1 00:00:15,920 --> 00:00:19,119 Speaker 1: This is Wall Street Week. I'm David Weston bringing you 2 00:00:19,400 --> 00:00:23,479 Speaker 1: stories of capitalism. This week an industrial boom south of 3 00:00:23,520 --> 00:00:26,360 Speaker 1: the border. We went to Mexico to see how tariffs 4 00:00:26,400 --> 00:00:30,120 Speaker 1: and transportation costs are reshaping the American supply chain and 5 00:00:30,520 --> 00:00:33,600 Speaker 1: what that could mean for China. But we begin with 6 00:00:33,680 --> 00:00:37,199 Speaker 1: a biography, not one about a person or even really 7 00:00:37,320 --> 00:00:37,720 Speaker 1: a thing. 8 00:00:38,320 --> 00:00:39,080 Speaker 2: How do we. 9 00:00:39,080 --> 00:00:43,519 Speaker 1: Classify something like artificial intelligence, something that has never existed 10 00:00:43,560 --> 00:00:47,559 Speaker 1: before and something that might be beyond our understanding. This 11 00:00:47,640 --> 00:00:51,080 Speaker 1: is a story about identity, who we are as humans, 12 00:00:51,280 --> 00:00:54,680 Speaker 1: whether our identity is being merged with computers, and if so, 13 00:00:55,280 --> 00:00:56,440 Speaker 1: what we need to do about it. 14 00:00:58,760 --> 00:01:00,440 Speaker 2: There's nothing special that people. 15 00:01:00,720 --> 00:01:04,560 Speaker 3: They are very wonderful things, and incredibly we're incredibly complicated, 16 00:01:05,880 --> 00:01:08,160 Speaker 3: and we're very wonderful to other people. 17 00:01:09,600 --> 00:01:12,199 Speaker 2: That's people and what we care about. We've evolved that way. 18 00:01:13,800 --> 00:01:18,560 Speaker 2: But there's nothing that you can't replicate in other material. 19 00:01:19,240 --> 00:01:22,400 Speaker 1: If anyone should know about whether computers can do everything 20 00:01:22,440 --> 00:01:25,760 Speaker 1: that humans can, it would be Jeffrey Hinton, the Computer 21 00:01:25,880 --> 00:01:29,640 Speaker 1: scientists and University of Toronto professor emeritus, is known as 22 00:01:29,680 --> 00:01:33,360 Speaker 1: the godfather of AI. After working for over a decade 23 00:01:33,360 --> 00:01:35,880 Speaker 1: at Google, he quit his job in twenty twenty three 24 00:01:36,040 --> 00:01:39,440 Speaker 1: and became outspoken about the risks of the technology. He 25 00:01:39,440 --> 00:01:44,000 Speaker 1: helped pioneer this, even as he continues to receive accolades 26 00:01:44,040 --> 00:01:46,920 Speaker 1: for his work. We sat down with him after he 27 00:01:46,959 --> 00:01:50,160 Speaker 1: had just received the Nobel Prize in physics. 28 00:01:50,840 --> 00:01:52,720 Speaker 3: I was in a hotel in California and my phone 29 00:01:52,720 --> 00:01:54,800 Speaker 3: went off at one in the morning, and I thought 30 00:01:54,800 --> 00:01:56,800 Speaker 3: about not answering it, and then they said I won 31 00:01:56,840 --> 00:02:01,160 Speaker 3: the Nobel Prize in physics, and I thought, wait a minute, 32 00:02:01,240 --> 00:02:02,400 Speaker 3: I'm not a physicist. 33 00:02:02,200 --> 00:02:03,240 Speaker 2: This can't be right. 34 00:02:03,680 --> 00:02:06,440 Speaker 1: But it was right. And in December it will be 35 00:02:06,480 --> 00:02:10,240 Speaker 1: Professor Hinton, along with John Hopfield, who will be receiving 36 00:02:10,280 --> 00:02:13,160 Speaker 1: the award at a ceremony in Stockholm for their breakthrough 37 00:02:13,200 --> 00:02:17,480 Speaker 1: work on machine learning. That generative AI is reshaping the 38 00:02:17,520 --> 00:02:22,280 Speaker 1: business landscape is beyond doubt. Bloomberg Intelligence expects it to 39 00:02:22,400 --> 00:02:25,640 Speaker 1: drive big tech firms to nearly double their CAPEX spending 40 00:02:25,680 --> 00:02:29,320 Speaker 1: from twenty twenty three to twenty twenty five, with revenue 41 00:02:29,360 --> 00:02:32,480 Speaker 1: from AI set to approach a trillion dollars by the 42 00:02:32,600 --> 00:02:35,120 Speaker 1: end of the decade. But before we get to the 43 00:02:35,160 --> 00:02:38,520 Speaker 1: future of AI, we turned to its past. We asked 44 00:02:38,560 --> 00:02:41,200 Speaker 1: Jeffrey Hinton to go back to the beginning, when this 45 00:02:41,360 --> 00:02:45,040 Speaker 1: new creature that can do what humans can do first arrived. 46 00:02:46,280 --> 00:02:50,320 Speaker 3: I would say it was born in about nineteen eighty 47 00:02:50,440 --> 00:02:56,160 Speaker 3: three with something that Terry Sinowski and I did, which 48 00:02:56,240 --> 00:02:59,320 Speaker 3: was called the Boltsham Machine. It never really panned out, 49 00:02:59,400 --> 00:03:01,800 Speaker 3: but it was a generaty of AI. It was a 50 00:03:01,800 --> 00:03:04,520 Speaker 3: model that would generate images, not very good images, but 51 00:03:04,520 --> 00:03:05,399 Speaker 3: it would generate them. 52 00:03:05,560 --> 00:03:06,160 Speaker 4: Well. 53 00:03:06,400 --> 00:03:08,639 Speaker 5: There could be a debate about the birth of it, 54 00:03:08,680 --> 00:03:12,519 Speaker 5: but I would go all the way back to Alan 55 00:03:12,600 --> 00:03:16,120 Speaker 5: Turing and then John von Neumann, so that would be 56 00:03:16,520 --> 00:03:19,200 Speaker 5: in the thirties, forties and fifties. 57 00:03:19,720 --> 00:03:22,680 Speaker 1: Marty Chavez of Sixth Street has spent his career developing 58 00:03:22,720 --> 00:03:25,519 Speaker 1: AI and he traces its birth back to the very 59 00:03:25,560 --> 00:03:27,720 Speaker 1: foundations of computer science. 60 00:03:28,480 --> 00:03:32,200 Speaker 5: So it's been around a while, right, people thinking about 61 00:03:32,560 --> 00:03:37,680 Speaker 5: what is computation, What is the nature of computation? And 62 00:03:38,240 --> 00:03:42,720 Speaker 5: do human beings do computations? Is everything that's happening in 63 00:03:42,720 --> 00:03:46,000 Speaker 5: our brains or our minds just a form of computation. 64 00:03:46,120 --> 00:03:50,240 Speaker 5: People were asking these questions really almost almost one hundred 65 00:03:50,320 --> 00:03:52,200 Speaker 5: years ago and starting to work on it. 66 00:03:52,320 --> 00:03:55,040 Speaker 1: Whatever the beginnings of AI, most agree that it took 67 00:03:55,200 --> 00:03:58,400 Speaker 1: many people in many decades to get here, But they 68 00:03:58,400 --> 00:04:01,880 Speaker 1: also agree that in just the last few years the 69 00:04:01,960 --> 00:04:05,480 Speaker 1: game has changed. A series of rapid breakthroughs has brought 70 00:04:05,520 --> 00:04:09,000 Speaker 1: what were far out predictions much closer to reality. 71 00:04:09,720 --> 00:04:14,520 Speaker 3: The big milestone is the large language models, which work 72 00:04:14,600 --> 00:04:18,160 Speaker 3: by looking at a context and then generating guesses about 73 00:04:18,160 --> 00:04:19,560 Speaker 3: what they think the next word should be. 74 00:04:19,960 --> 00:04:21,000 Speaker 2: That's why they generative. 75 00:04:22,000 --> 00:04:24,800 Speaker 3: Then that was followed by things like Dali from open 76 00:04:24,839 --> 00:04:27,680 Speaker 3: Ai that could generate images. Before that, there were things 77 00:04:27,680 --> 00:04:30,880 Speaker 3: that could generate captions for images, so they would look 78 00:04:30,880 --> 00:04:33,360 Speaker 3: at an image and then they would generate a string 79 00:04:33,360 --> 00:04:34,760 Speaker 3: of words that describe the image. 80 00:04:36,560 --> 00:04:38,880 Speaker 2: Those were the early versions of generative area. 81 00:04:39,080 --> 00:04:43,600 Speaker 3: Now we're getting multimodal chatbots which can generate everything. They 82 00:04:43,600 --> 00:04:45,240 Speaker 3: can generate images and language. 