WEBVTT 1 00:00:01.837 --> 00:00:03.979 Hi, this is Technically Creative's Week in Review. 2 00:00:04.399 --> 00:00:08.302 This week, two of the biggest advertising companies on Earth had a fight on Instagram. 3 00:00:08.803 --> 00:00:12.486 A company announced it can put a billboard in your favorite show after the show is wrapped. 4 00:00:13.006 --> 00:00:17.550 Fox spent $22 billion on the thing you used to use to change the channel away from Fox. 5 00:00:17.710 --> 00:00:23.656 Elon Musk spent nearly half a billion dollars on an election, lost, and then joked that his space lasers must be broken. 6 00:00:23.676 --> 00:00:24.196 Yeah. 7 00:00:24.296 --> 00:00:29.197 And a professor got arrested for impeding a delivery robot because apparently the robot has rights now? 8 00:00:29.977 --> 00:00:33.358 Somewhere, a screenwriter is being told that all of this is far too fetched. 9 00:00:33.498 --> 00:00:36.499 Welcome to our Week in Review, where we'll cover all that in the next 20 minutes. 10 00:00:36.679 --> 00:00:41.540 Technically Creative is brought to you by Kubrick Labs, the AI and IT transformation partner for creative companies. 11 00:00:41.640 --> 00:00:42.820 So let's start with the news. 12 00:00:42.940 --> 00:00:47.962 Publicists and WPP are acting like the real housewives of New Jersey, and they're just fighting on Instagram. 13 00:00:47.982 --> 00:00:48.962 So that's amazing. 14 00:00:49.842 --> 00:00:53.546 WPP and Publicis are two of the biggest advertising holding companies in the world. 15 00:00:53.946 --> 00:00:58.450 The giant parent companies that own hundreds of agencies, some you've heard of, plenty you haven't. 16 00:00:59.110 --> 00:01:07.318 On WPP's earnings call on October 29th, its chief executive, Cindy Rose, was asked why WPP didn't pitch for Coca-Cola's North American media business. 17 00:01:08.080 --> 00:01:11.124 WPP had just won Koch's media almost everywhere else. 18 00:01:11.625 --> 00:01:16.672 And she said it was a tough call between six more weeks of pitching and actually doing the work that it had already won. 19 00:01:16.872 --> 00:01:17.873 And that was all fine. 20 00:01:18.614 --> 00:01:20.076 And that was all strategically fine. 21 00:01:20.177 --> 00:01:24.582 But then she had a dig at rivals who pick up big accounts without pitching at all. 22 00:01:25.083 --> 00:01:27.505 She called it a CEO handshake kind of thing. 23 00:01:27.585 --> 00:01:32.629 Publicist, which has won a run of big accounts without competitive pitching, took that personally. 24 00:01:32.829 --> 00:01:36.952 On November 5th, it posted an open letter on Instagram, and it called Cindy out. 25 00:01:37.292 --> 00:01:42.837 It said, you can't be a passionate defender of pitch culture and then plead exhaustion in the same breath. 26 00:01:43.457 --> 00:01:50.764 And then it said, if an extra six-week run was enough to sideline your teams, it's not an endurance problem, it's a capability one. 27 00:01:50.964 --> 00:01:54.347 That's one multinational corporation telling the other one to get in shape. 28 00:01:54.808 --> 00:01:55.689 And the cherry on top? 29 00:01:56.089 --> 00:02:00.853 The Coke North America pitch WPP skipped on was won by Omnicom, who took it from publicists. 30 00:02:01.714 --> 00:02:07.058 Two companies whose entire business is persuasion and the best they could manage was a subtweet on a letterhead. 31 00:02:07.198 --> 00:02:09.480 And I'm being flippant, but there's a real point here. 32 00:02:09.900 --> 00:02:14.524 Publicis in its letter made the case that the biggest accounts sure aren't one via a pitch. 33 00:02:15.064 --> 00:02:24.772 They're one on data, identity, technology, and months of diligence about whether your systems plug into the client's systems and whether they fill real strategic gaps. 34 00:02:25.572 --> 00:02:33.301 If that's right, the pitch, that strange ritual where agencies give away their best thinking for free, is losing ground to the plumbing underneath it. 35 00:02:33.341 --> 00:02:38.747 So what started as snark is actually pointing us to a very real change happening at the advertising holding company level. 