1 00:00:03,160 --> 00:00:07,360 Speaker 1: Media. I'm mad as hell and I'm not gonna take 2 00:00:07,360 --> 00:00:23,279 Speaker 1: it anymore. Welcome to bear Offline. I'm your host ed Zetron. Now. 3 00:00:23,280 --> 00:00:25,280 Speaker 2: This is the second part of a two part series 4 00:00:25,280 --> 00:00:25,759 Speaker 2: about how. 5 00:00:25,720 --> 00:00:29,080 Speaker 1: Open Ay's stupid, bloody growth myth has gone the entire 6 00:00:29,160 --> 00:00:32,800 Speaker 1: media ecosystem, forcing people who really should know better to 7 00:00:32,840 --> 00:00:37,960 Speaker 1: repeat absolutely insane things as though they're actually perfectly reasonable. Now, 8 00:00:37,960 --> 00:00:40,199 Speaker 1: if you haven't already listened to the last episode and 9 00:00:40,240 --> 00:00:43,639 Speaker 1: come back, just go, just go. I don't I'll need 10 00:00:43,680 --> 00:00:45,479 Speaker 1: you to listen to it otherwise, Let's talk about why 11 00:00:45,479 --> 00:00:47,760 Speaker 1: this bubble actually inflated. And I realized I did a 12 00:00:47,760 --> 00:00:51,200 Speaker 1: monologue about this, but it's so obvious that I needed 13 00:00:51,240 --> 00:00:54,360 Speaker 1: to do a longer episode. But let's start simple. The 14 00:00:54,440 --> 00:00:57,840 Speaker 1: term artificial intelligence is bastardized to the point it effectively 15 00:00:57,880 --> 00:01:00,880 Speaker 1: means nothing and everything at the same time. When people 16 00:01:00,920 --> 00:01:03,400 Speaker 1: hear AI, they think of an autonomous intelligence that can 17 00:01:03,440 --> 00:01:06,240 Speaker 1: do things for them, and generative AI can do things 18 00:01:06,240 --> 00:01:08,320 Speaker 1: for you, like generate an image or a text from 19 00:01:08,360 --> 00:01:11,640 Speaker 1: a simple prompt. As a result, it's easy to manipulate 20 00:01:11,680 --> 00:01:14,600 Speaker 1: people who don't know much about tech, or even try 21 00:01:14,640 --> 00:01:16,480 Speaker 1: it in the first place, into believing that this will 22 00:01:16,560 --> 00:01:18,759 Speaker 1: naturally progress from It can create a bunch of text 23 00:01:18,800 --> 00:01:20,160 Speaker 1: for me that I have to write for my job 24 00:01:20,280 --> 00:01:22,440 Speaker 1: just by me typing in a prompt. Too, it can 25 00:01:22,480 --> 00:01:25,759 Speaker 1: do my job for me just by typing in a prompt. Basically, 26 00:01:25,840 --> 00:01:29,200 Speaker 1: everything you've read about the future of AI extrapolates generative 27 00:01:29,240 --> 00:01:32,039 Speaker 1: AI's ability to sort of generate something a human would 28 00:01:32,120 --> 00:01:35,280 Speaker 1: make and turns it into a It can do whatever 29 00:01:35,319 --> 00:01:37,800 Speaker 1: a human can do, or because some tech in the 30 00:01:37,840 --> 00:01:41,160 Speaker 1: past has sometimes been bad at the beginning and linearly 31 00:01:41,200 --> 00:01:45,679 Speaker 1: improved as time drags on. This illogical thinking underpins the 32 00:01:45,840 --> 00:01:48,640 Speaker 1: entire generative AI boom because we found out exactly how 33 00:01:48,720 --> 00:01:50,559 Speaker 1: many people do not know what the fuck they're talking 34 00:01:50,640 --> 00:01:52,920 Speaker 1: about and are willing to believe the last semi intelligent 35 00:01:52,920 --> 00:01:55,680 Speaker 1: person that they talk to. Generatif AI is kind of 36 00:01:55,720 --> 00:01:59,240 Speaker 1: a remarkable karner, just good enough version of human expression 37 00:01:59,280 --> 00:02:01,920 Speaker 1: to get it past the gatekeepers in finance and the media, 38 00:02:02,040 --> 00:02:04,520 Speaker 1: knowing that neither will apply a second gear of critical 39 00:02:04,560 --> 00:02:07,080 Speaker 1: thinking beyond hook as we'll do an AI now. The 40 00:02:07,160 --> 00:02:10,799 Speaker 1: expectation that general IVAI will transform into this much more 41 00:02:10,840 --> 00:02:14,400 Speaker 1: powerful version requires you to first ignore its existing limitations, 42 00:02:14,639 --> 00:02:16,440 Speaker 1: believing it to be more capable than it is, and 43 00:02:16,480 --> 00:02:18,280 Speaker 1: also ignore the fact that these models have yet to 44 00:02:18,320 --> 00:02:21,880 Speaker 1: show meaningful improvement over the past few years unless you're looking, 45 00:02:21,919 --> 00:02:24,760 Speaker 1: at course, at