83 00:04:45,480 --> 00:04:50,480 Speaker 5: The huge milestone that I think of is one that 84 00:04:50,560 --> 00:04:55,960 Speaker 5: we hit twenty sixteen twenty seventeen. This was a milestone 85 00:04:56,080 --> 00:04:59,279 Speaker 5: in the evolution of the of the neural networks. It 86 00:04:59,360 --> 00:05:03,240 Speaker 5: maybe didn't get a lot of public attention, but it 87 00:05:03,360 --> 00:05:07,240 Speaker 5: certainly got the attention of everybody in the field. And 88 00:05:07,920 --> 00:05:11,839 Speaker 5: this milestone came about because of the work of many 89 00:05:11,880 --> 00:05:15,360 Speaker 5: of the researchers. I mentioned a few of them for 90 00:05:15,520 --> 00:05:19,720 Speaker 5: over decades, but another contributor to the milestone was the 91 00:05:19,760 --> 00:05:25,240 Speaker 5: Internet and the a sudden availability of vast amounts of 92 00:05:25,839 --> 00:05:27,120 Speaker 5: labeled data. 93 00:05:27,640 --> 00:05:30,520 Speaker 1: The breakthroughs that Hinton and others worked to achieve have 94 00:05:30,600 --> 00:05:33,960 Speaker 1: developed artificial intelligence to the point that now, in a 95 00:05:34,000 --> 00:05:38,119 Speaker 1: text conversation, it might be indistinguishable from a human being. 96 00:05:38,720 --> 00:05:41,440 Speaker 1: A study this year by cognitive scientists that you see 97 00:05:41,480 --> 00:05:44,719 Speaker 1: San Diego showed that people mistook chat GPT for a 98 00:05:44,800 --> 00:05:48,039 Speaker 1: human more than fifty percent of the time, clearing the 99 00:05:48,080 --> 00:05:51,920 Speaker 1: bar for the so called touring test. Humans themselves only 100 00:05:51,960 --> 00:05:54,479 Speaker 1: passed the test about two thirds of the time. But 101 00:05:54,560 --> 00:05:58,480 Speaker 1: as surprising as that is, it could be just the beginning. 102 00:05:59,200 --> 00:06:02,160 Speaker 1: How far along is AI in its life? Again, if 103 00:06:02,160 --> 00:06:06,160 Speaker 1: we're anthropomorphizing, here, is it a child? Is an adolescent? 104 00:06:06,200 --> 00:06:06,640 Speaker 1: Where is it? 105 00:06:06,560 --> 00:06:12,880 Speaker 3: It's more like a toddler, maybe like a three year old. 106 00:06:13,240 --> 00:06:17,000 Speaker 5: Let's imagine that we're in that gangly, awkward teenager phase 107 00:06:17,440 --> 00:06:23,440 Speaker 5: of AI. What happens next? There's a huge debate. 108 00:06:23,680 --> 00:06:26,919 Speaker 1: Whether AI is a toddler or an adolescent. Today. The 109 00:06:27,000 --> 00:06:29,920 Speaker 1: question is how far can it go? How much can 110 00:06:29,960 --> 00:06:33,120 Speaker 1: it learn? And perhaps most important, will it catch up 111 00:06:33,160 --> 00:06:36,599 Speaker 1: to us humans? For Jeffrey Hinton. It's a matter of when, 112 00:06:37,120 --> 00:06:42,360 Speaker 1: not if. Does that mean that general of artificial intelligence 113 00:06:43,160 --> 00:06:46,200 Speaker 1: will be able to do whatever humans can do? Is 114 00:06:46,240 --> 00:06:49,800 Speaker 1: there anything that it cannot do that we humans can do? 115 00:06:50,200 --> 00:06:53,799 Speaker 3: Yes, most people have the belief there's something very special 116 00:06:53,839 --> 00:06:58,039 Speaker 3: at people, which is consciousness or sentience or subjective experience, 117 00:06:58,520 --> 00:06:59,599 Speaker 3: and we have this in a theater. 118 00:07:00,120 --> 00:07:01,080 Speaker 2: Probably all nonsense. 119 00:07:01,440 --> 00:07:04,279 Speaker 3: The history of it is, these large language models predicting 120 00:07:04,279 --> 00:07:07,640 Speaker 3: the next words using backpropagation were initially designed as a 121 00:07:07,680 --> 00:07:10,840 Speaker 3: model of how people work. Now there's a whole bunch 122 00:07:10,880 --> 00:07:13,000 Speaker 3: of people who believe in symbolic logic and that's how 123 00:07:13,000 --> 00:07:15,720 Speaker 3: people must think. They think it must be true, who 124 00:07:15,720 --> 00:07:18,200 Speaker 3: think people work a completely different way, and they're wrong. 125 00:07:18,240 --> 00:07:20,200 Speaker 3: They never managed to produce stuff that works nearly as 126 00:07:20,240 --> 00:07:24,120 Speaker 3: well as these big language models. So they'll tell you 127 00:07:24,160 --> 00:07:26,320 Speaker 3: these big language models are quite unlike people. They're just 128 00:07:26,440 --> 00:07:28,680 Speaker 3: using correlations and so on, and it's all just not 129 00:07:28,880 --> 00:07:32,800 Speaker 3: really understanding. No, our best model of how people understand 130 00:07:33,160 --> 00:07:35,160 Speaker 3: is exactly the same model as how these big language 131 00:07:35,160 --> 00:07:38,360 Speaker 3: models understand. They're just like us, and they're much more 132 00:07:38,400 --> 00:07:41,160 Speaker 3: like us than they are like standard computer software. So 133 00:07:41,360 --> 00:07:44,320 Speaker 3: understanding is all about converting words into features, having the 134 00:07:44,320 --> 00:07:48,800 Speaker 3: features interact, and having those predict the next word. We 135 00:07:48,840 --> 00:07:50,600 Speaker 3: do it pretty much the same way they do it. 136 00:07:50,640 --> 00:07:54,960 Speaker 1: What about creativity? Can general AI be as creative as 137 00:07:55,000 --> 00:07:55,520 Speaker 1: humans are? 138 00:07:56,200 --> 00:07:58,840 Speaker 3: As someone else might have said, There you go again, 139 00:08:00,240 --> 00:08:03,920 Speaker 3: you want people to be special. If you take standard 140 00:08:03,920 --> 00:08:07,160 Speaker 3: tests of creativity, AIS already do better than most people. 141 00:08:07,520 --> 00:08:11,680 Speaker 1: What about in a large corporation, you have worker bees 142 00:08:11,760 --> 00:08:13,840 Speaker 1: we often have heard of them. Then you have middle management, 143 00:08:14,200 --> 00:08:17,080 Speaker 1: upper management and CEOs? Could they replace CEOs? 144 00:08:17,800 --> 00:08:20,440 Speaker 3: Okay, so there's a good scenario for the future which 145 00:08:20,480 --> 00:08:21,040 Speaker 3: goes like this. 146 00:08:21,600 --> 00:08:22,880 Speaker 2: Suppose you have a big. 147 00:08:22,600 --> 00:08:26,440 Speaker 3: Company with a rather dumb CEO. Maybe he's the son 148 00:08:26,480 --> 00:08:30,720 Speaker 3: of the previous CEO, but he has an executive assistant. 149 00:08:30,280 --> 00:08:32,800 Speaker 2: Who's very good, almost certainly a woman. 150 00:08:33,040 --> 00:08:36,719 Speaker 3: This executive assistant is asked to achieve things by the 151 00:08:36,800 --> 00:08:39,319 Speaker 3: dumb CEO, and she achieves them in ways he doesn't 152 00:08:39,360 --> 00:08:42,000 Speaker 3: understand and would never have thought of, so he becomes 153 00:08:42,040 --> 00:08:45,320 Speaker 3: much more effective. He gets what he wants, he thinks 154 00:08:45,360 --> 00:08:49,240 Speaker 3: he's doing it. It all happens behind the scenes and 155 00:08:49,360 --> 00:08:53,080 Speaker 3: very efficiently. That's how it could be if we all 156 00:08:53,200 --> 00:08:56,719 Speaker 3: had our own AI assistants who were smarter than us, 157 00:08:56,760 --> 00:08:58,680 Speaker 3: and they were thoroughly benevolent, and they never wanted to 158 00:08:58,720 --> 00:09:02,720 Speaker 3: take over because somehow achieve that that's the good scenario. 159 00:09:03,160 --> 00:09:05,760 Speaker 1: I had a boss once in business who said, a 160 00:09:05,880 --> 00:09:09,960 Speaker 1: danger is wishing yourselves to success. Do we wish ourselves 161 00:09:10,080 --> 00:09:14,280 Speaker 1: to a safer world by hoping that General AI cannot 162 00:09:14,280 --> 00:09:16,360 Speaker 1: do everything we can do, because it might be very 163 00:09:16,400 --> 00:09:18,600 Speaker 1: threatening to many people if we believed it can do 164 00:09:18,720 --> 00:09:19,679 Speaker 1: everything we can do. 165 00:09:20,880 --> 00:09:23,720 Speaker 3: Yes, it is very threatening and I'm very worried about it, 166 00:09:24,679 --> 00:09:29,920 Speaker 3: and we should definitely put huge effort right now into 167 00:09:29,960 --> 00:09:31,760 Speaker 3: figuring out whether we can keep control of it. 168 00:09:32,080 --> 00:09:34,319 Speaker 1: What effort should we put forward, what should we do? 169 00:09:34,840 --> 00:09:38,560 Speaker 3: Okay, so for something like climate change, everybody knows what 170 00:09:38,600 --> 00:09:41,600 Speaker 3: we should do. There's a very simple solution, just don't 171 00:09:41,600 --> 00:09:45,400 Speaker 3: burn carbon. It's just we don't have the political will. 172 00:09:45,640 --> 00:09:47,240 Speaker 2: We still give big subsidies to. 173 00:09:47,200 --> 00:09:51,839 Speaker 3: Oil companies, which is lunatic. So we know what to do, 174 00:09:51,880 --> 00:09:56,120 Speaker 3: we just won't do it with AI. To keep AI 175 00:09:56,280 --> 00:10:00,160 Speaker 3: safe from the long term danger of more INtime and 176 00:10:00,240 --> 00:10:02,960 Speaker 3: things than us just taking over from us, we don't 177 00:10:03,000 --> 00:10:05,000 Speaker 3: even know what to do. What we need to do 178 00:10:05,080 --> 00:10:09,360 Speaker 3: now is understand that there really is this danger. These 179 00:10:09,400 --> 00:10:11,600 Speaker 3: things really do understand and they really are going to 180 00:10:11,600 --> 00:10:14,600 Speaker 3: get smarter than us. But also how are we going 181 00:10:14,679 --> 00:10:17,280 Speaker 3: to keep control of them? We have some chance because 182 00:10:17,400 --> 00:10:18,439 Speaker 3: we're creating. 