WEBVTT - How Chat GPT and AI interact with the World 0:00:02.520 --> 0:00:06.840 Bloomberg Audio Studios, podcasts, radio news. 0:00:07.160 --> 0:00:10.440 You're listening to Bloomberg Business Week with Carol Messer and 0:00:10.520 --> 0:00:14.000 Tim Stenebeck on Bloomberg Radio. Look, I said we're going 0:00:14.040 --> 0:00:16.760 all in on AI. The author comes to our four 0:00:16.800 --> 0:00:20.280 o'clock hour today. Alphabet just reported we got that news 0:00:20.600 --> 0:00:23.120 just now, Caroline Hyde and man Deep Singh joining us 0:00:23.760 --> 0:00:27.720 Microsoft's GitHub agreeing to bank artificial intelligence models from Anthropic 0:00:27.760 --> 0:00:30.520 and Google into a coding assistant used by millions of 0:00:30.520 --> 0:00:33.240 software developers. And then there was that Wall Street Journal 0:00:33.240 --> 0:00:37.160 report earlier today Carol elon Musk's XAI is in talks 0:00:37.159 --> 0:00:39.960 to raise funding at a value of forty billion dollars. 0:00:40.040 --> 0:00:42.120 Yeah, that's real money. So much of the reason we 0:00:42.159 --> 0:00:45.320 are talking about AI is because of Jeffrey Hinton, Nobel 0:00:45.360 --> 0:00:47.599 Prize winner in physics for work in AI, known as 0:00:47.600 --> 0:00:50.960 the godfather of AI. He was on the most recent 0:00:50.960 --> 0:00:53.600 episode of Wall Street Week with David weston. 0:00:54.200 --> 0:00:58.640 These big eye systems, great sub goals. Now, the problem 0:00:58.640 --> 0:01:00.680 with that is if you give something the ability to 0:01:00.720 --> 0:01:04.760 create subcoals, it will quickly realize this one particular subcoal 0:01:04.800 --> 0:01:09.160 that's almost always useful. So if I have the goal 0:01:09.200 --> 0:01:12.280 of just getting more control over the world. That will 0:01:12.280 --> 0:01:15.360 help me achieve everything I want to achieve. So these 0:01:15.360 --> 0:01:17.679 things will realize that very quickly. Even if they've got 0:01:17.680 --> 0:01:20.119 no self interest. Then I understand that if I get 0:01:20.120 --> 0:01:22.640 more control, I'm it'll be better at doing what they 0:01:22.640 --> 0:01:24.560 want me to do, and so they will try and 0:01:24.560 --> 0:01:28.600 get more control. That's the beginning of a very slippery slope. 0:01:29.200 --> 0:01:32.080 A slippery slope. Indeed, perhaps our next guest work with 0:01:32.160 --> 0:01:35.360 Jeffrey Hinton developing the Boltzmann machine. It is the first 0:01:35.400 --> 0:01:38.920 learning algorithm for Mueller neural networks. And if you're scratching 0:01:38.959 --> 0:01:41.160 your et duh, you're going to get an explainer in 0:01:41.240 --> 0:01:41.760 just a moment. 0:01:42.160 --> 0:01:45.600 We're very pleased to have a doctor Terrence Sanowski with 0:01:45.680 --> 0:01:48.920 us today's Francis Crick Chair at the Salk Institute for 0:01:48.920 --> 0:01:52.600 Biological Studies, Distinguished Professor at the University of California at 0:01:52.600 --> 0:01:56.080 San Diego. He's also president of the Neural Information Processing 0:01:56.120 --> 0:01:59.720 Systems Foundation. He joined us from San Diego. His new 0:01:59.720 --> 0:02:03.240 book out. It's called chat GPT and the Future of AI, 0:02:03.480 --> 0:02:07.600 The Deep Language Revolution. Terry, good to have you with us. 0:02:07.640 --> 0:02:10.960 Congratulations on the book. You have been thinking about this 0:02:11.000 --> 0:02:15.239 stuff for more than three decades. At this point, I 0:02:15.320 --> 0:02:17.359 think a lot of people have been talking about it 0:02:17.800 --> 0:02:21.560 and thinking about investing in it for two years. At 0:02:21.560 --> 0:02:25.720 this point, are you just saying every day, finally people 0:02:25.760 --> 0:02:28.680 are taking seriously what I've been looking at for a generation. 