« StarTalk Radio

Will AI Replace Us? with Matt Ginsberg

2023-07-21 | 🔗

Is artificial intelligence taking over? Neil deGrasse Tyson and co hosts Chuck Nice and Gary O’Reilly discuss deepfakes, AI hallucinations, and whether AI really is intelligent with software engineer at X, the moonshot factory, Matt Ginsberg. 

NOTE: StarTalk+ Patrons can listen to this entire episode commercial-free.

Thanks to our Patrons Kathleen Kussman, Craig Hamilton, Denis de Oliveira, Jim, Ryan, and Krishna for supporting us this week.

Photo Credit: mikemacmarketing, CC BY 2.0, via Wikimedia Commons

This is an unofficial transcript meant for reference. Accuracy is not guaranteed.
For it is electrifying its icons to keep the soul of driving alive while shifting towards an electric vehicle future the- all electric mustang mark ii gives you the bold style of mustang. Along with the practicality of anne S, you ve an electric vehicle for people who want to them to experience the thrill of the drive sure their needs are met, get the x I met a mustang and the space to share it with five seats and frank, a k, a front trunk, that's right, it has room for you, your people and your things more about the all electric mustang marquis suv at four dotcom, slash, mustang, mark ii, a c h e again, that's for dot com, slash, mustang, mark ii,. Want to learn a new language and who doesn't well spearing its immersive lessons from the most trusted language app rosetta,
stone. You know you keep telling yourself. You want to learn a new language, but true axiom feature even guess: feedback on your pronunciation, so you can speak the language like a native fine lessons as short as ten minutes, making it easy for you to learn any time anywhere. Don't put off learning that language. There's no better time the right now to get started and roses. The stone makes the perfect holiday gift for very limited time. Startup radio listeners can get rosetta stones lifetime membership for fifty percent of that. Fifty percent of us but if access to twenty five language courses for the rest of your life, redeem. Your fifty percent off at rosetta stone, dot com slashed star talk today, coming up on start or special edition will ay I The end of civilization, what but before
happens will help us coach sports better. Will it make deep fakes that will destabilize All that we know and trust also would help us explore space. to welcome start off your place in the universe. Where science and pop culture collide knocked off begin right now. This is the start of extra special edition, an entire show right now devoted to a ah, we will now is a good or bad add on earth. You earn ah bout, it everybody's opined. In on it. I do visited price I saw her or does it all create problems. I, how do we use it in
a certain media in an all kinds of places. Here we are back with Jake I jack, how you doing man, my co host, sorry to tell you, but this is not me, I'm not here. Ah he he, he see. It since this is my I double gander, that's your double gearing right back. I have ways the test for that problem. The area has me tied up in the closet. Help me place some that are made dairy cow you do a man, I'm good, I'm interested in because he's not a subject. I have any real depth of knowledge about, which is most subjects But tat is my lesson has not carry that lie ahead, tied up in a closet. Ghana is right. I asked to be all right. I shall do my best.
You're too double gag serious doubts gary set the scene. What do we have here? Alright, today's guest has a doctor in mathematics from oxford. That said, the englishman, was it researching astrophysics and then decided to switch to I, author, yes published author of a book in twenty eighteen called factor. Man develop ten. I love this thing. Developed a computer program called doktor feel less EF. I double l. That's The crossword puzzle puzzles and doktor phil has competed in professional crossword puzzle torments successfully, and now these gentlemen works for the delightful people, Google, so today's guest, none other than a dear friend, meant this book was returning to start menu, would welcome back, not greater gratefully tie you all again. I think. At the end, we probably should have a vote on whether we want a lack shocking gary out of the closet or whether we prefer the dockworkers
and we think that I am right there, whether whether these are better versions right so met you. I wouldn't what are you doing with Google right now? or do you want some kind of nba non disclosure agreement? So I work for an organization called acts. It is part of alphabet We are also moonshine factory, which means just just be clear. Alphabet is the whole company or google outdated holy coming of google and other organizations so way for example, is part of google or x in alphabet somebody and alphabet x is an alphabet. We're not I have good enough of it I bet you something for a guy who went to oxford. I ain't so impressed that you know that at is in the alphabet. If there s we do the hardest stuff, we can think of it. Unbelievably fun place to work. The people are incredibly smart. We
A project like how to rise in it in that sphere is the x prize was money for its citizens, I should say moon shot, something there who thought you could make this happen, and you do we put enough smart people, I'd funded, enough money and then there it is. Is it true that they call it the failure division of of google because they don't care? If you fail it's all about, you know the disk. ray of information and and advancement through doing stuff, You would never otherwise attempt. So we this project called tapestry. The goal is to de carbonized retrograde everybody uses renewables and huge, climbing a pact that is so hard that if we can't do it, nobody's gonna, be stun
specific project, that acts is probably more likely to fail and succeed, but there are some amazing successes so way mo which alphabet self, driving card evasion came out of it if another project called mineral that just graduated. That has these weird view. those that drive around the tracks between crops on a farm, farm in these cameras and machine vision, to figure out how in so doing and just generally to make farming more efficient. All these things are things it there were just as hard establish free when they started but they had actually succeeded in other out, as I mean we their bats, their party. Alphabet, so we call them. That's an deserts. Other division weber like that's it, but not an eighth its. I have to tell you it's great working for a company that expect you to do unbelievably hard things and realizes that Try and do things that hard you're, not.