183 00:10:18,080 --> 00:10:22,600 Speaker 1: Them, and that's where we turn next in the biography 184 00:10:22,600 --> 00:10:25,840 Speaker 1: of AI. How much good can it do? And what 185 00:10:25,880 --> 00:10:28,120 Speaker 1: do we need to do to make sure it becomes 186 00:10:28,120 --> 00:10:31,320 Speaker 1: a responsible citizen instead of a threat to us. 187 00:10:31,320 --> 00:10:35,640 Speaker 3: All many people say, you know, I'll create more jobs 188 00:10:36,320 --> 00:10:39,840 Speaker 3: for this particular thing. I'm not convinced of that. What 189 00:10:39,880 --> 00:10:45,199 Speaker 3: we're doing in the Industrial Revolution we make human strength irrelevant. 190 00:10:46,200 --> 00:10:50,440 Speaker 3: Now we're making human intelligence irrelevant, and that's very scary. 191 00:10:54,000 --> 00:10:57,400 Speaker 6: You're listening to Bloomberg Wall Street Week with David Weston 192 00:10:57,559 --> 00:11:06,240 Speaker 6: from Bloomberg Radio on November fifth, America designs. 193 00:11:05,760 --> 00:11:10,600 Speaker 1: A remarkable seismic shifts in American politics, and Bloomberg is covering. 194 00:11:10,280 --> 00:11:14,000 Speaker 6: Every angle now through election day live from our nation's capital. 195 00:11:14,040 --> 00:11:16,720 Speaker 7: Our polling shows more enthusiasm among black voters. 196 00:11:16,720 --> 00:11:19,480 Speaker 6: Spell and s of Pop with Jim Mathew and Kayley Lyne. 197 00:11:19,720 --> 00:11:22,800 Speaker 8: Where voters see Trump and Harris on the issues. 198 00:11:22,920 --> 00:11:24,720 Speaker 1: How important is higher turnouts? 199 00:11:24,760 --> 00:11:27,840 Speaker 6: Listen live weekdays at noon and five pm Eastern, or 200 00:11:27,880 --> 00:11:31,839 Speaker 6: on demand wherever you get your podcasts. Bloomberg Radio. Context 201 00:11:31,920 --> 00:11:32,880 Speaker 6: changes everything. 202 00:11:33,840 --> 00:11:37,400 Speaker 8: This is the Bloomberg Green Report. A company called form 203 00:11:37,640 --> 00:11:39,960 Speaker 8: Energy has developed what could turn out to be the 204 00:11:40,040 --> 00:11:43,160 Speaker 8: right product at just the right time. It's a utility 205 00:11:43,240 --> 00:11:46,160 Speaker 8: sized battery that can feed the electricity to the power 206 00:11:46,200 --> 00:11:49,760 Speaker 8: grid for one hundred hours straight. That's twenty five times 207 00:11:49,800 --> 00:11:52,680 Speaker 8: longer than most of the grid tide batteries in use today. 208 00:11:53,080 --> 00:11:56,160 Speaker 8: The batteries will start arriving at power plants from Colorado 209 00:11:56,240 --> 00:11:59,079 Speaker 8: to Virginia next year. They're coming at a time when 210 00:11:59,160 --> 00:12:01,640 Speaker 8: electricity did de manned, is being driven up by new 211 00:12:01,679 --> 00:12:05,600 Speaker 8: factories and data centers, and extreme weather events like Hurricanes 212 00:12:05,640 --> 00:12:09,640 Speaker 8: Helene and Milton have triggered massive blackouts. Those are potent 213 00:12:09,720 --> 00:12:13,200 Speaker 8: selling points. Form Energy is based near Boston. It has 214 00:12:13,280 --> 00:12:16,120 Speaker 8: raised one point two billion dollars from investors who are 215 00:12:16,160 --> 00:12:20,000 Speaker 8: betting big on the company's timing and technology. Bloomberg NEF 216 00:12:20,040 --> 00:12:23,079 Speaker 8: says that's especially remarkable in light of a decline in 217 00:12:23,160 --> 00:12:27,400 Speaker 8: overall climate tech financing since last year. Venture capital has 218 00:12:27,440 --> 00:12:32,319 Speaker 8: been strained by high interest rates and macroeconomic headwinds. Jeff Bellinger, 219 00:12:32,480 --> 00:12:33,440 Speaker 8: Bloomberg Radio. 220 00:12:33,800 --> 00:12:35,959 Speaker 9: Out Here in the middle of these acres, it can 221 00:12:35,960 --> 00:12:38,480 Speaker 9: feel like you're the only person on earth. That's how 222 00:12:38,520 --> 00:12:40,800 Speaker 9: it feels when you're struggling with your mental health. But 223 00:12:40,920 --> 00:12:43,800 Speaker 9: you don't have to feel alone. Find more information at 224 00:12:43,800 --> 00:12:46,280 Speaker 9: Love youormind Today dot org, brought to you by the 225 00:12:46,360 --> 00:12:48,360 Speaker 9: Huntsman Mentuealth Institute in the AD Council. 226 00:13:03,720 --> 00:13:07,960 Speaker 6: This is Bloomberg Wall Street Week with David Weston from 227 00:13:08,080 --> 00:13:09,640 Speaker 6: Bloomberg Radio. 228 00:13:12,200 --> 00:13:15,360 Speaker 1: The life of artificial intelligence is connected to the lives 229 00:13:15,400 --> 00:13:19,679 Speaker 1: of people and also to their livelihoods. AI is already 230 00:13:19,720 --> 00:13:22,880 Speaker 1: being used to make processes more efficient. What could this 231 00:13:23,000 --> 00:13:27,480 Speaker 1: mean for productivity and therefore economic growth and who will 232 00:13:27,480 --> 00:13:31,240 Speaker 1: benefit from it? We pose those questions to Jeffrey Hinton. 233 00:13:33,080 --> 00:13:36,400 Speaker 3: It will be a wonderful thing for productivity, that's true. 234 00:13:36,679 --> 00:13:39,360 Speaker 3: Whether it be a wonderful thing for society is something else. Altogether. 235 00:13:40,040 --> 00:13:44,320 Speaker 3: In a decent society, if you increase productivity a lot, 236 00:13:44,520 --> 00:13:48,360 Speaker 3: everybody's better off. But here, what's going to happen if 237 00:13:48,400 --> 00:13:53,160 Speaker 3: you increase productivity a lot. The rich and the big 238 00:13:53,200 --> 00:13:57,400 Speaker 3: companies are going to get much richer, and ordinary people 239 00:13:57,440 --> 00:13:59,400 Speaker 3: are probably going to be worse off because they lose 240 00:13:59,440 --> 00:13:59,960 Speaker 3: their jobs. 241 00:14:01,080 --> 00:14:04,800 Speaker 1: AI's ability to reshape productivity could also give new life 242 00:14:04,840 --> 00:14:08,960 Speaker 1: to developed economies whose productivity has stalled. US productivity growth 243 00:14:08,960 --> 00:14:11,600 Speaker 1: has been significantly lower in the twenty first century than 244 00:14:11,640 --> 00:14:15,200 Speaker 1: it was between nineteen thirty and two thousand. But for 245 00:14:15,320 --> 00:14:18,520 Speaker 1: all the talk of the sweeping effects of AI, economic 246 00:14:18,600 --> 00:14:22,520 Speaker 1: change has been slow to materialize. So far. Goldwyn Sachs 247 00:14:22,560 --> 00:14:25,560 Speaker 1: reports that only five percent of companies claim to use 248 00:14:25,680 --> 00:14:30,000 Speaker 1: generative AI in regular production, with tech and information businesses 249 00:14:30,080 --> 00:14:34,080 Speaker 1: leading the way. When wider adoption does come, Hinton anticipates 250 00:14:34,080 --> 00:14:37,040 Speaker 1: it will have very different impacts on different parts of 251 00:14:37,080 --> 00:14:37,760 Speaker 1: the workforce. 252 00:14:38,320 --> 00:14:41,680 Speaker 3: Many people say, you know, illegrate more jobs for this 253 00:14:41,760 --> 00:14:42,560 Speaker 3: particular thing. 254 00:14:43,120 --> 00:14:44,160 Speaker 2: I'm not convinced of that. 255 00:14:44,680 --> 00:14:48,920 Speaker 3: What we're doing in the Industrial Revolution we make human 256 00:14:49,000 --> 00:14:54,520 Speaker 3: strength irrelevant. Now we're making human intelligence irrelevant, and that's 257 00:14:54,640 --> 00:14:59,480 Speaker 3: very scary. So there's some areas where demand is very elastic. 