the benchmarks that they are built specifically 46 00:02:24,760 --> 00:02:28,760 Speaker 1: to pass. They still halluciname, they're still ungodly expensive to run, 47 00:02:28,840 --> 00:02:31,440 Speaker 1: they're still unreliable, and they still don't really do that much. 48 00:02:31,960 --> 00:02:34,720 Speaker 1: We're still chat GPT's growth as galvanized these people into 49 00:02:34,760 --> 00:02:37,800 Speaker 1: believing that this is a legitimate, meaningful movement rather than 50 00:02:37,840 --> 00:02:41,840 Speaker 1: the most successful public relation campaign of all time. Think 51 00:02:41,880 --> 00:02:44,480 Speaker 1: of it like this, and some of you really don't 52 00:02:44,560 --> 00:02:47,519 Speaker 1: like this argument is something couldn't be this big just 53 00:02:47,560 --> 00:02:48,399 Speaker 1: because of the media. 54 00:02:48,639 --> 00:02:49,280 Speaker 2: It really could. 55 00:02:49,440 --> 00:02:52,440 Speaker 1: If almost every single media outlet talked about one thing, 56 00:02:52,560 --> 00:02:54,720 Speaker 1: this thing being generally of AI, and that one thing 57 00:02:54,800 --> 00:02:57,720 Speaker 1: was available from one company, Open Ai, wouldn't it look 58 00:02:57,760 --> 00:03:01,079 Speaker 1: exactly how things look today? Got open ai with hundreds 59 00:03:01,120 --> 00:03:03,000 Speaker 1: of millions of monthly active users, and then a bunch 60 00:03:03,000 --> 00:03:05,519 Speaker 1: of other companies, including big tech firms with multi trillion 61 00:03:05,520 --> 00:03:08,320 Speaker 1: dollar market caps, with somewhere between ten and sixty nine 62 00:03:08,360 --> 00:03:11,240 Speaker 1: million monthly active users, and I don't count Gemini because 63 00:03:11,240 --> 00:03:14,440 Speaker 1: they're cheating. You can't slap it on Google Assistant. That's unfair. 64 00:03:15,040 --> 00:03:17,840 Speaker 1: What we're seeing here is one company taking most of 65 00:03:17,840 --> 00:03:20,120 Speaker 1: the users and money available and doing so because the 66 00:03:20,200 --> 00:03:23,440 Speaker 1: media fucking helped them. People aren't amazed by chat GPT. 67 00:03:23,639 --> 00:03:24,320 Speaker 2: They're curious. 68 00:03:24,360 --> 00:03:26,680 Speaker 1: They're curious about why the media won't shut up about it, 69 00:03:27,280 --> 00:03:30,440 Speaker 1: and we have to realize and recognize and accept that 70 00:03:30,520 --> 00:03:33,360 Speaker 1: part of the reason this bubble was inflated was the 71 00:03:33,360 --> 00:03:36,800 Speaker 1: failure of Google Search. Everyone I talk to that uses 72 00:03:36,840 --> 00:03:39,120 Speaker 1: chat gpt regularly uses it as either a way to 73 00:03:39,160 --> 00:03:42,240 Speaker 1: generate shitty limericks or as a replacement for Google Search, 74 00:03:42,280 --> 00:03:44,400 Speaker 1: a product that Google is deliberately made worse as a 75 00:03:44,400 --> 00:03:47,880 Speaker 1: means of increasing profits. Listen to our award winning episode 76 00:03:48,320 --> 00:03:50,520 Speaker 1: The Man That Destroyed Google Search. I'm going to say 77 00:03:50,520 --> 00:03:53,600 Speaker 1: that forever, and by the way, that's absolutely what you 78 00:03:53,640 --> 00:03:56,760 Speaker 1: should listen to. It's wherebe award winning the best Business episode. 79 00:03:56,800 --> 00:03:57,640 Speaker 2: We won it, folks. 80 00:03:57,760 --> 00:03:59,840 Speaker 1: Anyway, I hate to say this, I really don't like 81 00:03:59,840 --> 00:04:01,800 Speaker 1: saying this. I don't like to give it to them, 82 00:04:01,800 --> 00:04:05,160 Speaker 1: but chat GPT, if I'm honest, is better at processing 83 00:04:05,200 --> 00:04:07,880 Speaker 1: search strings than Google Search, which is not so much 84 00:04:07,880 --> 00:04:10,640 Speaker 1: a sign that chatchpt is good at something as it 85 00:04:10,720 --> 00:04:13,280 Speaker 1: is that Google has stopped innovating in any meaningful way. 86 00:04:13,720 --> 00:04:15,800 Speaker 1: Over time, Google search should have become something that was 87 00:04:15,840 --> 00:04:18,640 Speaker 1: able to interpret searches into kind of like a perfect result, 88 00:04:18,640 --> 00:04:21,120 Speaker 1: which would require the company to improve how it processes 89 00:04:21,160 --> 00:04:25,400 Speaker 1: your requests. It should be how Google used to feel magical. Instead, 90 