It's gonna, pull it off what course, science in general has many many failures. The press only talks about the successes, so you starting out in science, This would not have been a foreign concept to other than that there is a whole company, that's cool with. It fits normally The joint debate on the bottom line you're on the street in the next the next quarter, exactly No, I tell my friends, in projects that eventually I doubt it didn't work. I say you know you. Always learn more from a failure and from a success, so we should celebrate the failures and we do we actually when it should come to an end. Get shot down. There's a big meaning! Everybody applause is like a party because we know that we ve learned stuff. We know that we ve tried star. and we know that now.
Try something new and enough of it works that the whole enterprise is something that can justify that. To you know the short sighted desires of shareholders, because honestly, don't know, I don't know how you do it. You out. We have two different questions from at one percent of google's. annual revenue, gets directed towards exe. That is covered by an area is an already know. That is india. It is It is an irony effort that tries to do the crazy things all this that we're gonna talk about about general. Today I is based on technology called transformers. They came out of a part of Google com, rain brain came out of it, so
way while came out of it. We we ve done amazingly impact for thing. We have a long time frame, so we start a project. It's not it's common for us to say this about take tenure we might kill it into because we can tell but its not working, but if it succeeds, it's gonna take ten years. We're! Ok with that, so The hard part from from the shareholders is not arguing that routing value, I think, I think, were clearly adding value. We have to get them to be patient enough see that value materialise, so you're lucky man, because there's a cult, there's a culture there. I saw it once or twice when I was praying where If you d lose defeated heavily people went away lick. Their will. but considered where the things went wrong. Came back with solutions. If you build culture, You can achieve things by using that way, but the pressure to get results in a shop was so right in pointing out its dollars. I don't have time
I'm permitting such a phenomenon amount of money to this, and if we do, too many games. My coach gets fired, also to think so. it's a really really fascinating place if you can developer sustained cultural viper we do have the culture. An alphabet has been fantastic about, Ok, that There is room in an entity as successful as alphabet to have a bunch of people. To let them take the long view to let them try and do incredibly hard things and see what happens and and and obviously we can't just keep failing. Some stuff eventually has to work, but some stuff does work and that the people at acts who decide what are we gonna work on when we gonna kill it? When are we going to keep pushing it? seem to be very good about. Ensuring that net net warehouses a positive reflect on a recent new stuff,
about a john lennon song, wherein aid sample John Lennon's voice. And then had him finish. The song is, he died before it was recorded. Yeah can you were fighting on is a good thing or a bad thing or that's a complicated thing. I think that on say more serve you. What would Jesus do we say? What would John Lennon do without it If he were here, would you you in the nose, like nobody do more beetle songs, I think, are an undeniably good thing. I think that expected did he had ability? Your two things you wanna take away from us. First, What they chose to synthesize is emphasised, Lenin's voice.