36 00:02:38.787 --> 00:02:43.292 So if you're a smaller agency wondering how these deals are being made, think about it. 37 00:02:43.652 --> 00:02:44.453 Think about data. 38 00:02:44.753 --> 00:02:45.815 Think about the technology. 39 00:02:46.275 --> 00:02:49.857 On Friday, Fox closed its $22 billion acquisition of Roku. 40 00:02:50.457 --> 00:02:58.600 Fox, if you're not in the US, is the Murdoch family's American TV company, the broadcast network, Fox Sports, Fox News, and the free streaming service Tubi. 41 00:02:58.780 --> 00:03:04.762 And Roku makes streaming sticks, TVs, and software that more than 100 million households use to decide what to watch. 42 00:03:05.363 --> 00:03:06.563 The Justice Department cleared it. 43 00:03:06.683 --> 00:03:12.926 The Murdoch spent half a century building one of the great television empires, and then spent $22 billion buying the menu. 44 00:03:13.586 --> 00:03:19.893 which honestly makes perfect sense, especially since they sold the rest of Fox to Disney a year or two ago. 45 00:03:20.734 --> 00:03:25.079 In streaming, owning the shows matters less than owning the screen where people choose the shows. 46 00:03:25.520 --> 00:03:31.186 Whoever controls the home screen controls what gets to be seen, and then who gets the data and who sells the ads around it. 47 00:03:31.727 --> 00:03:34.350 To get it approved, Fox has promised to keep Roku open. 48 00:03:34.690 --> 00:03:42.137 It'll carry non-Fox content, won't rig search results against rivals, and it will firewall its competitors' data for at least three years. 49 00:03:42.218 --> 00:03:46.522 And I gotta tell you, I would love to be in the room for the year four planning meeting. 50 00:03:46.860 --> 00:03:50.422 And then are you excited for me to tell you about the latest in product placement news? 51 00:03:50.983 --> 00:03:55.285 Last week, we talked about clients changing a finished music video after the director had gone home. 52 00:03:55.446 --> 00:03:57.727 This week, that power is becoming a business model. 53 00:03:57.747 --> 00:04:04.151 Two ad insertion companies, Whisper World and Rembrandt, are merging to form something called WR Labs. 54 00:04:04.311 --> 00:04:05.112 And this isn't new. 55 00:04:05.512 --> 00:04:11.456 Whisper World has spent more than a decade putting virtual billboards into TV shots after the actors have left the frame. 56 00:04:11.936 --> 00:04:14.298 And around 30,000 insertions a year. 57 00:04:15.098 --> 00:04:19.045 And the reason it works, as it's been described, is that the director doesn't need to approve it. 58 00:04:19.185 --> 00:04:22.150 Rembrandt does the exact same thing, just with AI. 59 00:04:22.431 --> 00:04:25.997 So they're a match made in heaven, or purgatory, or whatever the other one is. 60 00:04:27.215 --> 00:04:29.838 Unilever, Nestle, Procter & Gamble are already buying. 61 00:04:30.398 --> 00:04:36.104 So for rights holders, old shows can earn new money, and placements can change by country or campaign. 62 00:04:36.304 --> 00:04:42.830 But what's interesting is if there's a poster on the wall of a scene about grief, that's now ad inventory sold by the week. 63 00:04:42.890 --> 00:04:51.338 And I'd be interested to know if DPs and showrunners are having to make lensing decisions based on the ability to sell inventory in a scene. 64 00:04:52.179 --> 00:05:00.463 As many of you may know, a long lens, which is where you film from quite far away, it gets the person's face in focus, but everything behind is blurry. 65 00:05:01.104 --> 00:05:07.707 I wonder if we'll start seeing a lot more things shot with shorter lenses so that the thing can be seen in the back. 66 00:05:07.767 --> 00:05:14.450 And according to a reporting on the deal, the merger also hands Rembrandt a decade of premium TV and film footage to train its AI on. 67 00:05:14.931 --> 00:05:19.353 So now your movie or show is now on an ad server and most likely in a training set. 68 00:05:19.713 --> 00:05:22.236 And then Europe decided to ask a pretty good question. 69 00:05:22.316 --> 00:05:24.799 It asked whether chatbots owe publishers money. 