it works using the exact same technology basically that McCulloch and Pitt's kind of imagined in 1943 lots of us worked on in the early eighties but no one that works and so question is what was it what what happened that made it work now didn't let it work that well if you look at what the system actually does today there it may need 5,000 controllable nouns in English common nouns which you can make pictures of maybe 10,000 if you include so much specialized things like special kinds of plants and peoples and things that people can with some frequency recognize well the thing that we can now do is we you know we train it on 30 million images of all these kinds of things and it's this big complicated messy neural network probably doesn't matter much what the details of that neuron network are we do this training takes about a quadrillion GPU operations to do the training and at the end of it it does a pretty good job of recognizing 10,000 kinds of things and we as humans are impressed by this because it's pretty much what we humans can do and it pretty much had about the same training data that we have it's about the same number of images that you know a human would see in the first couple of years of their life it's about the same number of operations that have to be done to do the training it's about the same number of neurons and the kind of at least the first levels of our visual cortex the details are all different the actual way that these artificial neurons work is is little to do with the way that that actual neurons in the brain work but it's conceptually similar and there's a certain kind of user you know a salty to what's going on that you have this essentially a really what it is at a sort of more mathematical level it's it's this that's a composition of a very large number of functions that have certain continuity properties that allow you to effectively use calculus methods to incremental a train the thing and once you have those attributes you you it seems can end up with something that does the same kind of job that we do in doing the geologic recognition and you know it's actually interesting because back in the 80s people have very successfully done OCR optical character recognition so they were able to take you know the 26 letters of the English alphabet and so on and say okay is that a is that B is that a C and so on that could be done for 26 different possibilities but it couldn't be done for 10,000 possibilities it's really just a matter of the scale of the whole system that makes that possible today as a kind of a a have we talked to AI yet um you know this is these are important components I mean there are basically a few of these this there's no object recognition there's voice to text and there's language translation and those are kind of three kinds of things which humans manage to do with varying degrees of difficulty I can't do language translation for any I mean human language really may be massive a little bit but you know people can learn to do human language translation um these other two you know voice to text well people learn to do that and there's a lot of recognition people are sort of they learn the first couple years of life like it or not to do that so so the thing that we're these become essentially these are these are some of the missing links to how do we make machines that are kind of human-like in what they do and for me one of the interesting things has been sort of incorporating those capabilities into a precise kind of symbolic language and there's a whole lot of stuff to say about some that is a kind of a 500-year story about what we now need to do in terms of having a symbolic language to represent the everyday world we now have the capability to say you know this is a glass of water or something now we actually have to find a good but we can we can go from picture of glass of water to the concept of a glass of water now we have to have some actual symbolic language to represent those things and you know in my own efforts I you know started off trying to represent kind of mathematical technical kinds of knowledge and then went on to lots of other kinds of knowledge I'm sort of I think we've got a pretty good job done now of sort of systematic objective knowledge in the world now the question is to represent kind of everyday discourse and the kinds of things that people say to each other in a precise symbolic way and there are certain kinds of you know you might have enough in a precise symbolic representation you might say X is greater than 5 okay that's a sort of predicate you might also say you know I want a piece of chocolate that's also a predicate it has the but it has an I want in it rather than a you know sort of chocolate has higher calorie value that and such and such and so we have to try to find a sort of symbolic representation a precise representation of these kinds of things that we have traditionally expressed in in human natural language and actually I've been interested in this this is one of the things like thinking about these days and it's kind of kind of interesting because I you know I'm I like I like to do my homework and I like to find out what other people figured out about this so I start reading literature about this and most of the literature points back to the 1600s and there was a lot of people like wide nets in the late 1600s and I got John Wilkins these were the people who had there's this period when there were these things that they called philosophical languages and the idea of a philosophical language would be it was essentially what I'm now trying to do as a symbolic representation of the world but one thing that I really like is I look at the philosophical language of John Wilkins and you know you can see how did he divide things that were important in the world and it's both it's somewhat sobering but somewhat pleasing in some ways some aspects of the human condition have been the same since the sixteen hundreds that's the same types of issues that come up and some are very different I mean the the whole section on on you know depth and various forms of human suffering is huge at that time and in sort of today's ontology would look a lot smaller big achievement there are also other other aspects of it's interesting to see you know how would a philosophical language of today differ from a philosophical language from the mid 1600s this is a this is a measure of progress I think to see to see how that you know what kind of difference there is that you know so this is one of the things that I'd like to be able to do is to have a sort of symbolic representation of everyday discourse in the way that we now have a symbolic representation of sort of systematic discourse it's that there are many of these sort of attempts at formalization that have happened over the years you know I think in in mathematics for example Whitehead and russell 1910 principia mathematica that was the