0:02:30.960 --> 0:02:33.920 Well, it's great to be here to have this opportunity. 0:02:34.280 --> 0:02:38.600 My book launched today, and I'm really excited about it. 0:02:38.680 --> 0:02:43.800 I think that what's happening right now is really unbelievable 0:02:43.919 --> 0:02:48.520 in terms of the breath and the depth and the excitement. 0:02:48.639 --> 0:02:52.320 And so I was there at the beginning. Jeffrey Hinton 0:02:52.320 --> 0:02:56.240 and I collaborated in the nineteen eighties. All the learning 0:02:56.320 --> 0:03:00.000 algorithms that are being used today for these large languages 0:03:00.080 --> 0:03:05.200 models and deep learning were developed by us back in 0:03:05.200 --> 0:03:08.400 that era. And of course what we didn't have back 0:03:08.440 --> 0:03:12.280 then were computers that were fast enough that could scale 0:03:12.360 --> 0:03:16.960 up these models to be you know, be able to 0:03:17.000 --> 0:03:22.040 solve these very difficult problems and artificial intelligence. But this 0:03:22.240 --> 0:03:26.440 you mentioned, the Neural Information Processing Systems Foundation on the 0:03:26.440 --> 0:03:30.960 President reorganized the biggest AI meeting and in December, we're 0:03:30.960 --> 0:03:36.520 expecting sixteen thousand researchers to descend on Vancouver. After by 0:03:36.560 --> 0:03:41.720 the way Taylor Swift, she's the big headliner on Sunday, 0:03:41.760 --> 0:03:43.000 but we have the rest of the week. 0:03:43.840 --> 0:03:45.520 So I want to ask you, how did you think 0:03:45.520 --> 0:03:48.920 about neural networks in large language models in the nineteen eighties. 0:03:48.960 --> 0:03:50.240 How do you think about them today. 0:03:52.320 --> 0:03:58.280 Well, we actually had a premonition that these large language models, 0:04:00.080 --> 0:04:02.600 I should say not the large they were small language models. 0:04:02.640 --> 0:04:06.400 Back we're really good at language. And that was a 0:04:06.440 --> 0:04:09.800 particular project, a summer project for a gratitude in my 0:04:09.920 --> 0:04:13.480 lap called net talk, which was trained on a dictionary 0:04:13.880 --> 0:04:16.440 to be able to pronounce English text. You know, if 0:04:16.480 --> 0:04:18.920 you give it an article from the Wall Street Journal 0:04:19.320 --> 0:04:21.640 and they would pronounce it in an understandable way. And 0:04:21.960 --> 0:04:26.040 this in linguistics is a very difficult problem because English 0:04:26.120 --> 0:04:29.520 is so irregular. There are a lot of regularities, but 0:04:30.360 --> 0:04:33.760 you also have irregularities, and then you have rules for 0:04:33.839 --> 0:04:37.400 the irregularities. But it really was amazing that a small, 0:04:37.440 --> 0:04:40.240 tiny network with just a few hundred units and tens 0:04:40.320 --> 0:04:44.000 of thousands of weights, the parameters, the connections between the 0:04:44.120 --> 0:04:47.560 units could do that. It was like an amazing compression 0:04:47.720 --> 0:04:52.880 of complexity. And now we know that these large language models, 0:04:52.880 --> 0:04:56.760 the deep learning networks, they love language, and they are 0:04:56.880 --> 0:04:59.839 capable of things that we never could have imagined. 