Yup what made the beatles so magical is, I think the words what they chose to say: that's much harder. Synthesizing a person's voice is relatively speaking, easy. So the first thing I think we need to think about is synthesizing a voice and since synthesizing, the idea of the essence different thing, and the second thing is the fact that you consent. someone's voice is scary, because it's gonna make deep figs so much more of a problem than they currently are, because now we can, the picture of whoever doing whatever, and we can even attacks some ways to it. To then she, but it really happened when in fact it didn't really happen. So I think that's a big few. That huge that will need to be addressed if we're going to keep all of society for tend to be grounded in reality as opposed to these fabrications, but there's about it right now, because, if the, if we
come up with some way to watermark this technology? On a digital level, I mean what is going to what is going to stop people from utilizing it in the most heinous ways possible? Nothing. Nothing! So that's! Your actually right! Water marking is tricky because Emma. You can pass a law saying all dangerously created. Images must be watermark, and then somebody creates an asian country, it doesn't have a law. and it's all you and now. What do you do? I think that the way we have to deal with this, and if there were a couple of things what he is, we need to develop the technology so right now there are programmes that can recognize barred which is do those charity they I can talk to, and if you give it a try
it was written by bar it if they can say yeah. Ninety five percent that was written by barred not by human. We need to use those to understand. what was trading and always now and the same thing can be said of images. There are none watermark traces that we need to better understand and better make take advantage The second thing is, we need to recognise that there are trusted sources, and we need pay attention to this came from somewhere that. I actually unwilling to believe if they say they took the picture he took the picture, it actually happen so a news agency potentially conserve this role in the way that some guy some far away, who just creates a picture, might not be entrusted or is, and we as a society have to be more suspicious. Sadly, and we have been that just because
You show me a picture. Doesn't even strip as true met. We discussed some of the positive at it. Some of the negativity, but if we take intelligence part of I have. We actually sweeping started to play with their development? pointing in the right direction to do the right. Things or have we kind of just wasted our time with a sofa. I don't think we're wasting our time. Ok, first of all, I think there are. There are lots of applications and ai that have been phenomenally successful, incredibly important. The car that you just bought was probably manufactured mostly by roma. The cars have fewer defect. Coming off the line, absolutely air cheaper to a good. Are they well, there's innovation. I think pathfinder right, we send a robot to mars because we can
Don't have the technology to some people might yet, but we Send our agents there form of these automated devices. They have to be pretty independent because round trip message time tomorrow, this law You know the rower has to avoid a rock all by itself, because you he don't talk to it. You can't talk to it or clear, quickly. Watch our live twenty minutes later, it's too late. The marcy clearly more extreme at an early hour, since I think we ve done good things. I think that a high is waiting very fast at the moment in wine indulging moves quickly. Its challenging. It's always been challenging for society to keep up and for society to say, ok I have to think about. Here is how the world has actually changed and what it it. important that society distinguished from the apparent changes which are actually grounded in in reality, from and actually has happened that matters so, for example,
all these shared forums barred and others do not actually smart. Actually know what's going on, but it's very easy for people to think they're, smart and too some for explanations about which there completely poorly equipped to respond, as opposed to asking the things that they can respond to. So, for example, Somebody asked me recently if barred understood causality they wanted to understand. some causal thing or lack of a couple of things, those it could barter understand causality, and I said: ok and I went to bed and I said, is there correlation between? phase of the moon and the amount of chicken eaten in denmark, and it said absolutely it quoted a a paper that had never actually been written in a survey that had never actually done it just made all this stuff up. That was- and I told my son about this. He said dad. Why are you asking barn questions like this? It's not going to answer them, but if
want a new business, and I wanna have a website, and I can't afford it develop I can go to bard and say: make me a website that does this and Bartel do great, so we all of us need to understand what these entities can do, what they can't do, what they're going to be effective at what they're gonna be ineffective, at, and we need to ask them to do what they are good at, which is a lot it's just. Not everything you're not going to you know, replace all us All the good news in awe of all in the operative well, shall we realized you apparently because you're in the closet near this is just the chuck avatar. Well, love blade mats, yes,.
hey- you ve been searching for a hotel for that epic trip that you ve been planning, but there are so many options that you can't decide which one to book the two So, hotels, family, a brands, has over seventy five hundred locations with a whole. Tell for every type of stay like Cambria hotel. with offerings like rooftop bars, spotlight, bathrooms and convenient downtown locations or a comfort in for the whole family to enjoy Cambria hotels, offer spy light bathrooms with many that have bluetooth, mirror connectivity. So you can listen to your favorite podcast and we all know that start talk or songs. While you get ready, H, Cambria hotel us curated with locally inspired design elements, providing guess with a sense of space and place, In the city you are visiting, Cambria hotels offer hyper local craft beers on tap, ok, yeah. Had me at craft beer? How about
move, top bars lounges an outdoor spaces, forgets to enjoy comfort offerings include a free, hot breakfast, great pulls for the entire family and spaces rooms make your travels come true today, when you book direct at choice, hotels, dot, com, that's choice, hotels, dot, com. Where travels come true,. It seems that we are talking about robots mars, what would you have to build into a programme for it to look into deep space? And fine things that we don't look, don't know to look for just yet How do you go about tweaking? I