sort of the great effort to the most the most the biggest show of effort at least had been previous efforts by Frager and NPR know that were a little more modest in their in their presentation to try and see how would you formalize in that case mathematics in a precise system it's sort of interesting what they managed to do right what they did wrong ultimately they were wrong in the idea of what they thought they should formalize they thought they should formalize some kind of process of mathematical proof which turns out not to be the thing that most people care about but you had asked about what will be in modern Turing test or will be a modern analog with your intestines the interesting question I mean I think that the sort of the they being able to have the conversational BOTS which is kind of Turing's idea that's definitely still out there that one hasn't been solved yet it will be solved the only question is what's the application for which it has solved and for a long time I have been you know kind of like like a why do we care type thing because I was thinking the number one application was going to be customer service and well that's a great application it's you know in terms of a a my favorite way to spend my life that isn't particularly high enough on the list what I realized though about the Turing test and things like customer service because customer services is precisely one of these places where you're trying to interface between you know you're trying to have a sort of conversational thing happen one thing I realized is that the one big difference between time and our time is that our method of communicating with computers there's one huge difference which is in his time what he imagined was it's a conversation you say some things to it or you type some things to it type some stuff back in today's world it shows you a screen back and actually the the case that I was curious to see a few years ago was you know you go to a movie theater there's you know you can buy a movie ticket from a person and buy a movie ticket from a machine so there was the question of at what point you know people like me you always like to use the latest you know techno toys you know as soon as those things appear to movie theaters right I was using the Machine only and for a long time there's nobody else using the Machine and then in urban movie theaters you started to see more and more people using machines and and now most people I think use the machines but one thing it's interesting about those machines is how is the transaction with the Machine different from the transaction of the human and the main answer is there's a visible display on the machine so you know you say you know it's it might ask you something that you just press a button you can see immediately you know you can use your eyes to understand something the visual system to interpret something it's a little different and so for example North nothi you know if you ask it something and like you know in when it's used inside Siri Siri if there is a short answer we'll say back the answer the short answer to you what most people want is the visual display of the big report that shows you know that the infographic of this or that so this is a this is something which is sort of interesting because it's it's a non human form of communication that turns out to be richer than traditional human communication that is you know if we were all incredibly fast perfect artists we could you know as we're talking we could draw them infographic and say this is what I'm talking about but in fact in most human human communication where we're left with pure language whereas in human to computer to human communication we have this much higher bandwidth channel of visual communication that turns out to be important and so so the traditional Turing test I think is a little bit it's all funny because many of the most powerful applications kind of fall away because have this additional communication channel so you know I've been interested so for example here's one that we're actually trying to pursue right now is a a bot to communicate about writing programs so you say I wanna write this program I wanted to do this says you know it'll say well I'm written this little piece of program is what it does is this what you want blah blah blah it's a kind of a back and forth but there's also other kinds of bots that we've looked at other things like tutoring BOTS or it's like okay you know you should understand this piece of chemistry or something and how do you you know and that's interesting because it's actually actually kind of a an interesting problem because you have to make a model of the human that is if you're trying to explain now what's the right thing to say at this point you know do you explain this okay what is the human confused about you have to have a model of the human to know what they're confused about and so on I think it's a no but but well it's been difficult for me to understand it is in the case of you know when do you achieve a sort of Turing test AI type of thing it's like as there's there isn't the right motivation there's not the right you know one could as a toy one could make a little chat bot that people can chat with but I don't think and I think that will be the next kind of you know we can see the current round of deep learning particularly recurrent neural networks and so on I can make pretty good models of human speech and so I'm human writing and so on so it's pretty easy to type in you know you say how are you feeling today and it kind of knows that most of the time when somebody asks somebody how are you feeling today this is the type of response you did and you know for example one of the one of the toys actually playing over but I want to figure out what I can automate responding to my email I know the answer is now the sign when you know a good Turing test for me will be why can I have a bottle respond to most my email and it's um this is a and that's sort of an interesting you know there that's a it's a tough test because some you know some aspects of email like a I don't care about this throw in the spam folder type thing that's comparatively easy but if it's somebody says you know what should we do about this inconsistency in some design of our product of this and that and the other to be able to answer that in any way that's um you know to be able to say do you approve this thing to be able to answer that with any reasonable degree of confidence is hard and the thing to realize about that is most of those answers one has to learn them from the human that the email is connected to those big I mean I I I think I might be a little bit of ahead of the game because I've been collecting data on myself for a it's not about twenty-five years so I have I have every piece of email every keystroke I've typed for every piece of email for 25 years of a keystroke but maybe twenty years or something and