0:05:00.440 --> 0:05:03.160 Well, we're gonna we have a few minutes with you now, 0:05:03.320 --> 0:05:05.240 and then we're going to come back and have even 0:05:05.240 --> 0:05:07.359 more time with you. But that's really what I wanted 0:05:07.360 --> 0:05:12.839 to talk about the idea of super intelligent AI. What 0:05:12.920 --> 0:05:16.320 are we not thinking about? What's the threat out there? 0:05:18.200 --> 0:05:22.240 So you know, my good friend Jeff is very concerned, 0:05:22.279 --> 0:05:25.760 and I think he's one of the smartest people I've 0:05:25.800 --> 0:05:29.640 ever met. And if he's worried about it, then there's 0:05:29.640 --> 0:05:34.520 some as a concern. However, I think that even if 0:05:34.560 --> 0:05:38.760 you're concerned, it's very difficult to know when that's going 0:05:38.800 --> 0:05:42.280 to happen, if it ever will happen. And there are 0:05:42.360 --> 0:05:46.000 super forecasters out there, and this is from the Economist magazine, 0:05:46.440 --> 0:05:50.360 who are much better at people who are experts at 0:05:50.400 --> 0:05:53.279 predicting you know, if and when there may be a 0:05:53.320 --> 0:05:58.400 catastrophic or existential threat, and it turns out that in fact, 0:05:58.440 --> 0:06:05.000 they're not as a concerned as the experts in AI. 0:06:06.400 --> 0:06:11.159 I'm happy that someone is thinking about the worst case 0:06:11.240 --> 0:06:14.279 outcome because if not, then if it ever happens, we're 0:06:14.279 --> 0:06:18.200 in trouble. But right now, I'm more concerned about trying 0:06:18.240 --> 0:06:23.760 to understand how they work mathematically and also to learn 0:06:23.960 --> 0:06:27.400 more from the brain. After all these were designed. Back then, 0:06:27.600 --> 0:06:30.719 Jeff and I looked at the brain the only existence 0:06:30.720 --> 0:06:33.480 proof you could solve any problem in AI. And you know, 0:06:33.560 --> 0:06:36.880 we tried to build something that was based on similar principles. 0:06:37.320 --> 0:06:39.520 So now we can continue. There's a lot more in 0:06:39.560 --> 0:06:40.680 the brain we can learn from. 0:06:40.760 --> 0:06:42.599 But paint that picture for us, because I think a 0:06:42.600 --> 0:06:45.520 lot of people are worried about doomsday scenarios here, and 0:06:45.839 --> 0:06:49.599 if Jeffrey Hinton is worried about that stuff, I mean, 0:06:50.480 --> 0:06:52.159 should we we should be worried about it. 0:06:52.200 --> 0:06:58.400 You're saying, I think that we should be cautious, that 0:06:58.560 --> 0:07:03.760 is to say, we should be uh constantly thinking along 0:07:03.960 --> 0:07:06.320 the lines that Jeff is in terms of what could 0:07:06.360 --> 0:07:09.600 possibly happen, and you know, be cautious and put in 0:07:09.880 --> 0:07:12.520 precautions so that it can't happen. 0:07:12.560 --> 0:07:17.960 What sorry, I just yeah, go ahead. 0:07:18.840 --> 0:07:22.000 What I'm really concerned about are the unintended consequences, things 0:07:22.040 --> 0:07:25.320 that you cannot predict. Something may happen that you know, 0:07:25.400 --> 0:07:26.680 no one thought of, even Jeff. 0:07:27.560 --> 0:07:30.040 Yeah, and like you know, we have learned certainly right 0:07:30.080 --> 0:07:35.120 great financial crisis pandemic like the un the unthinkable can 0:07:35.240 --> 0:07:37.600 happen and you throw technology into it and you just 0:07:37.680 --> 0:07:41.760 kind of don't know where it's gonna go