258 00:15:00,040 --> 00:15:03,360 Speaker 3: An example would be healthcare. If I could get ten 259 00:15:03,400 --> 00:15:06,480 Speaker 3: hours a week talking to my doctor, I'm over seventy 260 00:15:07,720 --> 00:15:11,720 Speaker 3: I'd be very happy. So if you take someone and 261 00:15:11,800 --> 00:15:14,360 Speaker 3: make them much more efficient by having them work with 262 00:15:14,440 --> 00:15:18,640 Speaker 3: a very intelligent AI, they're not gonna become unemployed. It's 263 00:15:18,640 --> 00:15:20,360 Speaker 3: not that you're now only gonna need a few of them. 264 00:15:20,560 --> 00:15:22,240 Speaker 3: You're just gonna get much more healthcare. 265 00:15:22,760 --> 00:15:23,080 Speaker 2: Great. 266 00:15:23,880 --> 00:15:26,560 Speaker 3: So in elastic areas, it's great. The some areas that 267 00:15:26,640 --> 00:15:31,920 Speaker 3: are less elastic, like I have a niece who answers 268 00:15:32,280 --> 00:15:35,280 Speaker 3: letters a complaint to a health service. She used to 269 00:15:35,280 --> 00:15:37,760 Speaker 3: take twenty five minutes to answer a letter. Now she 270 00:15:37,840 --> 00:15:41,760 Speaker 3: can just scan the letter into chat GBT, it'll give 271 00:15:41,760 --> 00:15:43,440 Speaker 3: an answer. She'll look at it, check it is okay, 272 00:15:43,480 --> 00:15:47,840 Speaker 3: that's five minutes now. I suspect they'll need less people 273 00:15:47,920 --> 00:15:50,920 Speaker 3: like that. It may be they can just everybody can 274 00:15:50,920 --> 00:15:53,240 Speaker 3: complain a lot more, but I suspect they'll need less. 275 00:15:53,080 --> 00:15:53,640 Speaker 2: People like that. 276 00:15:53,760 --> 00:15:56,400 Speaker 3: So some jobs are Some jobs are elastic, others aren't. 277 00:15:56,640 --> 00:15:59,640 Speaker 3: The non elastic ones, I think people will lose their jobs. 278 00:16:00,320 --> 00:16:04,680 Speaker 3: And what's going to happen is the extra wealth created 279 00:16:04,720 --> 00:16:07,440 Speaker 3: by the increasing productivity is not going to go to them. 280 00:16:08,080 --> 00:16:11,200 Speaker 1: Hinton's prize in physics wasn't the only Nobel awarded this 281 00:16:11,280 --> 00:16:15,880 Speaker 1: year for artificial intelligence. Two Google DeepMind scientists, a professor 282 00:16:15,920 --> 00:16:19,360 Speaker 1: at the University of Washington receive Nobels in chemistry for 283 00:16:19,520 --> 00:16:22,600 Speaker 1: using AI to predict the structure of millions of proteins 284 00:16:22,840 --> 00:16:26,560 Speaker 1: and even to invent a whole new protein. Marty Chavez 285 00:16:26,600 --> 00:16:30,360 Speaker 1: A sixth Street says that tool called alpha fold shows 286 00:16:30,400 --> 00:16:31,920 Speaker 1: the power of machine learning. 287 00:16:32,400 --> 00:16:37,040 Speaker 5: As important a breakthrough as this Alpha fold work is 288 00:16:37,480 --> 00:16:40,880 Speaker 5: if you think about the problems in biology, the problems 289 00:16:40,920 --> 00:16:44,440 Speaker 5: in biology are so much more complicated. 290 00:16:44,600 --> 00:16:44,840 Speaker 10: Right. 291 00:16:45,040 --> 00:16:48,520 Speaker 5: Proteins are the basic building block of life. But then 292 00:16:48,560 --> 00:16:53,400 Speaker 5: those proteins organize themselves into organ eels of a cell, 293 00:16:53,560 --> 00:16:57,880 Speaker 5: and then cells, which organize themselves into tissues, which organize 294 00:16:57,920 --> 00:17:02,720 Speaker 5: themselves into organs which go to organ systems human beings, 295 00:17:03,120 --> 00:17:04,960 Speaker 5: populations of human beings. 296 00:17:05,440 --> 00:17:09,040 Speaker 1: Beyond biology and health sciences, generative AI could also have 297 00:17:09,160 --> 00:17:12,640 Speaker 1: profound effects on the climate. For one thing, it will 298 00:17:12,640 --> 00:17:16,399 Speaker 1: require a lot more electricity to drive it. Wells Fargo 299 00:17:16,440 --> 00:17:19,159 Speaker 1: projects a five hundred and fifty percent surge in AI 300 00:17:19,280 --> 00:17:22,720 Speaker 1: power demand by twenty twenty six. Ireland is a poster 301 00:17:22,840 --> 00:17:25,919 Speaker 1: child for the extent of that demand. Today, it's growing 302 00:17:26,000 --> 00:17:28,720 Speaker 1: number of data centers used up twenty one percent of 303 00:17:28,760 --> 00:17:31,080 Speaker 1: the country's electricity in twenty twenty three. 304 00:17:31,680 --> 00:17:33,280 Speaker 11: The good news is it's not new in our history. 305 00:17:33,320 --> 00:17:35,600 Speaker 11: We've been through many periods as a country where we 306 00:17:35,680 --> 00:17:39,159 Speaker 11: have had significant energy demand. But what this is going 307 00:17:39,200 --> 00:17:41,080 Speaker 11: to mean is AI is now adding to this as 308 00:17:41,080 --> 00:17:44,080 Speaker 11: we electrify transportation, as we electrify our buildings, we now 309 00:17:44,119 --> 00:17:47,399 Speaker 11: have this big near term demand for energy to power 310 00:17:47,480 --> 00:17:48,320 Speaker 11: these data centers. 311 00:17:48,880 --> 00:17:52,000 Speaker 1: Brian Deese served as the Director of the National Economic 312 00:17:52,040 --> 00:17:55,360 Speaker 1: Council under President Biden and is now focused on studying 313 00:17:55,359 --> 00:17:58,000 Speaker 1: the effects of AI on energy at MIT. 314 00:17:58,640 --> 00:17:58,720 Speaker 12: SO. 315 00:17:58,840 --> 00:18:01,040 Speaker 11: In the near term, it creates this tension of how 316 00:18:01,040 --> 00:18:04,280 Speaker 11: do we bring that additional electricity online? And are we 317 00:18:04,320 --> 00:18:07,600 Speaker 11: bringing cleaner sources online or dirtier sources online, which obviously 318 00:18:07,600 --> 00:18:10,520 Speaker 11: affects emissions. But the longer term, the bigger picture way 319 00:18:10,560 --> 00:18:13,360 Speaker 11: these interact is AI is a technology which will then 320 00:18:13,359 --> 00:18:16,520 Speaker 11: get deployed in lots of aspects of our economy, including 321 00:18:16,560 --> 00:18:19,520 Speaker 11: our energy system. And so there are also opportunities where 322 00:18:19,640 --> 00:18:22,320 Speaker 11: these two technological developments coming at the same time could 323 00:18:22,320 --> 00:18:25,800 Speaker 11: actually be one plus one equals three help us find 324 00:18:25,880 --> 00:18:29,439 Speaker 11: more efficient ways to innovate in the energy space. 325 00:18:30,240 --> 00:18:33,119 Speaker 1: For all the good that generative AI could potentially do 326 00:18:33,240 --> 00:18:36,200 Speaker 1: for the human race, it also poses some real dangers, 327 00:18:36,640 --> 00:18:40,199 Speaker 1: perhaps none so apparent as its application to armed warfare. 328 00:18:40,680 --> 00:18:45,040 Speaker 1: Anya Manual's Aspens Strategy Group publisher report on this very subject. 329 00:18:45,560 --> 00:18:49,119 Speaker 7: AI by its nature is dual use because it is 330 00:18:49,160 --> 00:18:52,440 Speaker 7: a multipurpose technology. It's like electricity or the steam engine. 331 00:18:52,480 --> 00:18:56,639 Speaker 7: It's going to power everything, from the most amazing positive 332 00:18:57,000 --> 00:18:59,320 Speaker 7: use cases to the really dangerous ones. 333 00:19:00,160 --> 00:19:03,760 Speaker 1: All of my lifetime, the biggest geopolitical risk terror even 334 00:19:04,000 --> 00:19:08,520 Speaker 1: has been the threat of thermonuclear destruction. Recently was a 335 00:19:08,520 --> 00:19:11,000 Speaker 1: meeting over in South Korea with an attempt to have 336 00:19:11,040 --> 00:19:14,199 Speaker 1: everybody agreed that if there were to be nuclear weapons used, 337 00:19:14,280 --> 00:19:16,960 Speaker 1: a human had to be in that chain of command, 338 00:19:17,440 --> 00:19:21,199 Speaker 1: and the countries there agreed to it except for one China. 339 00:19:21,400 --> 00:19:22,320 Speaker 1: What do we make of that? 340 00:19:22,440 --> 00:19:26,439 Speaker 7: We are really on day one of this technology and 341 00:19:26,480 --> 00:19:29,640 Speaker 7: how it can be used visa v National security. There 342 00:19:29,640 --> 00:19:32,000 Speaker 7: have been a number of attempts in the last couple 343 00:19:32,040 --> 00:19:34,240 Speaker 7: of years, a lot of them thankfully led by the 344 00:19:34,320 --> 00:19:39,720 Speaker 7: United States to think carefully about AI and nuclear weapons 345 00:19:40,000 --> 00:19:46,080 Speaker 7: use one two. How much you would allow lethal autonomous weapons? 