lots of others other stuff like that so so in a sense I should be able to train an avatar in AI that you know will do what I can do perhaps better than me you know more easily the most you know one of things I think is kind of interesting think about it is in a world where a eyes are figuring out a lot of stuff for us people worry about the scenario oh my gosh the a eyes are going to take over my belief about that scenario is that something much more in a sense amusing will happen at least first which is it will quickly become the case that the AI can figure out it knows what you intend to do what you want to do and it's really good at figuring out how to get there and so just like you know with a car GPS we tell it we want to go to this destination and people like me just I don't know where the heck I am I just follow my GPS years ago when GPS is were much younger I've learned two things my children are always amused by the fact that I had a very early GPS and was like drive drive this way this way this way we actually were on a on one of these piers going out into Boston Harbor that was it's like I just follow the GPS but you know what will happen what to the point is is that Oh there'll be a an AI that knows our history and knows oh yeah you'll probably go to you know on this on this menu you're probably going to want all of this or on this you know you're talking to this person you should talk to them about this you know I've looked at your interests I know something about their interests these are the common interests that you have you know these are some great topics that you can talk to them about um and more and more you know people the AIS will suggest what we should do and I suspect people most of the time just follow what their eyes tell them to do because they're probably better than what they figured out for themselves so I think I mean you know to me this is the AI take of a scenario that happens is the laziness of the humans it's not really laziness it's like take good advice you know the AI is telling you what to do it's better than what you would have figured out for yourself just do what the AI says it's a you know there's a sort of a complicated interaction of in terms of Technology and so on it's this this question of that you know you can do terrible things with technology you can do good things with technology people will always be people and some people will try and do terrible things with technology and some people will try and do good things with technology I think one of the things that I really like about technology today is the kind of equalization that it's produced across lots of kinds of people I mean there was a time when I used to be very proud that I had the best computer of anybody on you but now I have the same computer that's pretty much anybody I know I'm you know we have the same smartphones and pretty much the same technology can be used by a decent fraction of the 7 billion people you know that exist um it's not perfectly flat but it's it's reasonably flat and I think we'll see the same type of thing of other areas of Technology whether it's medical technology other kinds of things well you know in a sense I don't know whether it's luck or whether it has to be that way that these you know these pieces of technology that one's producing are very broadly you know can be very brutally available it's not the case that there's a sort of the king's technology is different from everybody else's technology and I think that's a that's an important thing now in terms of how I mean one of the things that I I always notice because you know we we make stuff that we sell to people and people use all over the world and you know I've sometimes we've even thought about publishing these indices of how much does Mathematica get used how much does Wolfram Alpha get used in different countries about world because you know a huge amount you know which cities and you know you know all kinds of stuff but you know their countries which are really very technologically sophisticated there are countries where they really not you know there and I think you know we can and so it's sort of an interesting thing to me today the great frontier I think you know five hundred years ago was literacy today it's doing programming of some kind today's programming will be obsolete and not very long in other words for example when I was first using computers in the 70s people would say well if you're really a serious programmer you're going to be using a seminal language now I often ask these computer science graduates did you learn assembly language well yes I have knows of section one class about assembly language okay why do people not learn assembly language because basically computers are better at writing assembly language than humans are and it's only a very small set of people who need to know the details of how you know language gets compiled into a standard language and so on well a lot of what's being done by armies of programmers today is similarly mundane it's tough where the goals can be described much more succinctly and it turns into some giant blob of Java code of JavaScript code on and there's actually no good reason for humans to do writing all that stuff you know that's what people like me trying to do is to automate that so that we can automate the process of programming so what's really important is just going from what the human wants to do to getting the machine as automatically as possible to get that done now the thing that's interesting right now one of the things I'm really interested in right at this moment is this sort of the equalization that this is producing because it means that in the past if you wanted to write a serious piece of code a program that did something important and real there was a lot of work you have to you know you have to really know quite a bit about software engineering you had to invest you know months of time in it you had to you know if you were some you know you'd have to hire programmers who knew this you have to learn yourself whatever big investment now you know the big achievement from having automated a lot of the stack is that's not true anymore you know a one-line piece of code even a thing you can tweet sometimes already does something interesting and useful and that means that it sort of unlocks a vast range of people who couldn't previously make computers do what they do things for them to let them make computers do things for them and so now what happens well so one things I'm interested in is is at this point kids and fancy professionals are really at the same level in terms of what they can do in terms of teaching computers what I'm showing computers were telling computers what to what to do for them and so now one of the things I'm interested in is how do you how do you teach that kind of computational thinking and programming to a broader range of people in the world as possible and one