70 00:05:24.979 --> 00:05:31.688 On Tuesday, an advocate general at the EU's top court gave his opinion on the first case the court has taken on generative AI and copyright. 71 00:05:32.188 --> 00:05:37.555 A Hungarian publisher, like Company, sued Google after Gemini produced summaries of its articles. 72 00:05:38.965 --> 00:05:45.187 The Advocate General said that when a chatbot retrieves a publisher's article, it uses it to build an answer and serves that answer up. 73 00:05:45.487 --> 00:05:49.289 And the publisher's rights, quote-unquote, may be infringed in certain circumstances. 74 00:05:49.889 --> 00:05:55.931 On training, though, he said feeding copyrighted material into a model is permitted under Europe's text and data mining exemption. 75 00:05:56.511 --> 00:06:01.793 It's an opinion, not a ruling, but the court usually follows them, and a decision is expected in the coming months. 76 00:06:02.998 --> 00:06:06.162 For years, the fight has been about what AI was allowed to read. 77 00:06:06.522 --> 00:06:08.245 This is about what it's allowed to repeat. 78 00:06:08.445 --> 00:06:11.529 For a news publisher, the second question is the one that hits the business. 79 00:06:11.949 --> 00:06:17.516 Because a chatbot that answers with your journalism is a chatbot whose users never click through to you. 80 00:06:18.117 --> 00:06:21.718 I don't love the training half of this, but the retrieval half is a step forward. 81 00:06:22.218 --> 00:06:25.179 And then NBCUniversal is backing handheld television. 82 00:06:25.599 --> 00:06:39.823 On Tuesday, the Indian company LOLO, or LOLO, I like thinking that it's a hark back to the British LOLO, raised $80 million, co-led by NBCUniversal, at a reported valuation of around a half a billion dollars. 83 00:06:40.484 --> 00:06:44.365 LOLO makes microdramas, mostly through its app StoryTV. 84 00:06:45.145 --> 00:06:51.667 The company says it has more than 40 million monthly users across its apps, spending an average of 95 minutes a day. 85 00:06:52.308 --> 00:06:58.030 It's also built an AI production suite called Dolphin, which its founder says gives it the economics to make it. 86 00:06:58.810 --> 00:07:01.791 NBCUniversal says it'll help take the company to the U.S. 87 00:07:01.971 --> 00:07:04.612 So Hollywood spent a century making the screen bigger. 88 00:07:04.712 --> 00:07:11.155 Widescreen, Cinemascope, 70mm, IMAX, and the massive thing that's in my living room right now. 89 00:07:11.895 --> 00:07:17.457 But one of the fastest-growing entertainment formats on Earth is designed to be watched while you're queuing for coffee. 90 00:07:18.118 --> 00:07:21.359 For producers, it's the counterfactual question in its purest form. 91 00:07:21.899 --> 00:07:24.961 Micro dramas on this scale aren't replacing a prestige drama. 92 00:07:25.321 --> 00:07:29.022 At this price per episode, most of them wouldn't even exist at all without cheap production. 93 00:07:29.222 --> 00:07:32.304 So it goes to a thesis we've had on this podcast for quite some time. 94 00:07:32.364 --> 00:07:36.786 Yes, it's taking attention, but it isn't removing opportunities. 95 00:07:36.886 --> 00:07:40.007 And then I've been promising myself to do more news from the world of art. 96 00:07:41.453 --> 00:07:42.674 And this is a really good one. 97 00:07:43.054 --> 00:07:46.496 AI is selling Aboriginal art without Aboriginal artists. 98 00:07:47.156 --> 00:08:03.104 Sally Scales, a Pintjajajara artist, excuse me if I butchered that, who chairs the National Gallery of Australia's First Nations Advisory Group, told The Guardian this week that AI is turning First Nations art into, in her words, Timu Slop. 99 00:08:03.905 --> 00:08:10.488 AI-generated designs that imitate Aboriginal art are turning up for sale online with no connection to any artist or community. 100 00:08:11.313 --> 00:08:12.774 There's no joke in the harm here. 101 00:08:13.134 --> 00:08:18.277 These styles carry family lineage, country, and knowledge that copyright was never designed to protect. 102 00:08:18.877 --> 00:08:21.899 If there's anything absurd, it's the business model around it. 103 00:08:22.659 --> 00:08:30.344 We built technology that can imitate any artistic tradition on Earth, and now it's printing a culture it doesn't understand onto cheap phone cases. 