346 00:19:46,119 --> 00:19:48,439 Speaker 7: And this means really, you know, if you have a 347 00:19:48,480 --> 00:19:52,040 Speaker 7: predator drone, there's still a human directing it and deciding 348 00:19:52,080 --> 00:19:54,280 Speaker 7: who to target and who to shoot at. In the 349 00:19:54,440 --> 00:19:57,840 Speaker 7: Ukraine War, You're getting very close now to drone on 350 00:19:57,920 --> 00:20:02,160 Speaker 7: drone dog fights, drones doing their own targeting and shooting, 351 00:20:02,200 --> 00:20:06,200 Speaker 7: and very soon you have a situation that's very dangerous 352 00:20:06,560 --> 00:20:10,080 Speaker 7: where if the AI weapon is always going to be faster, 353 00:20:11,359 --> 00:20:15,879 Speaker 7: the incentive for everyone is to use lethal autonomous weapons 354 00:20:15,920 --> 00:20:18,920 Speaker 7: because they'll always win, and then you'll have huge escalation, 355 00:20:19,440 --> 00:20:22,520 Speaker 7: super dangerous. Lots of different people are trying to get 356 00:20:22,560 --> 00:20:25,200 Speaker 7: at this in different ways. The Korea Conference was one 357 00:20:25,240 --> 00:20:28,280 Speaker 7: of them. I'm not surprised that China hasn't signed on. 358 00:20:28,680 --> 00:20:31,560 Speaker 7: I would just give you one example. You use nuclear weapons. 359 00:20:32,080 --> 00:20:35,160 Speaker 7: You know, when the US came first in the nuclear race, 360 00:20:35,400 --> 00:20:38,439 Speaker 7: we proposed some limited limits on nuclear weapons, I think 361 00:20:38,480 --> 00:20:42,119 Speaker 7: as early as nineteen forty six. Then no real arms 362 00:20:42,119 --> 00:20:46,480 Speaker 7: controlled treaties were negotiated until after you had the Cuban 363 00:20:46,480 --> 00:20:48,520 Speaker 7: missile crisis and we got to the brink, and then 364 00:20:48,680 --> 00:20:51,040 Speaker 7: we had some pretty good ones. But what was going 365 00:20:51,080 --> 00:20:53,320 Speaker 7: on all the time in the background, and what is 366 00:20:53,359 --> 00:20:56,640 Speaker 7: going on now, and it's super important and bears emphasizing, 367 00:20:57,200 --> 00:20:59,960 Speaker 7: is quietly behind the scenes. There is some Chinese sign 368 00:21:00,000 --> 00:21:04,200 Speaker 7: scientists and American and Western scientists talking about these issues. 369 00:21:04,600 --> 00:21:06,560 Speaker 7: There are other track twos that are more on the 370 00:21:06,600 --> 00:21:10,920 Speaker 7: policy area. I'm part of one of them, quiet conversations 371 00:21:10,920 --> 00:21:13,760 Speaker 7: with the Chinese about the dangers of AI and the 372 00:21:13,840 --> 00:21:18,200 Speaker 7: Chinese interlock unders I've talked to are equally worried about 373 00:21:18,400 --> 00:21:20,800 Speaker 7: very similar things than the ones you and I are 374 00:21:20,840 --> 00:21:21,760 Speaker 7: talking about here. 375 00:21:22,200 --> 00:21:24,920 Speaker 3: And for a long time people have realized there's going 376 00:21:24,920 --> 00:21:26,480 Speaker 3: to be an arms race who can get the best 377 00:21:26,520 --> 00:21:30,119 Speaker 3: lethal autonomous weapons fastest. All of the defense departments of 378 00:21:30,160 --> 00:21:35,280 Speaker 3: the people who sell arms, like the US, China, Britain, Israel, Russia, 379 00:21:35,400 --> 00:21:39,320 Speaker 3: all those defense departments are busy working on autonomous lethal weapons. 380 00:21:40,520 --> 00:21:42,679 Speaker 3: They're not going to stop. If one of them stopped, 381 00:21:42,680 --> 00:21:46,439 Speaker 3: the others wouldn't. What we need for that is not 382 00:21:46,520 --> 00:21:50,200 Speaker 3: to stop working on it, but to have a Geneva convention. Now, 383 00:21:50,480 --> 00:21:53,400 Speaker 3: you don't get Geneva conventions things like the Geneva Conventions 384 00:21:53,400 --> 00:21:56,320 Speaker 3: for chemical weapons until after something very nasty has happened. 385 00:21:56,880 --> 00:21:58,919 Speaker 2: So I think realistically, very. 386 00:21:58,840 --> 00:22:01,160 Speaker 3: Nasty things are going to happen with ethan autonomous weapons, 387 00:22:01,320 --> 00:22:03,840 Speaker 3: and then maybe we'll be able to get conventions. So 388 00:22:03,920 --> 00:22:07,440 Speaker 3: for chemical weapons, the conventions have basically worked. Putin isn't 389 00:22:07,520 --> 00:22:10,240 Speaker 3: using them in the Ukraine and they haven't been used much, 390 00:22:12,119 --> 00:22:15,920 Speaker 3: so basically the conventions worked. I'm hoping they would work 391 00:22:15,920 --> 00:22:19,040 Speaker 3: for lethal autonomous weapons, although I'm less confident but nothing's 392 00:22:19,080 --> 00:22:21,639 Speaker 3: going to happen till after some very nasty things have happened. 393 00:22:22,359 --> 00:22:25,520 Speaker 1: Whether AI works for good or for ill, a fundamental 394 00:22:25,600 --> 00:22:28,440 Speaker 1: question is whether humans can maintain control or at least 395 00:22:28,480 --> 00:22:32,280 Speaker 1: influence over it. Professor Hinton says we shouldn't assume that 396 00:22:32,320 --> 00:22:34,359 Speaker 1: we can, at least for long. 397 00:22:34,840 --> 00:22:36,800 Speaker 3: As soon as you make agents which people are busy 398 00:22:36,800 --> 00:22:39,399 Speaker 3: doing now, things that can act in the world. To 399 00:22:39,440 --> 00:22:41,919 Speaker 3: make an effective agent, you have to give it the 400 00:22:41,960 --> 00:22:44,239 Speaker 3: ability to create subgoals. So if you want to get 401 00:22:44,280 --> 00:22:45,680 Speaker 3: to Europe, you have a sub goal of getting to 402 00:22:45,680 --> 00:22:48,240 Speaker 3: an airport, and you don't have to sort of think 403 00:22:48,280 --> 00:22:51,640 Speaker 3: about Europe while you're solving that subgoal. That's why subgoals 404 00:22:51,640 --> 00:22:56,240 Speaker 3: are helpful, and these big eye systems create sub goals. Now, 405 00:22:56,320 --> 00:22:58,280 Speaker 3: the problem with that is if you give something the 406 00:22:58,280 --> 00:23:01,959 Speaker 3: ability to create subgoals, it will quickly realize there's one 407 00:23:02,000 --> 00:23:06,800 Speaker 3: particular subcoal that's almost always useful. So if I have 408 00:23:06,840 --> 00:23:09,560 Speaker 3: the goal of just getting more control over the world, 409 00:23:09,880 --> 00:23:11,840 Speaker 3: that will help me achieve everything I want to achieve. 410 00:23:12,920 --> 00:23:15,199 Speaker 3: So these things will realize that very quickly, even if 411 00:23:15,200 --> 00:23:17,760 Speaker 3: they've got no self interest. Then understand that if I 412 00:23:17,800 --> 00:23:20,080 Speaker 3: get more control, I'm it'll be better at doing what 413 00:23:20,280 --> 00:23:22,240 Speaker 3: they want me to do, and so they will try 414 00:23:22,280 --> 00:23:25,359 Speaker 3: and get more control. That's the beginning of a very 415 00:23:25,400 --> 00:23:26,240 Speaker 3: slippery slope. 416 00:23:26,720 --> 00:23:29,800 Speaker 1: But that suggests times are wasting. Then, in fact, we 417 00:23:29,840 --> 00:23:32,359 Speaker 1: don't have that much time. We don't to be able 418 00:23:32,359 --> 00:23:34,600 Speaker 1: to get some control over general they are. 419 00:23:35,240 --> 00:23:37,600 Speaker 3: We control it right now, but we don't have that 420 00:23:37,720 --> 00:23:40,240 Speaker 3: much time to figure out how we're going to stay 421 00:23:40,240 --> 00:23:41,640 Speaker 3: in control when it's smarter than us. 422 00:23:42,040 --> 00:23:43,960 Speaker 1: How do we go about that. Let's take the United 423 00:23:44,000 --> 00:23:46,760 Speaker 1: States for example, before we go globally. In the United States, 424 00:23:47,160 --> 00:23:48,960 Speaker 1: we have a government. There are good people in the government, 425 00:23:49,040 --> 00:23:52,600 Speaker 1: smart people, some probably less smart, But are they up 426 00:23:52,640 --> 00:23:55,639 Speaker 1: to the job of really understanding what you're talking about 427 00:23:55,920 --> 00:23:57,240 Speaker 1: and getting their arms around. 428 00:23:57,520 --> 00:24:00,360 Speaker 3: I think with the current government they are taking it seriously. 