and to be able to achieve that, or is it not quite there? Yet I think it's mostly not quite carrier. So the way these things work is so his is a here's, how I often think of it the world, It breaks down into what I call fifty one. Forty nine problems where you want to be fifty fifty one percent right is good. So if you're playing the stock market- and you can, accurately picks oxford guy fifty one percent of the time you're about to be really rich, and then you have hundred zero problem. Where, if you ever sent, is not good enough. Ninety nine percent is not good enough. If you're trying to shut down a nuclear reactor in an emergency, you really need a hundred percent answer. all of these machine learning systems are incurred. Only good at the fifty one, forty nine sir, and there not so good at the hundred zero, so if you, a look in deep into deep space,
and you really interested in something that you thought was probability zero, an alien talking, tough, or a new kind of supernova or something that we have never seen before. What sort of a hundred zero ha. recognizing things whose probabilities actually zero and its EU huge machine learning systems are not created them but I don't agree, nerd fight in the progress you. I don't agree gloves on nerve right, we're going to tax evasion or afraid so that I agree with that in principle. Agree with you, but it is an important new ones here, because I can programme the computer to show me something way that I don't recognize is because I have a huge catalogue of things. I do recognise no matter what it is so
I know what stream abroad, but it's something that we haven't. We know we haven't seen it cause. The computer knows our entire catalogue correct the rain or knows everything that we now and something up there. We don't know- and I say, show me everything- we don't know, Because that has a novelist because anything beyond normalized wanted a gazelle gazillion thing right, yes or it could just be a glitch in the matrix, but it'll find it though, and so that's why I dont entirely agree. It's not good it at least astro physically in finding the d, the lone wolf out there. So the trick is we're Asking is find one that doesn't blot, basically correct, and These programmes will classify everything you give them things and they'll say home buyers this in pots- seven, oh, my god, so
whether you want it to or not as what you're saying is. It makes the decision to do so on its own, even if it creates a category that didn't exist and says well now it's in this category, well, it'll, probably put it in an existing airport. Ok, I want you to run now. We re pegana hang on hang on where you can you don't make me come out there, Somebody give us a clap corn. This note fights didn't really do. Is you can say if the frequency is outside, of a frequent I've ever seen flag it if the period the city is shorter than anything. I have ever seen flag it. You're doing now? Is you are actually creating sort of a new category of surprising thing That you are defining for the system. Is a couple of them accordingly, salute you? Can do that, but if you see a true surprise then do you actually had. No idea was gonna exists. So, for example,.
Imagine that you find a pulse are and miraculously The phase of the poles are One hundred percent correlated with phase of another pulsar to light years away? That's amazing! Something totally bizarre is going on. But the machine won't know because it has no idea to look because it doesnt how he told how round its not a new object. It's it's a new phenomenon in unison thing I write, it said I have arranged meetings that are better ran new things very logical get that it'll save pulsar next right wholly ogre I'll. Give you that that's all I'm saying is it does not. Look for it because we don't know to look for correct, when I was, I was being blunt about it and sang objects. Phenomena that are sort of singular that you just put in a catalogue with properties we
You, yes, I agree rural net searching through our data to find weird stuff all the time, but you are right. If there are two pulsars there's synchronize, we know it. Pulsars are, we know it there, but what their posts, like and there's synchronized good aliens again ready for the invasion? We would have no idea, nobody, I find that exactly, but but what would happen if this happen, people would say holy cow, pulse over synchronize, and then they would define a new thing which is a synchronize our pair all of a sudden that would go into our category. And now we would talk about that pulsar pair as an object and all of a sudden, the well, that would say: oh pulse, our pair I've seen that for this, the pulse, our hair, but the first time you see one of these fundamentally, new phenomena, which is what makes science fun in all honesty, machines thrown up it's too far outside you know, but they ve been trained to do
so we got to retrain them, already knows, I think so it puts a limit on a eyes' ability to explore for us. So it does so right now- and I have said this before the shop, what we are good at. and what machines are good at a different Incredibly good holy cow to synchronize pulsar is who'd, have thought it. I gotta pay attention to that were amazing. It the machines, their amazing at this day. But you can barely see over here. I looked at eighteen. Different ways is probably also They are better than we are Now we can do more with machines at our side than either of us can do in isolation, and I think it's great I'm incredibly optimistic because of that I dont see and the day is coming somewhere
way far away, but right now I don't see that then you, but without ass, any better living nearby without that were good to get this flip it into my backyard sports. Could I become live in game play. Culture, head coach could react in real time effort and I and I have I have built in anything like any foul play collar. I have running in simulation against the choices made by actual play, collars any crushes them it just annihilates and its easily fast, and I actually I played with it before joining at play than I had it. I actually was watching football games with it and it would put in real time this is what you should do, I would watch the coach villainy dead and- and I read
always simulations people are. Unfortunately, I guess not terribly good place always play calling as this giants, fix problems. Machines are gonna, be great at that great out it fast. However, does does europe programme take into account the adjustments that are made by quarterbacks who recognise coverage in real time? so the difference plays a car and the only answer is yes, it does, Yet in many more than one question, let me talk. There's one question: does your programme deflate the ball no further that said the answer that the answer, the chuck question is that part of my software was who's the quarterback, ok enough, because the quarterbacks that are effective at adapting to what They see when they come to the line. Exactly aren't going to have a slightly different statistics.