104 00:08:31.264 --> 00:08:35.827 It's an indictment of the marketplaces, it's an indictment of AI, it's an indictment of copyright. 105 00:08:36.874 --> 00:08:38.776 It's an indictment of what we're doing to artists. 106 00:08:39.236 --> 00:08:44.862 And anyone in designer retail who says, well, the AI made it, that doesn't settle where it came from. 107 00:08:45.502 --> 00:08:46.923 And that doesn't settle who it's hurting. 108 00:08:46.963 --> 00:08:50.667 So yeah, more from the world of art, because these sorts of questions are really interesting. 109 00:08:52.095 --> 00:08:54.717 Runway's AI actor can also join your Zoom. 110 00:08:54.938 --> 00:09:03.325 Runway, the AI video company, published research this week on its interactive characters, which turn a single image into a video character you can talk to in real time. 111 00:09:03.526 --> 00:09:07.589 It's a 57 billion parameter model generating 24 frames a second. 112 00:09:07.769 --> 00:09:14.295 And every live conversation runs on four NVIDIA H100 chips that cost tens of thousands of dollars of each. 113 00:09:15.176 --> 00:09:16.899 It's genuinely impressive engineering. 114 00:09:17.199 --> 00:09:20.824 It's also a feature where a character can join your Zoom, Google Meet, or Teams call. 115 00:09:20.944 --> 00:09:28.394 So Silicon Valley finally invented an actor who's not temperamental, is available around the clock, never needs a trailer, and can sit in on your Monday meeting. 116 00:09:29.275 --> 00:09:31.999 I wonder if Hollywood's wondering, why didn't we think of that? 117 00:09:33.841 --> 00:09:36.243 And then a professor took on the robots. 118 00:09:36.423 --> 00:09:41.768 Carolyn Marvin is a 78-year-old retired communications professor from the University of Pennsylvania. 119 00:09:42.228 --> 00:09:47.733 And she's been in the news again this week for standing in front of an Uber Eats delivery robot on Philadelphia sidewalks. 120 00:09:48.133 --> 00:09:52.477 In October, police handcuffed her, put her in a squad car, and cited her for disorderly conduct. 121 00:09:52.997 --> 00:10:01.511 Her argument is that private companies shouldn't get to take over public pavements with their machines, and that protesting corporate behavior in public spaces is her First Amendment right. 122 00:10:02.152 --> 00:10:03.875 I was impeding a machine, she says. 123 00:10:05.217 --> 00:10:06.860 She's also said she'll keep doing it. 124 00:10:08.223 --> 00:10:09.364 Here's the detail I love. 125 00:10:09.764 --> 00:10:14.168 Since 2020, Pennsylvania law has classified these robots as pedestrians. 126 00:10:14.528 --> 00:10:18.831 So legally, this was one pedestrian refusing to let another pedestrian pass. 127 00:10:18.931 --> 00:10:28.959 And I just think of all the billions that have been spent on computer vision and autonomous navigation, and that it's met its match from a retired professor who knows her constitutional rights. 128 00:10:29.059 --> 00:10:33.903 So this story delights me, except for the fact that she was thrown into a squad car and had handcuffs on. 129 00:10:33.923 --> 00:10:36.905 But it is quite charming, and I wish her all the luck in the way. 130 00:10:36.945 --> 00:10:51.876 And I don't think this will be the flashpoint for robots' rights versus human rights, but I think it is telling that because these robots are owned by very large corporations, that it might actually be another front in the discussion about corporate rights versus human rights. 131 00:10:52.678 --> 00:10:56.907 And then the last bit of news I'll talk about is that the Cybertruck production is being wound down. 132 00:10:57.468 --> 00:11:01.736 I presume it's because the Cybertruck can't do a backflip, and that's... 133 00:11:02.918 --> 00:11:05.419 the new standard for creating cars at Tesla. 134 00:11:06.199 --> 00:11:07.720 And so that was the news last week. 135 00:11:07.920 --> 00:11:09.601 But I think there's a much bigger story. 136 00:11:09.741 --> 00:11:14.423 I'm sure you've all seen the blue wave that happened during America's elections last week. 137 00:11:15.043 --> 00:11:18.485 And I think the biggest story of the week didn't look like a technology story. 