429 00:24:01,200 --> 00:24:04,760 Speaker 3: It's just very difficult to know what to do. One 430 00:24:04,920 --> 00:24:07,959 Speaker 3: clear thing they should be doing that I believe they 431 00:24:07,960 --> 00:24:12,000 Speaker 3: should be doing. We need many of the smartest young 432 00:24:12,040 --> 00:24:14,880 Speaker 3: researchers to be working on this problem, and we need 433 00:24:14,880 --> 00:24:17,800 Speaker 3: them to have resources. Now, the government doesn't have the resources. 434 00:24:17,840 --> 00:24:21,080 Speaker 3: The big companies have the resources. The government I think 435 00:24:21,800 --> 00:24:26,400 Speaker 3: should be insisting that the big companies spend much more 436 00:24:26,400 --> 00:24:29,720 Speaker 3: of their resources on safety, on this safety research or 437 00:24:29,760 --> 00:24:32,400 Speaker 3: how will we stay in control? Compared with what they 438 00:24:32,400 --> 00:24:34,399 Speaker 3: do now right now, they spend like a few percent 439 00:24:34,600 --> 00:24:38,400 Speaker 3: on that, and nearly all their resources go into building 440 00:24:38,800 --> 00:24:40,120 Speaker 3: even better, bigger models. 441 00:24:40,400 --> 00:24:41,960 Speaker 1: As you say, big companies don't like to be told 442 00:24:41,960 --> 00:24:44,160 Speaker 1: by the government how to spend their money, but they 443 00:24:44,160 --> 00:24:47,240 Speaker 1: are often. I mean, you have big accounting departments, for example, 444 00:24:47,280 --> 00:24:51,000 Speaker 1: to comply with various regulatory requirements and accounting from everything 445 00:24:51,040 --> 00:24:54,879 Speaker 1: you know. Everything you understand is general AI potentially an 446 00:24:54,920 --> 00:24:57,000 Speaker 1: existential threat to the human species. 447 00:24:57,119 --> 00:24:59,320 Speaker 3: I think that's what I've been saying. Yes, it really 448 00:24:59,359 --> 00:25:01,879 Speaker 3: is an exercenti threat. Some people say this is just 449 00:25:01,920 --> 00:25:04,800 Speaker 3: science fiction, and until fairly recently I believed it was 450 00:25:04,840 --> 00:25:06,960 Speaker 3: a long way off. I always thought it would be 451 00:25:06,960 --> 00:25:08,960 Speaker 3: a very long term threat, but I thought it would 452 00:25:08,960 --> 00:25:10,080 Speaker 3: be one hundred years before we had. 453 00:25:09,960 --> 00:25:12,200 Speaker 2: Really smart things. Fifty years. 454 00:25:12,240 --> 00:25:14,720 Speaker 3: Maybe we have plenty of time to think about it now. 455 00:25:14,760 --> 00:25:17,480 Speaker 3: I think it's quite likely that sometime in the next 456 00:25:17,560 --> 00:25:20,359 Speaker 3: twenty years these things will get smarter than us, and 457 00:25:20,400 --> 00:25:22,320 Speaker 3: we really need to worry about what happens. 458 00:25:22,359 --> 00:25:28,000 Speaker 1: Then, coming up a race for resources south of the border, 459 00:25:28,119 --> 00:25:30,560 Speaker 1: we go to Mexico, to see how global trade is 460 00:25:30,600 --> 00:25:34,200 Speaker 1: evolving in real time. That's next on Wall Street Week. 461 00:25:36,200 --> 00:25:39,640 Speaker 6: You're listening to Bloomberg Wall Straight Week with David Weston 462 00:25:39,760 --> 00:25:43,520 Speaker 6: from Bloomberg Radio. News When you want it, get the 463 00:25:43,600 --> 00:25:47,760 Speaker 6: latest headlines from Bloomberg News, updated continuously throughout the day 464 00:25:47,800 --> 00:25:51,399 Speaker 6: with Bloomberg News Now. Listen any time on Apple Card, playing, 465 00:25:51,440 --> 00:25:54,760 Speaker 6: Android Auto with the Bloomberg Business Aft, and anywhere you 466 00:25:54,840 --> 00:26:14,080 Speaker 6: get your podcasts. This is Bloomberg Wall Street Week with 467 00:26:14,240 --> 00:26:16,639 Speaker 6: David Weston from Bloomberg Radio. 468 00:26:19,160 --> 00:26:22,560 Speaker 1: This is a story about potential, the potential that further 469 00:26:22,640 --> 00:26:26,080 Speaker 1: economic integration between the United States and Mexico holds for 470 00:26:26,200 --> 00:26:30,160 Speaker 1: both countries, a potential for good, but also a potential 471 00:26:30,200 --> 00:26:31,800 Speaker 1: for unintended consequences. 472 00:26:32,280 --> 00:26:34,120 Speaker 13: The plastic comes from hyping. 473 00:26:34,520 --> 00:26:38,119 Speaker 1: Baldwin Britain is a manufacturer based in Monterey, trying to 474 00:26:38,200 --> 00:26:41,159 Speaker 1: seize what some are calling the Mexican moment. 475 00:26:41,640 --> 00:26:45,000 Speaker 13: And everything comes us as pellets and it gets melted 476 00:26:45,320 --> 00:26:46,440 Speaker 13: in this machine. 477 00:26:46,640 --> 00:26:50,159 Speaker 1: As CEO of Plastic Exports, he runs several factories in 478 00:26:50,200 --> 00:26:54,639 Speaker 1: northern Mexico making plastic components and finished consumer products in 479 00:26:54,680 --> 00:26:56,840 Speaker 1: his newest plant kitchen containers. 480 00:26:57,119 --> 00:26:59,919 Speaker 13: So this is for consumer products that we're selling to 481 00:26:59,920 --> 00:27:03,840 Speaker 13: the United States that basically came from near sharing projects. 482 00:27:03,960 --> 00:27:07,160 Speaker 13: So these products were previously made in China and they're 483 00:27:07,200 --> 00:27:08,680 Speaker 13: relocating into Mexico. 484 00:27:09,520 --> 00:27:12,800 Speaker 1: What was once offshoring to China and other exporters in 485 00:27:12,880 --> 00:27:16,040 Speaker 1: Ocean a Way is beginning to give way to nearshoring 486 00:27:16,280 --> 00:27:20,200 Speaker 1: from Mexico. The share of US imports coming from China, 487 00:27:20,280 --> 00:27:23,439 Speaker 1: which hit twenty one percent in twenty eighteen, is on 488 00:27:23,480 --> 00:27:27,480 Speaker 1: the decline, down to under fourteen percent last year. The 489 00:27:27,560 --> 00:27:30,880 Speaker 1: share coming from Mexico, over thirteen percent in twenty eighteen, 490 00:27:31,440 --> 00:27:38,280 Speaker 1: is on the rise, topping fifteen percent last year. This 491 00:27:38,520 --> 00:27:41,640 Speaker 1: closer integration between Mexico and the United States was jump 492 00:27:41,680 --> 00:27:45,000 Speaker 1: started by the North American Free Trade Agreement or NAFTA, 493 00:27:45,400 --> 00:27:48,679 Speaker 1: back in nineteen ninety four, covering Canada as well as 494 00:27:48,680 --> 00:27:51,880 Speaker 1: the US and Mexico. At that time, a young American 495 00:27:51,960 --> 00:27:55,360 Speaker 1: named David Eaton was an advisor on the treaty. The 496 00:27:55,400 --> 00:27:58,920 Speaker 1: promise of the agreement and the opportunities it created took 497 00:27:59,000 --> 00:28:02,760 Speaker 1: him to Monterey and he has never left. Now he's 498 00:28:02,800 --> 00:28:06,080 Speaker 1: the Mexico Director for Business development of the recently merged 499 00:28:06,119 --> 00:28:10,000 Speaker 1: Canadian Pacific Kansas City Rail forming what he calls the 500 00:28:10,119 --> 00:28:13,800 Speaker 1: near shoring backbone. His goal is to fuse together a 501 00:28:13,840 --> 00:28:17,200 Speaker 1: North American rail system reaching from Upper Canada through the 502 00:28:17,320 --> 00:28:21,199 Speaker 1: United States all the way to southern Mexico. How fast 503 00:28:21,400 --> 00:28:22,439 Speaker 1: is that need growing? 504 00:28:22,960 --> 00:28:23,360 Speaker 10: Ercardo. 505 00:28:23,760 --> 00:28:28,080 Speaker 12: We're investing and getting ready in the planning process right now. 506 00:28:28,440 --> 00:28:33,160 Speaker 12: Because you look at lots of the announcements out there, BMW, Volvo, 507 00:28:33,359 --> 00:28:38,920 Speaker 12: mattel Lego, Arsolo, Midal Turnium, the pipeline is really long. 508 00:28:39,760 --> 00:28:42,560 Speaker 1: Britain is willing to bet you'll find something he makes 509 00:28:42,640 --> 00:28:45,640 Speaker 1: in your home. As a longtime supplier of parts to 510 00:28:45,880 --> 00:28:49,920 Speaker 1: US and European home appliance companies, his new customers are 511 00:28:50,000 --> 00:28:53,240 Speaker 1: just as likely to be Chinese multinationals you've never heard of. 512 00:28:53,480 --> 00:28:56,480 Speaker 13: Thirty years ago we saw a lot of shifts of 513 00:28:56,520 --> 00:29:01,640 Speaker 13: business from Mexico to China. Four years we have seen 514 00:29:01,680 --> 00:29:03,680 Speaker 13: the shift come back to Mexico. 