then the quarter by air or not effective amigo. When the programme decide I don't want to colorado, I want to call pass. Where do I want to call pass to it? Does no report had where I am and what I missed something here since you how do you what he made you do better than the actual call players because you don't have an outcome that you can look at you you're saying The done that should have done what I said had he done. Would you said how do you know what the outcome would have been, so there don't. So there are two ways: first, I as any any coach is based on a simulation engine, so I could. I can run that simulation now, it's sort of herself I've been going because the single yes, it is a fact that way, one and I think you're still merit there, because you can test the accuracy of the simulation. The second thing I can do
I can just go back and look at the gate and say: okay, the play: where the actual human code made my play abc D, the place where he made a different play empty hiv? I can look at him and say a b c where How pretty well actually age J there were doubts and easy to look after the fact and see whether a play was a success or failure. But aren't you still just invoking statistics of past events I am in running their future event. it's reasonable right. So there's a reason why I'm going at it but yeah, that's the case. Isn't that what they're doing every single pitch in baseball today, They already thing, isn't it over analyzing a bright in baseball? Why are they not doing any football so maybe they are. I said that I dont know I You think they're baseball's a little more committed to the statistics of this more than the
as is happening between, but we also think that at some other we are asking is how hard is it to build? a machine, learning or other model. Tells you how effective a pitch will be, how effective football play will be. That is able to predict with reasonable accuracy as is able to predict the outcome of particular sporting choice,. The answer? Is it's not super hard and it's not super easy when you have to get there There is a lot of duration and bob. You haven't you reasonably modern techniques. That's what I did. in developing and it worked pretty well. I was able to tell with asia answer it a while. I think I was I was able to predict, run versus pass,
With very high accuracy- and I was able to predict exactly what play would be called something like twenty percent of the time- that's very, very high for football, twenty percent. That's insane, could you imagine a football team on this being on the sidelines, from a belgian team that stealing the signals, if you could imagine being on the sidelines and with high accuracy, knowing at twenty percent one out of Five plays. You would know that what they're going to do you want You're gonna do a lot harder game. This is part of why the software why the simulations indicated that a machine code is Is really going do very well against human cur,
I know, but what, if you put, what did you put one against another? They know my eyes better than your railway line. I always feel encouraged to those seeking out the deceptive cards versus the auto bud, Do they cancel each other out? They they sort of cancel each other out. So you not if you let's go to chat, which is sort of less I'm the marshal, so you can have even a relatively poor chest programming, is way better and the best human rights, but there, no better programmes and worse per right, so the same, thing can happen here now you have additional facet factors and sports, because The teams may well be simply more physically talented than another t true, and then the question becomes and a difference in the quality of the coach whether or not overcome the difference in the fiscal policies team and he won't allow that. I don't buy it right there.
we'll be better eyes and were say eyes how big those differences are. I don't know because right now, there's no way I've, but That's gonna be another facet of what makes a team good, How would you like the old man? Is there the spot, so organizations be whichever sport are already using I technology for, in situations, I think it's pretty small, when I joined acts, I was I was trying to talk, the nfl about using this offered it all for by calling- and it was it seem pre clear to me that they weren't doing anything like that. Yet I've been an extra get lears, sir we're looking back two years that eventually this is going to happen, but of course not yet, but it's moving towards that, because right now you're using nets, jan and next gems.