exactly. Okay, 0:07:41.760 --> 0:07:47.000 so now I'm terrified. Okay, Terrence Stoke go anywhere. We're 0:07:47.000 --> 0:07:49.480 gonna do some news. I do wonder what. 0:07:50.520 --> 0:07:53.480 Yeah, I think we have time for one more question 0:07:53.600 --> 0:07:54.160 before we go. 0:07:54.400 --> 0:07:57.040 Well, so you know, okay, we have a minute and 0:07:57.040 --> 0:07:58.680 then we're gonna take a break and come back. But 0:07:58.720 --> 0:08:01.679 I just do wonder. You know, when you talk to people, 0:08:01.720 --> 0:08:03.360 do you say, wait, this is really going to be 0:08:03.480 --> 0:08:04.680 net net a good thing? 0:08:07.280 --> 0:08:12.440 Look, all new technologies have good and bad consequences, and 0:08:12.600 --> 0:08:16.040 you try to mitigate the bad, and you you have 0:08:16.120 --> 0:08:19.520 to balance them, you know. Yeah, and right now it 0:08:19.560 --> 0:08:21.600 looks like the good is way way ahead of the 0:08:21.600 --> 0:08:23.920 bad in terms of the impact it may have on 0:08:24.040 --> 0:08:28.360 us and society and businesses. But you know, like I say, 0:08:28.400 --> 0:08:30.880 we have to be careful because we don't really know 0:08:30.920 --> 0:08:31.559 where it's heading. 0:08:32.160 --> 0:08:33.920 You know what worries me too, And we will talk 0:08:33.960 --> 0:08:35.520 about this maye when we come back. I feel like 0:08:35.559 --> 0:08:40.600 we throw around a lot of words, not you, but 0:08:40.720 --> 0:08:42.560 all of us in the general like you know, whether 0:08:42.600 --> 0:08:45.840 it's you know, AI, generative AI, and like you know, 0:08:45.920 --> 0:08:48.360 and I don't know that we really understand what's going on, 0:08:48.960 --> 0:08:52.000 and so it's hard to know what it could possibly become. 0:08:52.080 --> 0:08:54.320 So we're going to pick your brain a little bit 0:08:54.320 --> 0:08:55.440 more Terry when we come back. 0:08:55.600 --> 0:08:58.720 It's Terry Sanowski. He's a Francis Kirk Charity Sauk Institute 0:08:58.720 --> 0:09:02.520 for Biological Studies, Distinguished Professor at the University of California 0:09:02.640 --> 0:09:06.280 at San Diego, President of the Neural Information Processing Systems Foundation. 0:09:06.360 --> 0:09:09.079 The new book out now, Chat GPT in the Future 0:09:09.120 --> 0:09:13.080 of AI The Deep Language Revolution More with doctor Sanowski. 0:09:13.160 --> 0:09:15.000 Right after this, I want to get back to our guests. 0:09:15.000 --> 0:09:19.000 We're talking with doctor Terrence Sanowski. He is Francis Crick 0:09:19.080 --> 0:09:22.520 Chair at the Salk Institute for Biological Studies, Distinguished Professor 0:09:22.520 --> 0:09:25.679 at the University of California at San Diego. He's also 0:09:25.760 --> 0:09:29.600 president of the Neural Information Processing Systems Foundation, and he's 0:09:29.679 --> 0:09:32.320 joining us from San Diego on this Tuesday. His new 0:09:32.320 --> 0:09:34.800 book Chat CHEPT in the Future of AI, The Deep 0:09:34.880 --> 0:09:38.040 Language Revolution. I got to ask you because I am 0:09:38.080 --> 0:09:40.360 still trying to understand and I get worried that we 0:09:40.440 --> 0:09:43.920 throw these words around. Certainly not you, but we as 0:09:43.960 --> 0:09:47.080 we try to understand this with now, you know, not 0:09:47.200 --> 0:09:51.800 having full comprehension of what art artificial intelligence, the large 0:09:51.840 --> 0:09:54.520 language models that we're talking about today, where it takes us. 0:09:54.960 --> 0:09:58.800 Is it as subtle at evolving in life changing as 0:09:58.840 --> 0:10:00.920 the Internet was for us. 0:10:02.040 --> 0:10:07.360 So this is something that is emerging. And I have 0:10:07.760 --> 0:10:10.280 since the book was sent to press in the summer, 0:10:10.320 --> 0:10:13.319 I have a sub stack where I have tried to 0:10:13.400 --> 0:10:15.600 fill in with, you know, the new things that are happening. 