of my sort of little private things I really would like to see is for there to be you know a large number of random kids around the world and random countries who learn so that the new kind of capabilities of knowledge-based programming and so on and get to the point where they can produce code effectively that's as sophisticated as anybody you know in the fanciest kind of most educated places can and I think that's a I think this is within reach I think we've got to a point where so if anybody can learn to do sort of knowledge based programming and more importantly can learn to think computationally because the actual mechanics of the programming are pretty easy now what's difficult is imagining things in a computational way and thinking through how do we you know how do we how do we conceptualize this activity that we have in some computational way and so one of the things I'm interested in is how do you how do you teach computational thinking and this is you know we've had I mean mathematics for example there's a thousand years of history about how we teach mathematical thinking and we know to the level of you know which chapter of the book goes here and there it's like I was just asking yesterday actually might we're talking about some for some initiative we have I was was asking about calculus books I think they always have 14 chapters if I'm not mistaken I asked how long if they have those same 14 chapters and the claim was that that the very first calculus book written by Colin occurring in 1727 had some of the same structure many of the examples were the same so it's been something which has been developed over over a long period of time and is very precisely known sort of how you feed mathematics to humans so a couple of points to make first of all in the case of you know if you're writing Wolfram language code and I'm ultimately responsible for the design and structure of how the language works in the case of you know DNA code and biology there's nobody who can point to and say you're responsible for you know you designed this it's something that has evolved over a long period of time and much like a human natural language it is going to have some degree of there's some degree of complexity and on what it's going to do when you have a design language the what you know it's it should do what the designer thought it should do now which is not to say that it isn't super useful to program living systems not least because we are living systems and because living systems are the universe's well the only example we know a successful molecular computing we may you know there may come a time when we've managed to engineer things when we managed to design a lifelike thing that is as designed as a computer languages today but that's we're not at that point so we have to you know we have to be using the molecular computer that we have which is you know our biology now in terms of of how to do that programming I think it's a super interesting question I think that's been kind of you know if you look at the nanotechnology tradition there's been this kind of idea how do we achieve in our technology answer you know we take technology as we understand it in large scale today and we make it very very small so we we say you know how can we make a CPU chip that is you know on an atomic scale well where you'll like it mechanically but fundamentally we're using the same architecture as a CPU ship that we kind of know and love um I think that that isn't the only approach one can take and you know a lot of the things I've done looking at simple programs and what they do suggests that you can have even very very simple very kind of impoverished components and with the right compiler effectively you can make them do very interesting things and I think that is my own guesses are sort of I there's one of these projects sort of doing molecular scale computing that I've wanted to do for so long I just don't quite think that the ambient technologies to the point where one wouldn't have to spend a decade you know building ambient technology to get to the point I'm kind of hoping that there were almost of the day when it's possible for for you know somebody like me who isn't going to build all that ambient technology to actually do something with moto computing but I think my I guess about how how one would do that how one could do that is to say okay we've got these components these components are enough to make a universal computer and you might say well I don't know how to program with these components but by doing sort of searches in space of possible programs and so on one starts to build up a you know building blocks that one can then create a compiler for it and the surprising thing is that surprisingly impoverished stuff is capable of doing sophisticated things and the compilation step isn't Muhsin as one might expect in other words one might think oh gosh if if it's only you know a little I don't know like I have this very tiny Turing machine that's the simplest universal Turing machine that has two states and three colors and has a little tiny rule that you could write it in English it probably a sentence long sentence long but you could make a picture of it it's really tiny and simple um that Turing machine you might ask the question how can I actually compile a program that I might care about down to that Turing machine haven't done it but I think that what one will find is that there's a layer of nasty messy sort of machine code and then above that it gets pretty simple and that that layer of nasty messy machine code will be will add some inefficiency maybe a factor of 10,000 maybe more but a factor of 10,000 is nothing when you're dealing with the scale of molecules as compared to sort of large scale things and so you know I guess my own prejudice and thought would be that so that the searching a computational universe and trying to sort of find programs that are interesting fine building blocks that are interesting is a good approach I think that a more traditional engineering approach that says you know let's try to by pure thought figure out how we build stuff my guess is that's a that's a harder road to hoe it doesn't mean it can't be done but my guess is that one will be able to do some really amazing things by just saying these are the components we have a good representation for them what can we now let's search the possible programs we can make with these things now welcoming out now the question is one might say well we can get this combination of molecules will do all kinds of fun things it will make this big blob of stuff that will do this it will do that you might say but what do we care and then we have to answer the question you know it's back to this question about connecting kind of human purposes to what is available from from the system you know what one question is what what does the world look like when many people know how to code coding as a form