138 00:11:18.965 --> 00:11:20.245 It looked like a political one. 139 00:11:20.645 --> 00:11:24.207 And so the title of this episode is, what are we going to do with all this cloud? 140 00:11:24.667 --> 00:11:26.088 And that's a real question for me. 141 00:11:26.928 --> 00:11:31.810 On Tuesday, Democrats took control of both houses of Congress and picked up a handful of governorships. 142 00:11:32.311 --> 00:11:34.031 You'll hear it explained a few different ways. 143 00:11:34.371 --> 00:11:36.432 The main one was that it was about the cost of living. 144 00:11:36.813 --> 00:11:41.755 In the final Reuters poll before the vote, nearly half of registered voters called it their top issue. 145 00:11:42.235 --> 00:11:46.199 And about half of Americans say that their finances had gotten worse since January last year. 146 00:11:46.219 --> 00:11:48.060 It was also about billionaires. 147 00:11:48.201 --> 00:11:57.229 In the same poll, 93% of Democrats and 71% of Republicans said that it's inappropriate for billionaires to spend tens of millions of dollars to sway elections. 148 00:11:57.349 --> 00:12:00.492 And a bit more surprisingly, it was also about data centers. 149 00:12:01.673 --> 00:12:03.294 The AP went through the results. 150 00:12:03.655 --> 00:12:14.183 Of the 34 winning candidates for governor in races it had called, 32 support stronger regulation or restrictions on data centers, and 12 support bans or moratoriums. 151 00:12:14.624 --> 00:12:21.069 In local referendums on data centers, restrictions passed in 44 of the 45 votes where counting was finished. 152 00:12:21.730 --> 00:12:25.373 About three-quarters of voters said they'd oppose a new data center in their community. 153 00:12:25.997 --> 00:12:27.258 And it crossed party lines. 154 00:12:27.699 --> 00:12:34.225 One of the people who started a protest group in Lubbock, Texas, it was a Republican, and she said the data centers were on the ballot. 155 00:12:35.206 --> 00:12:42.513 Most coverage is treating those three separate stories as an economy story, an inequality story, and a weird local planning story. 156 00:12:43.013 --> 00:12:46.977 But I think they're one story, and it's one big technology story. 157 00:12:47.057 --> 00:12:48.038 So let's start with the money. 158 00:12:49.301 --> 00:12:54.143 Tech didn't cause your grocery bill to go up, but it is where new fortunes are being made. 159 00:12:54.924 --> 00:13:04.048 15 of the 34 newcomers to this year's Forbes 400 made their money in tech, and they account for nearly 60% of the newcomers' combined wealth. 160 00:13:04.828 --> 00:13:09.292 Forbes counted 45 new AI billionaires in this year to March alone. 161 00:13:09.472 --> 00:13:11.154 Now, a lot of that wealth is just on paper. 162 00:13:11.314 --> 00:13:16.999 It's shares in companies valued on what investors expect them to earn one day, not on what they earn now. 163 00:13:17.679 --> 00:13:20.082 That's the K-shaped economy in one sentence. 164 00:13:20.602 --> 00:13:23.805 If you own assets, the AI boom has been one of the best years of your life. 165 00:13:24.265 --> 00:13:28.229 If you're paying rent and buying groceries on a salary, you mostly get the price rises. 166 00:13:29.109 --> 00:13:30.089 Then the data centers. 167 00:13:30.649 --> 00:13:34.830 The same boom that's minting those fortunes needs somewhere to physically happen. 168 00:13:35.170 --> 00:13:40.732 And that's a very large, very loud warehouse that drinks water, eats electricity, and needs a new substation. 169 00:13:41.372 --> 00:13:42.612 The wealth goes national. 170 00:13:42.752 --> 00:13:43.952 The costs stay local. 171 00:13:44.233 --> 00:13:54.215 Texas elected its first Democratic agriculture commissioner since 1986, Clayton Tucker, who said fighting data centers was critical to his win because his ranchers were worried about water. 172 00:13:55.124 --> 00:13:56.005 Then the politics. 173 00:13:56.365 --> 00:14:03.692 In the state Senate race, Republican Ken Paxton wanted to make Texas the center of America's AI industry through rapid data center construction. 