0:10:15.600 --> 0:10:22.200 And I'm preparing something a new twelfth version the blog 0:10:22.400 --> 0:10:26.840 on the question of whether AI is overhyped or under hyped, 0:10:27.880 --> 0:10:30.120 and and you know, I've thought a lot about this, 0:10:30.320 --> 0:10:34.320 and you know, I think that it depends on the timescale. 0:10:34.360 --> 0:10:37.760 I think that on the short timescale it is overhyped. 0:10:37.760 --> 0:10:40.080 There's no doubt about it. There's just so much out there. 0:10:40.120 --> 0:10:42.959 I mean, every day the newspapers are filled with AI 0:10:43.080 --> 0:10:46.080 and your program. But I think in the long run 0:10:46.120 --> 0:10:49.080 it's actually under hyped. I think the real change in 0:10:49.120 --> 0:10:52.520 the Internet, for example, didn't occur within the first ten years. 0:10:52.559 --> 0:10:58.120 It was much later. Again, unintended applications that merge that 0:10:58.240 --> 0:11:02.000 you know, have enormous impact on our lives, like social media, 0:11:02.080 --> 0:11:04.200 So I think the same thing's going to happen with AI. 0:11:04.880 --> 0:11:09.240 But is it is it different? Like I guess, I'm like, 0:11:09.600 --> 0:11:12.319 what do you mean, Like, the Internet is not I 0:11:12.640 --> 0:11:15.120 wanted to say comfortable, but it's not because there's some 0:11:15.200 --> 0:11:18.680 really bad things that happen and we know that, right, 0:11:18.720 --> 0:11:20.559 and that's the battle we have with social media. And 0:11:20.600 --> 0:11:22.280 we want to talk to you about kind of regulatory 0:11:22.320 --> 0:11:24.800 oversight of AI in a moment. But I just I'm 0:11:24.840 --> 0:11:27.320 just trying to understand. You know, it does feel so 0:11:27.440 --> 0:11:30.240 seamless and just such a part of everything we do. 0:11:31.920 --> 0:11:35.000 But it hasn't necessarily replaced a ton of jobs. It's 0:11:35.040 --> 0:11:37.920 created jobs, It's replaced some jobs. I guess you could say, 0:11:38.679 --> 0:11:42.040 I'm just trying to understand, Like on what scale do 0:11:42.080 --> 0:11:45.360 you put it? You mentioned the internet, So is it 0:11:45.400 --> 0:11:47.679 apples to apples or is it something else? 0:11:48.920 --> 0:11:52.360 No, Well, first of all, it'll it uses the internet, 0:11:52.400 --> 0:11:57.600 So I mean that's like the the machinery that you 0:11:57.720 --> 0:12:00.880 need to reach to scale up and reach a large population. 0:12:02.080 --> 0:12:06.040 But it's more intimate than the Internet in the following 0:12:06.120 --> 0:12:09.079 sense that it talks to us, right, I mean It's 0:12:09.120 --> 0:12:11.920 as if an alien landed on the Earth and could 0:12:11.960 --> 0:12:15.679 talk to us in English, and it knew everything about 0:12:15.720 --> 0:12:19.320 you know, what, humans, history, everything, and the only thing 0:12:19.360 --> 0:12:21.920 we can be sure if it's not human. But it's 0:12:22.440 --> 0:12:25.480 really quite remarkable. Let me give you one example of 0:12:25.520 --> 0:12:28.760 something that I was really surprised at when they did 0:12:28.760 --> 0:12:32.360 a study of whether people who needed cognitive therapy preferred 0:12:33.040 --> 0:12:38.920 real humans or AI. They preferred AI, which was really 0:12:39.000 --> 0:12:41.360 quite remarkable. I didn't expect that. And part of the 0:12:41.360 --> 0:12:45.760 reason is that the AI is not judgmental like humans. 