of expression just like English writing as a form of expression you know men to me some simple pieces of code are quite poetic you know they express ideas in a very clean way that's very you look at and say ah that's that's there's this kind of a an aesthetic thing much as there is to expression in a natural language um but now now in general what we're seeing is is sort of this way of this way of expressing yourself you can express yourself in natural language you get the best result by drawing a picture you can express yourself in code one feature of code is that it's immediately executable it's not like when you write something somebody has to read it and the brain that's reading it has to separately absorb the thoughts that came from the person was writing it I mean a thing I've realized again one of the things I'm I'm thinking about right now that's I'm very frustrated I can't figure out I'm just sort of on the cusp but I haven't got that yet so what I realized is that you look at sort of how knowledge is transmitted in the history of the world so to speak one form of knowledge transmission is essentially genetic that is you have an organism and you know it's progeny has the same features that it had okay so that sort of level 0 level 1 is the kind of knowledge transmission that happens with things like visual object recognition where the new critter is born it has some girl network the neuron network doesn't have anything you know it has sort of random connections in it as the critter goes around the world and it starts recognizing different kinds of objects and it learns that knowledge so it's kind of something which without sort of without any and that's what throughout the animal kingdom you know critters have been learning as object recognition is that sort of the next level of knowledge then as a level of knowledge that was sort of a big achievement of our species which is natural language the ability to take knowledge and represent it abstractly enough that we can communicate it sort of this in a disembodied way brain to brain so to speak that we don't have to the individual brain doesn't have to relearn from the raw material the the knowledge can be taken out strata and can be to the next brain downline so to speak and you know arguably the the natural language it's kind of you know arguably the most important inventions of our species and in human history and it's what led to in many respects our civilization and many many other things so it's like really important well now we've actually got another level of this which is with and probably one day it will have a more interesting name but with essentially knowledge-based programming and so on we have a way of taking a representation of knowledge in the world it's an actual representation of the world it's not just a mathematics sort of computer language or something it's a thing that represents you know real things in the world but it does so in a precise symbolic way that has this feature that not only is it understandable by brains and communicate with other brains and to computers it's also immediately executable and I pretty sure that this is really a big deal and in a pretty sure that just as in some respects natural language gave us civilization that's the question of what will knowledge-based programming give us and B you know one bad answer is it will give us the civilization of the AIS that would be kind of disappointing for the humans that's kind of what we don't want to have happen because there could be a point at which the AI is a did a great job they're communicating with each other they're they're doing all these kinds of things and we're pretty much left out of it because we don't have there's no intermediate language there's no nothing sort of interfaces with our brains anyway so one of the questions that I'm I'm super interested in right now is this question of in this sort of fourth level of knowledge communication what you know what is the big thing that that would lead to it I kind of think you know if you were you know caveman allgo something and you were just realizing the language was starting it's like could you imagine civilization from that point and it's you know you'd have to and I I feel like what what should we be imagining right now and you know this relates to this question even for humans and most people could code what would the world look like and there are clearly many trivial things that would you know contracts are written in code you know restaurant menus might be written in code and you could say you know this is how the food's getting made okay I want to change this piece and that piece it's on things like this there are things that sort of simple things like that that change I think probably they're much more profound things that change mean they're you know the rise of literacy gave us some things that were chemists bureaucracy for example which didn't really you know it had existed in the past but I think it dramatically accelerated the the bureaucracy for metal worse gave us I think sort of greater depth of you know of governmental systems and so on for better or worse but I think the and so you know there's a question what does that look like in the case where most people could can code now when you when you ask about you know how does the sort of the coding world relate to the cultural world well so one of the things I've been thinking about recently is when you think about for example high school education and there's a question of okay you know how do you teach programming coding and that kind of thing computational thinking a high school level and one of the possibilities is well you have a course about that and you tack it onto all the many many many things that people are being taught today the other possibilities it's much more interesting is you just rethink all the existing areas and you say well if we also have computational thinking how does that affect how we study history how does that affect how we study languages social studies whatever else and the answer is it has great effect I mean in there's a lot of things that you can for example imagine you know you're writing your essay today the raw material for a typical kids essay is well I read something and this is the raw material and I'm going to write what I think about that it is not the case that kids can generate new knowledge very easily but so then the computational world that's no longer true it's very straightforward for a kid to go and you know if they know something about writing code to go over to the you know beautifully digitized historical data and so on and he'll figure out something new and then you're writing an essay about something where you say this is what I discovered today so to speak I can write an essay about it I think that's that's part of the the way that you know I don't think it's sterile at all the waiting this is the achievement of knowledge based