515 00:29:03,800 --> 00:29:06,320 Speaker 1: Are Chinese companies coming in to compete with you? 516 00:29:07,160 --> 00:29:10,120 Speaker 13: Yes, they are coming. They're setting up plants in Mexico 517 00:29:10,240 --> 00:29:12,840 Speaker 13: as a matter of fact, and Leveloon, the state of Levelone, 518 00:29:13,440 --> 00:29:16,080 Speaker 13: which is an industrial helb of Mexico. They're setting up 519 00:29:16,160 --> 00:29:19,440 Speaker 13: so that basically forces companies like us to be in 520 00:29:19,480 --> 00:29:22,040 Speaker 13: our a game to compete with them. 521 00:29:22,960 --> 00:29:26,800 Speaker 1: Today, industrial parks full of Chinese companies are sprouting up 522 00:29:26,840 --> 00:29:28,240 Speaker 1: around the Monterey Hub. 523 00:29:28,640 --> 00:29:32,160 Speaker 13: There's a city in China called sheen Sen. Thirty five 524 00:29:32,240 --> 00:29:35,240 Speaker 13: years ago there was a village to thirty thousand people. 525 00:29:35,320 --> 00:29:38,920 Speaker 13: Today they have over twenty million. There's a manyfacturing hub. 526 00:29:39,320 --> 00:29:41,080 Speaker 13: We see that happening to Mexico. 527 00:29:42,360 --> 00:29:47,960 Speaker 10: There is a philosophy among these companies coming from China 528 00:29:50,760 --> 00:29:57,400 Speaker 10: to go abroad, go to these other countries, learn how 529 00:29:57,400 --> 00:30:00,400 Speaker 10: to do this, integrate to move. 530 00:30:00,960 --> 00:30:04,560 Speaker 1: For decades, Alan Russell, CEO of TECHMA, has been in 531 00:30:04,600 --> 00:30:07,760 Speaker 1: the business of helping companies to open and operate factories 532 00:30:07,760 --> 00:30:11,840 Speaker 1: in Mexico, overcoming some of the friction and red tape. 533 00:30:12,080 --> 00:30:16,600 Speaker 10: Our technology is Mexico and we help companies skip the 534 00:30:16,640 --> 00:30:19,600 Speaker 10: pain and the time that it takes to learn that 535 00:30:19,680 --> 00:30:20,160 Speaker 10: on their own. 536 00:30:20,400 --> 00:30:24,520 Speaker 1: Over your thirty eight years, has there been some harmonization 537 00:30:25,040 --> 00:30:28,080 Speaker 1: of some of those differences in regulations and rules, tax flaws, 538 00:30:28,120 --> 00:30:28,680 Speaker 1: things like that. 539 00:30:28,840 --> 00:30:35,040 Speaker 10: They've gotten more strict, more rules and regulations. It's more 540 00:30:35,080 --> 00:30:37,640 Speaker 10: complex today than it ever has been. So we put 541 00:30:37,880 --> 00:30:41,720 Speaker 10: an umbrella over our client bring them into the country. 542 00:30:41,720 --> 00:30:45,000 Speaker 1: In mexicoolt Annie Ching is the plant manager at Leos, 543 00:30:45,520 --> 00:30:48,560 Speaker 1: one of the companies using Techma for its new operations 544 00:30:48,560 --> 00:30:51,480 Speaker 1: in Saltio, an hour and a half outside of Monterey. 545 00:30:52,000 --> 00:30:55,960 Speaker 1: Leos makes batteries for data centers, cars and golf carts. 546 00:30:56,560 --> 00:30:59,320 Speaker 1: It was established in China twenty five years ago and 547 00:30:59,360 --> 00:31:03,080 Speaker 1: still manuf factors there, but in twenty eleven Leoch relocated 548 00:31:03,120 --> 00:31:06,280 Speaker 1: its headquarters to Singapore, and just last year it opened 549 00:31:06,320 --> 00:31:09,720 Speaker 1: a new factory in Mexico with plans to open two more. 550 00:31:10,200 --> 00:31:14,280 Speaker 4: For the export process, we started the export of the 551 00:31:14,320 --> 00:31:17,680 Speaker 4: first containers to the United States last week. 552 00:31:17,840 --> 00:31:20,680 Speaker 1: Russell has seen an uptick in his business as companies 553 00:31:20,720 --> 00:31:24,000 Speaker 1: begin to shift their supply chains from China to Mexico, 554 00:31:24,640 --> 00:31:27,000 Speaker 1: at first because of the cost of fuel and the 555 00:31:27,040 --> 00:31:30,920 Speaker 1: delays and transportation, and then when the near shoring process 556 00:31:31,000 --> 00:31:35,320 Speaker 1: was turbocharged by the pandemic and tariffs, as companies sought 557 00:31:35,360 --> 00:31:36,680 Speaker 1: alternatives to China. 558 00:31:37,280 --> 00:31:41,280 Speaker 10: So there started to be a movement where the boardrooms 559 00:31:41,280 --> 00:31:45,200 Speaker 10: across the United States were saying, we should consider diversifying 560 00:31:45,240 --> 00:31:48,800 Speaker 10: from China. We have all of our eggs there. Coster 561 00:31:48,920 --> 00:31:51,160 Speaker 10: going up where a lot of these public companies actually 562 00:31:51,200 --> 00:31:56,160 Speaker 10: carry on their books a China risk factor until they 563 00:31:56,160 --> 00:32:00,760 Speaker 10: can adjust and minimize that risk by diverse ye into 564 00:32:00,800 --> 00:32:03,760 Speaker 10: Vietnam and the laos into India into Mexico. 565 00:32:04,440 --> 00:32:07,400 Speaker 1: We visited a Mattel plant in Monterey, a facility that 566 00:32:07,480 --> 00:32:10,120 Speaker 1: was close to closing at one point when production moved 567 00:32:10,120 --> 00:32:13,480 Speaker 1: to China, but since the pandemic, Mattel has taken a 568 00:32:13,600 --> 00:32:17,640 Speaker 1: u turn moving production closer to home. The Monterey plant 569 00:32:17,680 --> 00:32:20,640 Speaker 1: is now its largest in the world. The place where 570 00:32:20,640 --> 00:32:24,360 Speaker 1: they make, among other things, the Barbie Dreamhouse. They're all 571 00:32:24,400 --> 00:32:29,120 Speaker 1: destined for Christmas trees. The raw material for the Barbie 572 00:32:29,200 --> 00:32:32,680 Speaker 1: Dreamhouse comes from the United States by rail straight to 573 00:32:32,720 --> 00:32:33,720 Speaker 1: the factory doorstep. 574 00:32:33,840 --> 00:32:36,320 Speaker 12: Of course, toys are made of resin, and we've got 575 00:32:36,720 --> 00:32:38,880 Speaker 12: resin cars that are coming down with those cars are 576 00:32:38,880 --> 00:32:42,800 Speaker 12: pull Yeah Nindra sons of resin produced from the US 577 00:32:42,840 --> 00:32:44,080 Speaker 12: golf coast, primarily. 578 00:32:44,480 --> 00:32:47,520 Speaker 1: Although some have been quick to declare Mexico the victor 579 00:32:47,560 --> 00:32:50,360 Speaker 1: of the US trade war with China, Eaton says the 580 00:32:50,480 --> 00:32:53,959 Speaker 1: United States also benefits from deeper integration with its neighbor. 581 00:32:54,440 --> 00:32:57,000 Speaker 1: He points to the auto sector under the United States 582 00:32:57,120 --> 00:33:01,480 Speaker 1: Mexico Canada Agreement the USMCA as a future blueprint for 583 00:33:01,600 --> 00:33:05,760 Speaker 1: trade that will help US, Canada, and Mexico thrive together. 584 00:33:06,200 --> 00:33:09,400 Speaker 12: You really can't talk about a US made car or 585 00:33:09,440 --> 00:33:12,600 Speaker 12: a Canadian made car a Mexican car. These autos are 586 00:33:12,640 --> 00:33:15,360 Speaker 12: really North American. You know, the cars and the parts 587 00:33:15,400 --> 00:33:19,440 Speaker 12: and components across the three borders thousands and times. Literally 588 00:33:19,640 --> 00:33:22,120 Speaker 12: before you produce a before you produce a car. 589 00:33:22,320 --> 00:33:27,000 Speaker 1: Auto sector is really big here in Northern Mexico. Give 590 00:33:27,040 --> 00:33:29,080 Speaker 1: us a sense of where that is and where that's going. 591 00:33:29,160 --> 00:33:32,880 Speaker 12: The USMCA says that seventy or seventy five percent of 592 00:33:32,960 --> 00:33:36,800 Speaker 12: the value or content of a car must meet with 593 00:33:37,000 --> 00:33:40,760 Speaker 12: North American content requirements to get duty free treatment under 594 00:33:40,800 --> 00:33:44,880 Speaker 12: the USMCA, and that's creating incentives for a lot of 595 00:33:44,920 --> 00:33:48,680 Speaker 12: companies to change their sourcing. For example, the steel company 596 00:33:48,800 --> 00:33:52,720 Speaker 12: Urnium is now investing billions of dollars here in Northern 597 00:33:52,760 --> 00:33:58,320 Speaker 12: Mexico to process their own slab on site. They're going 598 00:33:58,440 --> 00:34:02,640 Speaker 12: towards the production of an auto coil that meets the 599 00:34:02,760 --> 00:34:07,320 Speaker 12: USMCA standards, and so instead of purchasing raw slab from Brazil, 600 00:34:07,760 --> 00:34:10,840 Speaker 12: they're going to start processing that slab here on site, 601 00:34:11,080 --> 00:34:14,640 Speaker 12: consuming iron ore and other products from the United States. 