that's a wonder, use what they're looking at is percentage like likelihood and probabilities for certain plays at certain times you ve never seen, more to point convergence in the nfl than you have right now that is a direct result of statistically you should do it. You ve never seen more teams going for forth and whatever. Because his dick low, you should join so help, I'm so embarrassed erection show about that specifically about I was eight years ago. I did some statistical work for the organ donor. and I told the coach at the time. I gave him these huge printing with what you should do in every situation. Fourth down and I suggest just punting between the thirty five- that's really what it all said, and he stopped
and I was the air with during that's where the best they were and the other college seems noticed, and they said funding between the very vague guidelines and you're actually write what has happened at the nfl while is people have noted so I see someone up hunting between thirty up going a fourth down I smile because it basically my work that this disease nor I did a while ago to yet the organ ducks the stopped. It and then you get it so so just just or understand to enforce down rather than punt to release the two too, returned the ball to the opposing team. they would go for first down and some personification. They don't. but if I am on my own thirty five, your line, I'm Hannah you the ball with the thirty five european right and you say that risk is not as great as his hand
over the ball, as it might have gotten a first down here on a purely starting ten ten yards further than than you would if I had, if you had made a fair catch. Yes, yet europe conceding ten yards, your conceding ten yards by going forward by not going for it. That's all you're really given up. You know unless there is a really good return now. Does it take that into account? Does that your exiting it took em into account. I guess you have you got a great return guy. You know that Well, it's also a function of how late again is it, and how much are you I had her behind by the actual rules it turns out. If we're doing, sports is don't put between the very fact that lines and always go for it. within one, even from your own. Ten. Well. Oh- and I told the oregon ducks coach, and you said I can't do that. I lose my job. I lose my job by giving I met my area from a straight person statistical perspective,
the environment is, if you put from your own ten you're still screw. still gonna have unbelievably good field possession. Yes, you know Go for it on your own ten, forty one, if a reasonably good chance getting it and now all of a sudden, your backing, but he just said you couldn't do that Let me just look me in the eyes, and I can't do that I'll lose my job so well Why did you lose your job anyway? he wins the nfl he had agreed. if the orthodoxy went the other what they should help. He lost his job upwards fantastic. It. You said earlier on mad about how machine learning programmes are getting quicker.
Are. We gonna get to the point where we can really start to predict some of the big natural disasters, the earthquakes the tsunami are in open, it always is, who does he gets right now, although let me be more precise, there you're, certain earthquakes, wouldn't be the best well, you're, but certainly storms, where we have limits to Heaven days in advance. You can predict the weather because there's some chaos taking over what? How does it, hey, I handled chaos any better than we ve ever handled it before so as a sort of the impression- and I think that actually comes- add to this fifty one, forty, nine versus hundreds arafat himself predicting an earthquake. That's a hundred zero thing. how many hurricanes they're gonna be this season? That's more like a fifty one. Forty nine thing dealing with chaos very much a fifty one. Forty nine thing: the stock market, a sort of chaos.
but I'm wanted I want to. I don't have to get it right all the time I just want to get it right. Most of the time. But hey I should just be cleared. Did my our unease knows more precisely what we mean by chaos, so what we learned back again in the seventies and eighties that you can start a system out with certain variables having certain values, and then you could run a system and not all systems would hey this way, but some systems you get a result, ok and then you can make it. I need to just meant in your initial parameters and then set it go. Let it go forward and you get a cup lately. Different result example that small changes in your initial conditions would not lead to small changes in your house it led to huge changes in your account, which meant that your ability to predict far into the future for some systems was essentially mathematically impossible. So this
that's what I mean by chaos here- a butterfly flaps, its wings in east africa and there's a hurricane six weeks later in the opponents and my my answer is: its again? It's two hundred zero versus fifty one, forty nine! I cannot tell you there will be a hurricane in the bahamas on october twenty, so I And tell you you're gonna be more hurricanes this year than average. in general, more hurricane, if you one forty nine, this specific hurricane that depends on that specific her fine, I don't know any better, then how about quantum computing? When you add that into the effect, because now you're looking at billions and billions of data points that are being fed to the eyes it is so far from hearing is good.