programming it's no longer sterile and the reason is because it's got the knowledge of the world sort of knitted into the language that you're using to write code so if you take mathematics as an example right so there's this right now there's this area of the people teach which is sort of a pure mathematical area but at least basic math gets into all kinds of places it hasn't gotten so much into the humanities but in you know it's it's something where where it's just part of the the way we think about things is so at least basic math so similarly computation is something that in these times is part of the basic way we should think about things and the great thing about computation is that if we think about things in terms of computation and then the things become sort of immediately executable they become things which where once we have the idea computationally and we know a little bit of a mechanics of how we write you know code it's pretty straightforward mechanics then given that idea once we've had it once we've formulated it computationally we can then get the machines to go do the work and a kid can get a machine machines do the work just the same way as the fancy researcher can do it I mean you know that I was saying earlier was I think this is the biggest suti an AI on its own does not have a goal goals are a human construct so the thing that came out of lots of science stuff I've done is this realization that sort of intelligence and computation are kind of the same thing and there's computation all over the universe whether it's in a turbulent fluid producing some complicated pattern of flow whether it's in some you know celestial mechanics thing of you know some interaction of asteroids with this that and the other whether it's in brains and so on and so this question of does it have a purpose right what is its goal you can ask that about any of these systems does the weather have a goal does climate have a goal does now and this is a very quickly I mean this unfortunately there's one of these things which people have been asking this since Aristotle right and this is the you know this is kind of the final cause question and so on for Aristotle and one can't unpack it a little bit so this whole sort of doesn't thing have a purpose so let me try and unpack it a little bit I haven't I don't claim to have completely unpacked this question but one question is can you tell if the thing has a purpose it was this made for a purpose so look at Stonehenge for example well Stonehenge made for a purpose was you know presented with a thing was it made for a purpose now a lot of stuff that we see today it's very obvious that it was made for a purpose by humans because it has a lot of the vernacular of human engineering history so you know we see a thing with cogs and no the Antikythera device when you know when people started looking at this lump of gunk that was you know dredged up from the you know of 100 BC 200 AD period you know shipwreck you know as it was it made for a purpose while you you know you when it was dropped and broken to their little cogs sticking out and we kind of immediately know this is made for a purpose and it isn't just a pile of gunk because that's part of the history of human engineering um and so given the history it's very easy to recognize human purpose in things I mean this is a little bit similar to this question of you know is it alive or not on earth it's very easy to answer the question is it a live one that you look at fairly easy you know we look at those that have RNA does it have cell membranes of these kinds of things that come from the history of life on Earth I remember when I was a kid you know the first first Mars Landers were landing right and I remember this is their life on Mars and is the green stuff that seems to happen every season you know vegetation or whatever and I remember I was really curious you know what would the tests be and you know from today's time he's a they're pretty amusing I mean the basic test that was was used was scoop up a piece of Martian soil feed it sugar and see if it eats it yeah that was the that was the top test of is it alive now I don't think any of us would believe that that life has to be something that eats sugar but you know the question of what is the abstract death life that's really hard you know there's a and it's very tends to be very sort of anyway the back to this question of purpose how do you recognize purpose so there are bunch of other huge examples you can get so one example is look at the Earth from space can one tell that there's anything with a purpose hanging out of the earth come and tell that there's civilization on the earth and it's a it's sorry I did this experiment 15 years ago now I asked astronauts what do you see on the earth that shows you that there's intelligence on the planet so to speak and the first thing I was told was in the Great Salt Lake in Utah there is a straight line that has turns out to be a causeway that divides what turned out to be two areas that have different very different colors of algae so it's you know it's very dramatic straight line it's like okay there's a straight line and I found that I was interested in where's the longest straight line made from lights as one knows that there's a road in Australia that's really long and straight and there's a railroad in Russia I guess in Siberia basically that's that's really long and it has sort of a you know lights bit go on you know when it stops at stations and things that are lights there so you'd see some straight lines and things and then and another another good example is some in New Zealand there's a Morris perfect circle baseball mountain rapist okay so this was I was doing this research I guess it made me go now before the web was common I'd say you couldn't just go to look up awesome statue into this so anyway we we got in touch with them we're trying to get maps of the big an autonomy we're in touch with the New Zealand Geological Survey and they said if you're writing a textbook please do not say that Mount Erebus is a circular volcano the circle does not come from the volcano the circle comes from a national park that was drawn around the volcano and there are sheep or something that graze you know inside the National Park but not out the way around and that's what leads to the circle so this is another example of human you know there's a piece of geometry that comes from the humans but it's pretty difficult to find really clear examples of sort of obvious purpose on the earth as viewed from space and I think you know the other question that comes up it's a question for the extraterrestrials so to speak okay you know if we want to recognize extraterrestrials out there how do we tell if a signal we're getting has a purpose so you know 1968 pulsars discovered you know every few milliseconds I guess knowing how long was the first of all saref again but something between no seconds