602 00:34:14,840 --> 00:34:19,799 Speaker 12: We're stronger together, frankly, when we manufacture in Mexico. It 603 00:34:19,920 --> 00:34:22,960 Speaker 12: creates more synergies and more jobs in the US. 604 00:34:23,239 --> 00:34:26,000 Speaker 1: The near shoring money coming into Mexico isn't just about 605 00:34:26,040 --> 00:34:29,920 Speaker 1: ways of transporting raw materials and finished goods. It also 606 00:34:29,960 --> 00:34:33,200 Speaker 1: requires real estate investment, which we saw in dozens of 607 00:34:33,320 --> 00:34:36,600 Speaker 1: large industrial parks through the area. But if you look 608 00:34:36,600 --> 00:34:38,880 Speaker 1: at this area, this industrial part right now, if you 609 00:34:38,920 --> 00:34:41,560 Speaker 1: came here twenty years ago, what would you see. Well? 610 00:34:42,760 --> 00:34:43,080 Speaker 2: Nothing. 611 00:34:43,440 --> 00:34:48,120 Speaker 1: Alberta Cretine spent a decade accumulating industrial real estate, some 612 00:34:48,200 --> 00:34:52,040 Speaker 1: forty two million square feet of it. Six bidders competed 613 00:34:52,080 --> 00:34:56,280 Speaker 1: to buy the company he led, Febra Tarafina, a competition 614 00:34:56,400 --> 00:34:59,480 Speaker 1: ultimately won by Prologis when it paid two point eight 615 00:34:59,520 --> 00:35:03,080 Speaker 1: billion dollar to top Blackstone's bid, doubling the size of 616 00:35:03,080 --> 00:35:06,239 Speaker 1: its portfolio in Mexico. Tell us about that bidding war. 617 00:35:06,480 --> 00:35:09,799 Speaker 14: Tearina was a public company, so there were some unsolicit 618 00:35:09,880 --> 00:35:12,960 Speaker 14: offers or a takeover office, you know, from one company, 619 00:35:13,000 --> 00:35:15,160 Speaker 14: and then an OI wanted, I wanted, and I allow 620 00:35:15,239 --> 00:35:17,200 Speaker 14: my investors to make the issue which way they wanted 621 00:35:17,239 --> 00:35:17,440 Speaker 14: to go? 622 00:35:17,880 --> 00:35:20,719 Speaker 1: Is a commercial estate for manufacturing in northern Mexico is 623 00:35:20,719 --> 00:35:23,040 Speaker 1: still a good investment if you were starting out today. 624 00:35:23,440 --> 00:35:24,360 Speaker 1: Would you invest in it? 625 00:35:24,560 --> 00:35:25,200 Speaker 14: Absolutely? 626 00:35:25,640 --> 00:35:30,320 Speaker 1: Despite his continued enthusiasm for the business, Cretine acknowledges some limits, 627 00:35:30,680 --> 00:35:33,600 Speaker 1: particularly when it comes to a reliable source of energy 628 00:35:33,719 --> 00:35:37,440 Speaker 1: to power the plants. Addressing the infrastructure needs with restrict 629 00:35:37,480 --> 00:35:39,880 Speaker 1: to electricity you talk about requires a fair amount of 630 00:35:39,880 --> 00:35:44,400 Speaker 1: capital investment. Yes, is there enough public investment to be 631 00:35:44,440 --> 00:35:45,960 Speaker 1: able to take care of that or do you need 632 00:35:46,239 --> 00:35:47,520 Speaker 1: private capital to come in? 633 00:35:47,680 --> 00:35:50,799 Speaker 14: Absolutely, we need a private capital to come in. The 634 00:35:50,840 --> 00:35:55,560 Speaker 14: administration acknowledges that the last administration almost have no investment 635 00:35:55,560 --> 00:35:59,600 Speaker 14: whatsoever in generation or in high tension lines, that there 636 00:35:59,600 --> 00:36:03,080 Speaker 14: had been interdeability of to continue to provide environment for 637 00:36:03,160 --> 00:36:04,880 Speaker 14: new companies to come to Mexico. 638 00:36:05,320 --> 00:36:08,640 Speaker 4: Part of Mexico's challenges are self made. Is its own 639 00:36:09,760 --> 00:36:12,080 Speaker 4: barriers that's putting in front. So it's things like not 640 00:36:12,200 --> 00:36:15,560 Speaker 4: having access to affordable, consistent, clean energy. 641 00:36:15,800 --> 00:36:19,080 Speaker 1: Shannon O'Neil is Senior Vice President of the Council on 642 00:36:19,160 --> 00:36:22,239 Speaker 1: Foreign Relations. She sees the window of opportunity others do, 643 00:36:22,640 --> 00:36:26,439 Speaker 1: but also some troublesome risks for investors. As you've talked 644 00:36:26,440 --> 00:36:30,200 Speaker 1: to investors in Mexico, how concerned are they about some 645 00:36:30,200 --> 00:36:33,200 Speaker 1: of the issues you identified like judicial reform, for example. 646 00:36:32,920 --> 00:36:35,440 Speaker 4: These really are front and center and investors' minds. So 647 00:36:35,520 --> 00:36:38,840 Speaker 4: the justice reform will mean that judges are now elected, 648 00:36:38,880 --> 00:36:41,840 Speaker 4: and so businesses worry that those judges could be bought 649 00:36:42,719 --> 00:36:45,360 Speaker 4: or influenced by the government in particular cases. 650 00:36:45,480 --> 00:36:47,799 Speaker 1: What about security in particular, because it wasn't that long 651 00:36:47,880 --> 00:36:50,160 Speaker 1: ago that it felt like some of the individual Mexican 652 00:36:50,160 --> 00:36:53,200 Speaker 1: states were almost in a state of civil war with 653 00:36:53,320 --> 00:36:53,960 Speaker 1: the cartels. 654 00:36:54,280 --> 00:36:58,160 Speaker 4: Security has worsened under this last government. Homicide rates have 655 00:36:58,239 --> 00:37:02,399 Speaker 4: remained at near record highs, but things like extortion, embezzlement, 656 00:37:02,600 --> 00:37:04,960 Speaker 4: and other kidnappings have grown. 657 00:37:05,200 --> 00:37:07,040 Speaker 1: One of the things an investor tends not to like 658 00:37:07,120 --> 00:37:08,439 Speaker 1: is uncertainty. 659 00:37:08,000 --> 00:37:11,200 Speaker 4: And it's building an uncertainty that happened under the last administration. 660 00:37:11,520 --> 00:37:14,319 Speaker 4: It's continuing and even deepening, and we've seen it in 661 00:37:14,360 --> 00:37:17,439 Speaker 4: the investment numbers. You know what, Mexicans were hoping twenty 662 00:37:17,480 --> 00:37:19,960 Speaker 4: twenty four would be a record highs in terms of investment, 663 00:37:20,239 --> 00:37:22,960 Speaker 4: and most of it has been frozen, both international investment, 664 00:37:23,000 --> 00:37:26,400 Speaker 4: foreign investment, but also domestic investments. So this is a 665 00:37:26,440 --> 00:37:31,400 Speaker 4: time of uncertainty in Mexico, even as companies are looking 666 00:37:31,400 --> 00:37:34,239 Speaker 4: for alternatives to Asia. 667 00:37:33,719 --> 00:37:37,600 Speaker 1: And if that weren't enough the USMCA itself, that successor 668 00:37:37,640 --> 00:37:41,480 Speaker 1: agreement to NAFTA underpinning the near showing system, will be 669 00:37:41,560 --> 00:37:45,120 Speaker 1: up for review in twenty twenty six, with new governments 670 00:37:45,160 --> 00:37:47,400 Speaker 1: in both the United States and Mexico. 671 00:37:47,600 --> 00:37:50,160 Speaker 4: I do think these are going to be pretty difficult negotiations. 672 00:37:50,280 --> 00:37:53,839 Speaker 4: The really looming issue for the review of the USMCA 673 00:37:54,040 --> 00:37:57,879 Speaker 4: is China and where does China fit into supply chains 674 00:37:57,880 --> 00:38:00,120 Speaker 4: in North America And where does China fit in to 675 00:38:00,200 --> 00:38:06,239 Speaker 4: production more broadly in North America. 676 00:38:13,040 --> 00:38:15,359 Speaker 1: That does it for this edition of Bloomberg Wall Street Week. 677 00:38:15,400 --> 00:38:17,719 Speaker 1: If you missed any part of today's program, you can 678 00:38:17,800 --> 00:38:21,160 Speaker 1: listen on demand with our Wall Street Week podcast. Find 679 00:38:21,160 --> 00:38:24,799 Speaker 1: that on Apple, Spotify, or anywhere else you get your podcasts. 680 00:38:25,080 --> 00:38:28,080 Speaker 1: I'm David Weston. Stay with us. Today's top stories and 681 00:38:28,239 --> 00:38:32,160 Speaker 1: global business headlines are coming up right now.