0:12:46.760 --> 0:12:49.120 Well wait, but isn't it depends on the data, Like 0:12:49.160 --> 0:12:50.360 we talk about, it wasn't. 0:12:50.160 --> 0:12:53.760 Getting trained on judgmental data. 0:12:54.520 --> 0:12:58.640 It was you know, Actually, it's a good question. What 0:12:58.760 --> 0:13:00.760 was it trained on. I think that it was fine 0:13:00.800 --> 0:13:06.120 tuned with you know, data from real subjects that we're 0:13:06.160 --> 0:13:11.880 that we're talking with a doctor. But even without that, 0:13:12.040 --> 0:13:15.199 I'll tell you something again, it's shocking is that it 0:13:15.559 --> 0:13:21.400 is actually empathy. These large language models can empathize with humans. 0:13:21.920 --> 0:13:23.960 And why is it? How is that? Well, it actually 0:13:24.040 --> 0:13:28.480 absorbed a lot of text out there, novels, you know, letters, 0:13:28.559 --> 0:13:32.080 and read it and so forth, and where empathy was 0:13:32.679 --> 0:13:37.400 being part of the discussion, and so it absorbed that 0:13:37.440 --> 0:13:38.559 too well. 0:13:38.559 --> 0:13:41.960 I wanted to hear a little bit of your thoughts 0:13:41.960 --> 0:13:44.760 on what we heard from Elon Musk a little earlier today. 0:13:44.760 --> 0:13:48.040 He actually participated in a surprise conversation at the Future 0:13:48.040 --> 0:13:50.839 Investment Initiative to discuss the future of AI. 0:13:51.240 --> 0:13:53.120 It's most likely going to be great, and there's this 0:13:53.280 --> 0:13:58.400 some chance which could be tense, that it goes bad. 0:14:00.200 --> 0:14:03.560 The chances on zero that it goes bad. But overall, 0:14:03.559 --> 0:14:05.840 at one point you said that covers eighty percent full 0:14:06.960 --> 0:14:08.240 is one positive way to look at it. 0:14:08.280 --> 0:14:12.800 Maybe ninety percent, okay, eighty or ninety percent positive. The 0:14:12.880 --> 0:14:17.880 question I have for you, professor, is do we need 0:14:18.760 --> 0:14:22.640 an international regulatory body? Do we need the largest, most 0:14:22.640 --> 0:14:25.160 powerful governments around the world to create some sort of 0:14:26.080 --> 0:14:29.720 standard to ensure or help ensure that this goes the 0:14:29.760 --> 0:14:30.160 right way. 0:14:31.720 --> 0:14:36.200 Well, as you know, in the UK they passed an 0:14:36.240 --> 0:14:40.080 AI Act which is like one hundred pages long and 0:14:40.240 --> 0:14:46.240 you know, incredibly detailed, and it's already absolute. I just 0:14:46.400 --> 0:14:50.120 moving blasting forward and you know you're trying to catch 0:14:50.160 --> 0:14:54.280 up with it. But I do believe that it's absolutely 0:14:54.360 --> 0:14:58.359 essential that it be regulated, and it should be regulated 0:14:59.000 --> 0:15:02.760 by people who a building it. The government, okay, is 0:15:04.240 --> 0:15:07.720 the business of protecting people. And we'll see how that 0:15:07.760 --> 0:15:13.240 plays out. But in for example, genetics, this happened, you know, 0:15:13.320 --> 0:15:16.640 back in the sixties seventies. They had a meeting where 0:15:16.640 --> 0:15:19.600 they came together at a solomar and they came up 0:15:19.640 --> 0:15:24.920 with a containment rules and regulations for doing experiments under 0:15:24.960 --> 0:15:30.200 the careful protection so that nothing leaks out, nothing gets out. 0:15:30.680 --> 0:15:32.160 And I think we need to do the same. 0:15:34.320 --> 0:15:37.200 Okay, So when does as you said, ten years for 0:15:37.240 --> 0:15:40.800 the internet to really kind of make its impact and 0:15:40.880 --> 0:15:45.240 presence really known and maybe you know, integrated into our lives. 