and seconds you know yeah you hear kind of this this flutter like sound that's a periodic thing and you know at the time it was you know the first question is is this a beacon because you know what would make a periodic thing like that it must be for a purpose well actually it turns out it's just a neutron star rotating but this question comes up over and over again what gives evidence of a purpose and in fact back in the early nineteen hundred's I guess Marconi and Tesla were both people who sort of listening to radio transmissions from away from the earth and sort of a question you know I think I've coming out a lot I think in the middle of the Atlantic you know could hear these you know we it sounds that sound a little bit like whale songs for them they're kind of you know they come from radio type things and the question I think Tesla was very much on the this is the Martian signaling us type thing how these went well in fact it's some modes of their own sphere that are effectively a magnetohydrodynamic phenomenon they're just physics and this is one of these cases of like the weather has a mind of its own so to speak how do you tell whether it's a thing which has a you know intelligence and purpose and all that kind of thing or whether it's just magneto hydrodynamics of the ionosphere a nap jump that's a so this is how do you tell if the thing has brothers it's hard on one protein that I think one can potentially apply is does the thing that Chi if you can identify a purpose that is it minimal and achieving that purpose that is if you see a thing and it's mostly you know a a for coming into it but it has incredibly elaborate ornamentation on it you'd say well its purpose is a fork but it also has all this ornament which is not relevant to its purpose now it's ornament the ornament may itself have a purpose to have people give people a different emotional reaction to their fork or whatever else it is but what what is some this question of is something minimal for its purpose and does that mean that it was built for the purpose so so when you look at the thing there are typically different explanations you can give for what happens one is the mechanistic explanation it does this because the ball rolls down the hill because at the next moment of time the gravitational pull will do this and this and this or the rope ball rolls down the hill because it's satisfying you know the principle of least action and it is globally trying to optimize this particular thing and you can you know they're typically these two explanations you can give for something via the mechanistic explanation and the you know to a logical explanation and the question of which is the winning explanation which is the right explanation there even know is so one possible criterion is the thing was built for a purpose if it is minimal and achieving that purpose the problem is that essentially all of our existing technology fails that that test we can imagine technology that works that way but most of what we build is absolutely steeped in technological history and is incredibly non minimal for achieving that purpose you look at a CPU chip there's no way that's the minimal way to achieve what the CPU chip achieves yet you know it's it's steeped in all this history of our engineering technology so so anyway this is this question of how do you identify if a thing has a purpose I think is things really hard I think it's really but for example for the extra-terrestrial question it's really important because it's like you know one good thought experiment is imagine that the extraterrestrials could arrange stars however they want how would they arrange them to show that they were arranged for a purpose but they put them in a straight line probably not because they can imagine all kinds of physical processes which might do that but they put them in it it shouldn't put them in equilateral triangles because that there's a particularly simple physical process that does that you know would they have a by Koch design that they have a you know some piece of you know alien artwork you would undoubtedly not recognize the alien artwork it's having a intelligent purpose and so on so it's a it's really a it's a you know I think it's an important question because when we look at you know radio noise from the galaxy it's like it's very similar to you know the CDMA transmissions from you know cell phones it's you know it's not fundamentally different from that and it's it's you know they those transmissions use pseudo noise sequences which happen to they have certain repeatability properties but they come across as noise and they are actually set up doesn't always for the purpose of not interfering with other channels and so on so it's it's a really a funny thing what you know how do we recognize sort of a a fundamental purpose and and the whole thing gets even more messy when we say we ask a question like so you know if we observe Prime's being generated from our sequence of primes being generated from pulsar we'd say what generated these that you need a whole civilization that would grow up and discover primes and make computers and did this and make radio transmitters or is there a another explanation that's just that some physical process makes primes and then that physical process may have all kinds of weird things going on inside I mean there's a little so an automaton I made up once that that makes Prime's and you know you can see how it works you take it apart of just has a little thing bouncing around inside it and I've come up sequence of primes and you know but that didn't need the whole history of civilization and biology and so on to get to that point so it's really a slippery thing whether you know when you observe something was it created for a purpose how do you tell if it has a purpose these kinds of things see I don't think there is an abstract sense of purpose I don't think there's an abstract meaning - in other words I think what you end up with is the universe if you're ending up with us weird thing where you have to say does the universe have a purpose then you're doing theology in some way you know does it there isn't an abstract purpose I think I think there's no meaningful sense in which there is an abstract notion of purpose that is that purpose is something that comes from history and it comes from so you know one of the things that might be true about computation might be true about our world that would be kind of disappointing is maybe we go through all this history of biology here civilization and so on and at the end of the day the answer is 42 or something and that's just you know that's the end so it's because that's that we got to the answer I would say you went through all of that you know what it crazy you know what a crazy place you went through all these four billion years of you know there is kinds of evolution and so on and all the stuff and then you got to 42 well actually