174 00:14:04.253 --> 00:14:06.975 Democrat James Tallarico campaigned for tighter rules. 175 00:14:07.676 --> 00:14:19.147 Elon Musk spent nearly half a billion dollars backing Republicans this cycle, and a Republican super PAC put $158 million of his money into supporting Paxton in September alone. 176 00:14:19.167 --> 00:14:19.347 Right. 177 00:14:19.707 --> 00:14:20.587 Tallarico won. 178 00:14:20.707 --> 00:14:25.590 The morning after, Musk posted, guess my election-interfering space lasers didn't work this time. 179 00:14:26.030 --> 00:14:33.633 And then, data centers in space are looking pretty appealing right now, which, to be fair, is the most honest thing anyone in the industry has said all week. 180 00:14:34.674 --> 00:14:36.655 If your neighbors won't have it, put it in orbit. 181 00:14:37.642 --> 00:14:41.444 Musk became a trillionaire in June when SpaceX went public. 182 00:14:41.564 --> 00:14:49.868 So $445 million for him is the same share of his net worth as $445 is to someone with a million dollars worth. 183 00:14:49.988 --> 00:14:52.369 It tells you the scale of what we're talking about. 184 00:14:52.769 --> 00:14:57.431 I know we look at that and we could be like, can you believe that Elon Musk is flaunting his wealth like that? 185 00:14:57.611 --> 00:14:59.754 I've given $450 to candidates. 186 00:15:00.375 --> 00:15:01.897 It didn't feel like a huge deal. 187 00:15:02.418 --> 00:15:05.602 And that's the level we're dealing with when we're talking about Elon Musk. 188 00:15:05.662 --> 00:15:06.924 And it shows the scale. 189 00:15:07.104 --> 00:15:12.852 Because a political campaign for the richest man on Earth costs about what a nice dinner costs a millionaire. 190 00:15:13.493 --> 00:15:14.253 So put it together. 191 00:15:14.613 --> 00:15:22.135 Technology is creating the new fortunes, needs the new buildings, and is funding the new political spending, all while most voters feel poorer. 192 00:15:22.275 --> 00:15:24.515 So this isn't three stories to me, it's one loop. 193 00:15:24.655 --> 00:15:26.576 But here's where I think a lot of the anger goes wrong. 194 00:15:27.016 --> 00:15:34.417 On Friday, protesters broke into a Microsoft data center construction site in Amsterdam, climbed a crane, and set off flares. 195 00:15:34.837 --> 00:15:37.798 Part of the campaign planned across 15 European countries. 196 00:15:38.238 --> 00:15:42.239 One of them told reporters, the only thing this will be used for is AI slop. 197 00:15:42.879 --> 00:15:43.520 It isn't, though. 198 00:15:43.960 --> 00:15:50.505 Data centers run your email, your cloud storage, your streaming, your render farm, your hospital records, and yes, your AI tools. 199 00:15:50.945 --> 00:15:56.409 And even the AI part isn't necessarily about all the chatbots making pictures of the Pope in a puffer jacket. 200 00:15:57.110 --> 00:15:58.311 Think about just your email. 201 00:15:58.771 --> 00:16:09.179 If AI means everyone now answers all of their emails instead of just 10% of them, and every one of those emails is a little bit longer, that's an enormous amount of extra data that has to be stored somewhere. 202 00:16:09.839 --> 00:16:13.161 You could call a lot of that slop, obviously, but it's slop you need. 203 00:16:13.601 --> 00:16:19.985 The day you go looking for an email exchange with somebody and it isn't there, you'll be furious, because that data has to live somewhere. 204 00:16:20.766 --> 00:16:23.127 And the rest of the world isn't pausing to think about it. 205 00:16:23.808 --> 00:16:33.133 China is building data centers across its cheap energy rural provinces at astonishing speed, with a plan worth about $300 billion over five years. 206 00:16:33.834 --> 00:16:38.317 So when some Democrats started calling for a national data center moratorium the day after the election— 207 00:16:39.117 --> 00:16:40.738 I do think that's the wrong answer. 208 00:16:41.418 --> 00:16:43.619 Stopping building doesn't make the demand go away. 209 00:16:43.719 --> 00:16:47.021 It just moves the computing somewhere else, probably overseas. 