0:15:45.360 --> 0:15:50.760 So is it a decade before we see LMS and 0:15:51.320 --> 0:15:54.280 AI at this level integrated into our lives. 0:15:56.000 --> 0:15:59.040 We are at a stage that aviation was at when 0:16:00.240 --> 0:16:04.560 at Kitty Hawk the Wright brothers made the first flight. 0:16:04.760 --> 0:16:07.640 It was ten feet up and one hundred feet long, 0:16:08.280 --> 0:16:12.360 and that really was the you know, something that then 0:16:12.680 --> 0:16:16.520 took decades and decades to build. And the most difficult thing, 0:16:16.560 --> 0:16:20.080 by the way that airplanes, you know, design of airplanes 0:16:20.120 --> 0:16:23.040 had to solve was control. How do you how do 0:16:23.080 --> 0:16:25.280 you make it go where you want to go without 0:16:25.320 --> 0:16:28.760 crashing and that's something that again it's like we're going 0:16:28.760 --> 0:16:32.240 through right now with AI. And yes, it will take decades. 0:16:32.280 --> 0:16:33.600 It's not going to happen overnight. 0:16:37.040 --> 0:16:37.280 I know. 0:16:37.680 --> 0:16:39.800 It's just it's kind of fascinating. We have a million 0:16:39.840 --> 0:16:41.760 more questions. Is it going to take all the jobs? 0:16:41.840 --> 0:16:43.160 Is it going to is it going to create jobs? 0:16:43.160 --> 0:16:45.080 Is it going to take jobs? Just got thirty seconds 0:16:45.320 --> 0:16:45.880 and yes. 0:16:45.760 --> 0:16:50.240 Yes, you know, I get asked that question whenever I 0:16:50.240 --> 0:16:53.480 give a public talk, and my answer is that you 0:16:53.520 --> 0:16:55.720 won't lose your job, but your job's going to change 0:16:55.760 --> 0:16:58.080 and you're going to need new skills, and you know 0:16:58.200 --> 0:17:01.480 it will morph over time. Now, you know these are 0:17:01.520 --> 0:17:03.680 people who are in the workforce now, but you know 0:17:03.760 --> 0:17:06.639 young people coming up, they'll have no trouble whatsoever finding 0:17:07.040 --> 0:17:08.560 new jobs in this new industry. 0:17:09.280 --> 0:17:12.440 We gotta get you back on the show, doctor Sanowski. 0:17:12.520 --> 0:17:13.199 Really appreciate you. 0:17:13.560 --> 0:17:15.199 Just say, Terry, what we always want to know is, 0:17:15.240 --> 0:17:17.560 like you know, people like us, you know anchors. 0:17:17.720 --> 0:17:19.280 We don't have time for him to answer that question. 0:17:19.480 --> 0:17:21.680 We don't have time for him to answer that question. Carol, 0:17:21.720 --> 0:17:23.359 I'm sorry, I got to give the book a play. 0:17:23.440 --> 0:17:24.880 Well, come back hopefully. 0:17:24.560 --> 0:17:26.960 The new book Chat GPT in the future of AI, 0:17:27.280 --> 0:17:31.720 the Deep Language Revolution, Doctor Terry Sanowski. He was there 0:17:31.760 --> 0:17:34.440 at the beginning. He knows it all. He's Francis Krickchair 0:17:34.560 --> 0:17:38.399 at the Salk Constitute for Biological Studies, among many other things. 0:17:38.440 --> 0:17:40.080 This is Bloomberg BusinessWeek.