I know less about it and I want to add that I wish that I certainly have the background Now conventional machines appear to be able to process the data we need them to process and quantum computer I think, will be helpful in other ways, so the corn staff appears to be best sudden doing and almost uncountable number of things in parallel. So wonderful than into german has certain properties, and I can sort of look at hundreds of billions of into your simultaneously using the bombs. That's cool, but the machine learning stuff. I'm just trying to look for properties in enormous datasets that involve sort of looking at all the data, seeing out interact with each other in that,
We seem at least so far to be able to do with islamic computing power of computing, but we seem to be able they were keeping. There might be a way to query that same set of data with the higher performance. In computing in ways we had not thought to even ask maybe of the data, maybe interesting abiding, about their butterflies. Remember that article in the journal, a reproducible results where, I think said this on another episode. This is a journal where its inferred idle scientists who have some crazy thought that is completely stupid, but they want to publish it anyway, and so it goes into the journal of reproducible results. So one of them was the calculation haven't is hotter than hell. It looked at the thermodynamics dynamics of souls and how many people I haven't worthy versus how worthy grey headed up all the energy of the souls we need to have an enemy Heaven? What are we talking,
it's a big problem, one of them, was a guy. I was just going to say: the problem is that hell does not have air conditioning and that Heaven does it so the the yeah. This is definitely a bot from that junk in the closet or the so the so. This paper has a photo of a butterfly, and it said it captured the butterfly that caused hurricane then what is in it, but the one on the prosecutor, one butterfly that went to held back. I did not read the paper. I'd say that your misery and as a good, it's fun, I'm so mad at what I think is most fearful for me. Is when I does not just the tasks we give better than we have ever done it, but when it self learns
and achieve what some mile version of what we might call consciousness and this Artificial general intelligence, I think, is the scariest part of a high that that gets that's been discussed. Months. Could you comment on where that is today? It is scary we need to understand, is what this technology is actually doing. These things that were dealing with these shared ray I programmes They have no notion of truth. They have no notion of reality. They have no notion of fat dealing reality even or morality or any. All we're doing is trying to predict what an expert would say just what words would come out of his mouth and as a result is heard. Dont know what they're doing they just know what someone might say. I don't
Think we're anywhere near a point where these things exhibit. True general how'd, we need it, your stand when we're act when we're interacting with them. These things don't understand, it only stand. There are facts that are going to stand. The interesting- and I don't even know what a great and so when I about the phase the moon and chicken in denmark. I got back this long study of didn't exist. you had no idea. I actually asked Are you sure and responded Well, not really that shark, as this was the only study I could find it just standing by its not facts, and it has no idea that this is not If you look at the world I seemed to would what they call mirage when certainly something elusive, hallucinations lose nation and its it doesnt know it's made something, right when we talk about making something up. Just a phrase is identifying it the between actual reality and whatever you're, saying
So another is an actual reality. it is not really necessary. Guardrail is, is it is. It is unnecessary, guardrail to imbue the dish intelligence, with the concept of these things, that it will do. You know like what reality as what a fact is. What truth is I mean? necessary it would be good, but I don't know how you do it. These things so divorced from the notion of around you. Can't you say: hey there are facts, remember that this is not how they. Work is that other programme had met. Could the problem for the problem be that he's language model, a I machines, our coming of age at a time when the internet is filled with non facts. So it's not! It's fall. We fed it junk food had come around right at the beginning,
of the internet where you didn't have she went on and all the rest of this could minded it before little better than a fact, and it is a fact but these things had no notion of truth would still be true. People have tried to cure the information. I would fare and some are not trained on nonsense and they still have this problem with loose thing back. The problem is I understand that there is an abstract and abject reality in which that of which they are a part, and I think you're right. The problem is not the programmes. The problem is we need to recognise these things are divorced from reality. We to remember. If I want a website created, it's gonna do well. If I want to ask it Is there a correlation between these two crazy things I pulled out of the air? It's gonna do badly and we shouldn't pay attention to something. When I too My son, I asked barred. Is there a correlation between, phase, the moon and the amount of chicken.
In denmark, his immediate response was don't ask for that. That's stupid, I do think. There's gonna be a job here, and it's gonna be an important job, which is Why take what I want to know and like prompt programme? What prompt do I give barred to get back the most useful answer I can and to avoid all the job. That's gonna be a thing we are going to be people who are good at it. They're going to be classes that teach you how to do it. It's going to be a real skill. We are going to need okay, so now, let's make that a given. What do you do, and this may be more philosophical them? You know them than europe you know qualified to answer? I am certainly not. What do you do with the people who helped us fully use the technology for the end of
misinformation, confusion and chaos, because, even if you do everything you just said, those agents can still utilise the technology to do sums. serious harm to society correct and I think they will- and I think this gets back to what we are came out before I think you need to have trusted sources. Trust is gonna, become much more valuable because lack of trust is can be set. So much for dangerous see need have pressed it forces. You need the extent you can. have technology that can help identify. Generated images, as opposed to real images serve as there is both a technical problem? Where can I produce offer that tells me this is fake. This is real technical problem and there is a social problem. How do I get