210 00:16:47.161 --> 00:16:48.841 But these are also two different stories. 211 00:16:49.222 --> 00:16:56.505 We need data centers is one thing, but that doesn't mean that we need this data center here on these terms and with your grid paying for it. 212 00:16:57.065 --> 00:17:03.408 Those are different arguments, and the industry has been getting away with treating them as one binary choice, data center or no data center. 213 00:17:03.968 --> 00:17:05.429 There are better ways to do this. 214 00:17:06.129 --> 00:17:16.475 Amazon and Viola are building a system in Mississippi that will cool a downer center with treated municipal wastewater, reusing more than 80 million gallons a year instead of pulling it from the local supply. 215 00:17:17.035 --> 00:17:19.017 You can build away from where people live. 216 00:17:19.377 --> 00:17:20.898 You can pay for your own power. 217 00:17:21.238 --> 00:17:24.039 Almost nobody seems to be fighting for that middle ground. 218 00:17:24.520 --> 00:17:26.701 It's either build anywhere or build nothing. 219 00:17:27.938 --> 00:17:30.039 This week, for the first time, somebody flinched. 220 00:17:30.559 --> 00:17:35.342 On Thursday, two days after the election, OpenAI published what it calls a community compact. 221 00:17:35.762 --> 00:17:41.025 It says OpenAI will cover the energy costs of the infrastructure it builds rather than passing them on to the public. 222 00:17:41.405 --> 00:17:43.386 It'll warn communities about the impacts. 223 00:17:43.586 --> 00:17:53.451 It'll publish reports on the jobs it expects to create and stop using non-disclosure agreements with public officials, which tells you that they were using non-disclosure agreements with public officials. 224 00:17:55.172 --> 00:17:57.513 OpenAI calls them compute campuses now. 225 00:17:58.637 --> 00:18:00.819 And I'm sure you'll all feel much better about that. 226 00:18:01.620 --> 00:18:06.024 Two days after the data centers lost an election, they got rebranded as universities. 227 00:18:06.845 --> 00:18:08.687 These are promises and they need checking. 228 00:18:09.087 --> 00:18:15.673 But paying for your own electricity bill and telling the neighbors what you're building is a far better place to start than either extreme. 229 00:18:15.934 --> 00:18:20.538 If any of you know me, you'll know that I'm very seldom on the side of open AI and it's making me uncomfortable. 230 00:18:21.800 --> 00:18:25.342 But if you make films and ads or music or books, this is your problem too. 231 00:18:25.522 --> 00:18:32.967 If these buildings get harder to build, I think creative businesses will feel it in what they pay for rendering, storage, AI tools well before most people notice. 232 00:18:34.688 --> 00:18:37.750 The question was never whether this technology can create enormous wealth. 233 00:18:37.810 --> 00:18:38.690 It plainly can. 234 00:18:39.111 --> 00:18:42.132 The question is whether the people who live next to the machine see any of it. 235 00:18:42.793 --> 00:18:47.376 On Tuesday, a lot of them answered, and I think everybody needs to pay attention to that. 236 00:18:47.576 --> 00:18:51.358 The question is, can we now have a really adult, nonpartisan debate about it? 237 00:18:52.640 --> 00:18:53.060 I doubt it. 238 00:18:54.721 --> 00:18:56.482 So yeah, the cloud has to live somewhere. 239 00:18:56.502 --> 00:18:58.764 It turns out that somewhere has a vote. 240 00:18:59.604 --> 00:19:00.445 So thanks for listening. 241 00:19:00.705 --> 00:19:08.390 Tomorrow's interview episode is with Douglas Clark, who's a wonderful engineer who's made a tool that helps actors with casting and self-tapes at home. 242 00:19:08.870 --> 00:19:12.312 Makes them feel a lot more part of the process and activates it a little bit. 243 00:19:14.033 --> 00:19:19.717 You know, not every role is flying Jacob Adolorty and Jack Loudon out to test for it. 244 00:19:20.537 --> 00:19:21.958 A lot of people are doing this at home. 245 00:19:22.038 --> 00:19:25.922 And so this is a great tool that's going to help them do that and turn in better performances. 246 00:19:26.483 --> 00:19:28.825 So thank you so much for listening, and we'll see you next week.