nothing like that will happen because there's this notion of computation with irreducibility which is kind of the thing that comes from girls there are universal computation and so on but there are computational processes that you can go through and that things often go through where there's no way to shortcut that process in other words you can't just no you can't say oh you were wasting your time I mean much of science has been about shortcutting computation done by nature so for example if we want to you know we're doing celestial mechanics we say let's predict where the planets will be a million years from now well you know we could just follow the equations follow each step and see what happens step by step but the big achievement of you know when we think there's a prediction in science it's because we're able to shortcut that and just jump from you know from where we are now and reduce the computation we're able to be smarter than the universe and figure out you know this is the end point without going through all the steps and that's been the sort of story of prediction in science but the good news is in a sense it's bad news for science it's good news for us having meaningful lives so to speak that there isn't a way to just say okay we can shortcut ever you know with a smart enough machine and smart enough mathematics we can always just jump ahead and get to the end point without going through all the steps we actually have to irreducibly follow through those steps and that's in a sense that's why that's why history means something if it was the case that we could get to the end point without going through the steps we would you know in a sense history would be in some sense pointless wouldn't you know I think I think this this fact that you know bad for science because we can't make these predictions but good for the meaningfulness of the history of civilization and so on it's that you know these details sort of are reducible and I think in a sense when one realizes that sort of ever think and have these attributes like intelligence and so on one realizes that if we are going to distinguish that the thing that has to be special about us is all of these details about us it's not going to be some big feature like it's not going to be the case as I thought it once that it was that there's us that's intelligent and there's everything else in the world that's not it's not going to be some big abstract difference between us and the clouds and the you know the cellular automata and the whatever else it's not an abstract different it's not something where we can say look you know this brain like neural network is just you know qualitatively different and this cellular automata thing rather it's a detailed difference that you know this brain like thing was produced by this long history of civilization etc etc whereas this Sonor ton of them was just created by my computer in the last microsecond you know I think I think it said so but my belief about you know the problem of the abstract AI is very similar the problem of extraterrestrial intelligence you know it's the recognition of when is a thing when does the thing have a purpose when is a thing intelligent you know these are these are I think these are again just questions I I don't get so that answer I'll be very you know it's a you know of course one of the great things in science is why have we not found that a extraterrestrials why you know how could we possibly be that's unique and you know what extent you know that maybe maybe that's a silly question because maybe there's intelligence all over the universe and we have to then ask well just how close is it you know does it have RNA does it have you know did it invent a notion of democracy or something and you know a lot of these other attributes like that we think of a lot of what we when we start trying to break down and say well it will be intelligent you know I I will be intelligent if it can do blah blah blah if it can find Prime's if it can produce this and that and other you know there are many other ways to get to those results and that's a consequence of the fact that there just isn't a bright line between intelligence and mere computation I think that's the I mean in a sense it's a very it's a disappointing you know it's a it's another part of the Copernican story so to speak you know we used to think you know as the center of the universe all this kind of thing and now at least we think gosh we're special because we have intelligence that nothing else does and I'm afraid you know the bad news in the sense is that that that really isn't a distinction and by the way that lack of a distinction I think is pretty critical for thinking about the future of the human condition because his is one of my I sort of I don't know scenario that I'm I'm really curious about it you know let's say there's a time when human consciousness is readily upload of all into digital form everything can be virtualized and so on and pretty soon we have you know a box of a trillion Souls there are trillions souls there in a box they're you know all virtualized and we look at this box and in the box there'll be you know hopefully nice molecular computing immediately derived from biology in some sense but maybe not but there'll be all kinds of molecules doing things that Ron's doing things the boxes doing all kinds of elaborate stuff and then we look at the rock that's sitting next to the box and inside the rock there's all kinds of elaborate stuff going on all kinds of electrons doing all kinds of things and we say what's the difference between the rock and the box of a trillion Souls and that's where that's the and the answer will be well the box of the trillion souls has this long history and the details of what's happening there were derived from the history civilization and you know people watching videos made in 2015 or whatever and you know all these kinds of things whereas the rock well it came from this geological history but it's not the history that is like you know it's not the particular history of our civilization but I think this is the kind of this question of you know realizing that there isn't sort of this distinction between intelligence and mere computation leaves you leads you to this these these things like imagine the future of civilization ends up being the box of a trillion souls and then what is the purpose of that what do we you know for example from our current point of view in that scenario it's like every soul is playing video games basically forever and the question is what do they you know what what are they you know what's the kind of the end point

Channel: The Artificial Intelligence Channel

Views: 30,310

Rating: 4.8671098 out of 5

Keywords: singularity, transhumanism, ai, artificial intelligence, deep learning, machine learning, immortality, anti aging

Id: cbu_bCQ2Lkg

Length: 91min 9sec (5469 seconds)

Published: Tue Oct 24 2017