people to care that you're looking at real information is opposed to sort of garbage. And my life, that's half my life as an educator. I believe you, but it's a and it's I think it's it is a huge part of what scientists need to do in our responsibility to do. It is even greater now, in its ever been they also make them all mom facet. This is more than one facet, I'm guessing now, because it is yes scientists, every programmers having responsibility but the lead. slave. Should I mean you, can make it governmental? Because if I go do something in another country that governments got no power So, are you gonna get space forced to oversee it now that one happen so
who will bring to bear legislation and these bad actors kept them in their place. It's gonna be a part was out of that have to be part of what that was talking about india, recognition portion. You will have to be able to recognise where these bad actors are like most of the dock web. You know where the people are. located. You you actually know where they are. You know Then they are in a country. That's not gonna do anything to them so you know. I just wonder if, if bad actors here put civilization at risk, that calls for some kind of international
Yes, oversight total, like stillness. I think that that so first of all, the technical problems here are are hard and are challenging and they're important, identifying generated text versus don cherry, and I am I am thrilled because I'm a technologist and I get to spend my productive time working on technical problems, and I don't have to solve the social problems. that's apparently needles drop it well. I m so glad to note felt for you that idea about I care about the need to inform people about what the technology can do. I do think that this particular check out he's gonna take this problem that you mentioned, that we have right. Truth has become more elusive and I think that this technology,
potentially can make it more elusive still, but it is still up to us- and I think it's still possible for us to say no we're gonna be committed, actually knowing whether guys blue before we start going, tell us telling, our friends and neighbours or color the sky is. We should check. this guy really is blue. Here's, why I'm convinced here's my source? Yes, it's trusted! So that's why I'm willing to talk to you about a part of me thinks that if ai has these, Destinations- and it doesn't really know what truth is- is nothing intelligent about it at all, so it's been misnamed, it's it's a disruptive force, in our culture and our society and maybe rebranded as artificial idiocy. I knew the bitter, yeah you were trying to get me killed in this closet,
at the back in the closet or Larry, just bring some somebody remarks here and and get your final reaction Matt. It seems to me if deep fakes become oh good, that no one can trust them, then that basically, the the internet as any source of information, and that has a positive side to it. Does it mean, for example, but let's look at two alone, we've chewing on won't I believe this stuff is long that it thought was true because it doesn't trust it is that the level of misinformation would be so total that people who were previous misinformed. We'll be worried that they be misinformed. I think that the
out of misinformation can go up. I think there will always be, I think, as likely to always be an internet. I think it's like they always have valuable factual act. not if you know what she's, which information and the trick is gonna, be to find it, and if you think about the stuff that I talked about on the top, ingleside, we need the ability to find it and on the social side we need Have the desire to find it, and I think that those thing isn't it a positive outcome? If she went on even believed this, the stuff- that's not true- it's not about that, might be a positive outcome in isolation, but if you alone A problem then so do all trying to effect positive change and well did hear that that brings us to something that we have been touched upon, and that is our ability, as
size as a society where the majority of people are scientifically literate and trained, as it's called thinkers. They know exactly where to place their trust, that that is really where the problem is Can I put it back on me that I got a train everybody to think this way come on Lila you that you know that that that's your job! I wanted We, the blame on mass. Now you might have always before we need to show Neil met you saying how eyes learning a quicker spades and he's gonna get their sooner eccentric cetera willing. then solve its own, solve this problem for us. Therefore it thank you. I told you. I don't see a reason for to expect that it will. It will fail this problem out, I check I do agree with you. I think that people don't have to be trained. The scientists
but I think they need to be trained as thinkers. They need to be able to and the information with which their presented valuator relatively dispassionately, and I think that education, broadly education is, can become much more important as we. need it more. It's off the case that, as machine start doing, the drudgery education becomes more important because we ve been, freed from the died. Tragically, we get to do the fun hard stuff People need to understand what that is, how they can contribute, and all I m all that That's gonna happen. I buy you mention education immensely to cut across you put people at handing in their homework or their assignments chris by I all we know, are we not heading down it? but where generated generations in the future will not have any desire to self educate. I think that I think there That is no. I think this thing.
Kids are handing in homer britain bye, bye, barter? What have you? I think this is a relatively I hope is a relatively temporary a normal agreeing, a greater and we'll will sort that out there. Honestly, I went out about it just means that the school system- values your grades more than you value, learning a day and god, oh you given braids, and so that that battle shift in a good way how this school systems place their value. on what it is to teach you something. Yet it may awaken students it may awaken in students a reckoning or a recognizing of that value themselves. Exactly now a exact, which I think I shall before I'd, takes us over in an exterminates us. Do these good sides of wine, There are some areas on a short term. Here's some short term, but I mean the fact that there can be so many deep face is gonna, be a short term up.
You know, problems are always opportunities in disguise. So the commitment to being able to recognise what truly is the realisation that there are facts that something that I can see society, embracing more than it has because it has to because it had all believe in reality. You get overwrought, so maybe at the end of this week, come out better our right. That's a good put them to end the sun Thank you for. Finally, brain is around, so there is some hope. Possibly I get chuck and carry out of the closet. Please it's been good. The talking matt always good to have you on the shelf. Thanks for coming coming around again and this will not be the last time we reach out to you- call them always fun. get the abbe check jerry. This has been start talks special edition, Neil digress, thyssen here, your personal ads,
will. This is looking.
Transcript generated on 2023-12-30.