00:04:25,400 --> 00:04:28,120 Speaker 1: Google search has become dramatically worse, mostly because the company's 91 00:04:28,120 --> 00:04:30,479 Speaker 1: incentives change from help people find something on the web 92 00:04:30,520 --> 00:04:33,000 Speaker 1: to funnel as much traffic and show as many add 93 00:04:33,080 --> 00:04:37,560 Speaker 1: impressions as possible on Google dot Com. It does little 94 00:04:37,560 --> 00:04:40,119 Speaker 1: annoying things like ignoring random words in the search string, 95 00:04:40,240 --> 00:04:42,120 Speaker 1: even though those words were there for a reason. And 96 00:04:42,160 --> 00:04:46,120 Speaker 1: Google doesn't know better than you what you want ideally, 97 00:04:46,160 --> 00:04:48,239 Speaker 1: though they should when it comes to a search product. 98 00:04:48,279 --> 00:04:51,760 Speaker 1: That's ideally what you get someone to search something for you, 99 00:04:51,760 --> 00:04:54,600 Speaker 1: You want them to bring back the best thing. And 100 00:04:55,480 --> 00:04:58,640 Speaker 1: by this point, Google search should have been far more magical, 101 00:04:58,720 --> 00:05:01,040 Speaker 1: more capable of taking a de witched question and turning 102 00:05:01,040 --> 00:05:02,919 Speaker 1: it into a great answer. Would said answer being a 103 00:05:02,920 --> 00:05:06,400 Speaker 1: result from the internet right. Note that nothing I'm saying 104 00:05:06,440 --> 00:05:09,039 Speaker 1: here is actually about generating a results. It's about processing 105 00:05:09,080 --> 00:05:12,240 Speaker 1: a user's query and presenting an answer. The very foundation 106 00:05:12,320 --> 00:05:14,239 Speaker 1: of computing and the thing that Google at one point 107 00:05:14,279 --> 00:05:16,400 Speaker 1: was the best in the world at doing. Thanks to 108 00:05:16,440 --> 00:05:18,599 Speaker 1: Propagar Ragavan, the former head of Ads that led a 109 00:05:18,640 --> 00:05:20,719 Speaker 1: coup to become the head of Search, Google is pulled 110 00:05:20,720 --> 00:05:23,720 Speaker 1: away from being the meaningful source of information on the Internet. 111 00:05:24,000 --> 00:05:26,960 Speaker 1: It really is actually quite sad when you think about it. 112 00:05:27,839 --> 00:05:30,920 Speaker 1: And I'd argue that chat GPT filled the void left 113 00:05:30,920 --> 00:05:33,279 Speaker 1: by Google Search by doing the thing that people wanted 114 00:05:33,320 --> 00:05:36,839 Speaker 1: Google Search to do, answer a question even if the 115 00:05:36,920 --> 00:05:40,160 Speaker 1: user isn't really sure how to ask it. Google Search 116 00:05:40,160 --> 00:05:42,520 Speaker 1: has become clunky and messy, putting the burden of using 117 00:05:42,560 --> 00:05:44,719 Speaker 1: the service on the user rather than helping fill the 118 00:05:44,720 --> 00:05:47,480 Speaker 1: gap between a query and an answer in any meaningful way. 119 00:05:47,880 --> 00:05:50,320 Speaker 1: Google's AI summaries don't even try to do what chat 120 00:05:50,320 --> 00:05:53,320 Speaker 1: GPT does. They just generate summaries based on search results 121 00:05:53,320 --> 00:05:56,000 Speaker 1: and say, uh, mate, is this what you want? I 122 00:05:56,000 --> 00:05:58,920 Speaker 1: don't fucking know? Only may I only mate latsambillion does 123 00:05:58,960 --> 00:06:01,680 Speaker 1: a core give me a fucking break? But one though 124 00:06:01,720 --> 00:06:04,840 Speaker 1: on Google's AI summaries, they're designed to answer a question 125 00:06:04,960 --> 00:06:07,440 Speaker 1: rather than provide a right answer. There's a distinction that 126 00:06:07,480 --> 00:06:09,520 Speaker 1: needs to be made here because it speaks the underlying 127 00:06:09,640 --> 00:06:13,280 Speaker 1: utility of search itself. One good illustration came earlier this 128 00:06:13,360 --> 00:06:15,320 Speaker 1: week when someone noticed that you could ask Google to 129 00:06:15,320 --> 00:06:17,560 Speaker 1: explain the meaning of a completely made up phrase and 130 00:06:17,600 --> 00:06:20,680 Speaker 1: it would dutifully obey you. Two dry frogs in the situation, 131 00:06:20,800 --> 00:06:22,480 Speaker 1: Google said, refer to a group of people in an 132 00:06:22,480 --> 00:06:26,320 Speaker 1: awkward or difficult social situation. Not every insect has a mortgage, 133 00:06:26,320 --> 00:06:28,600 Speaker 1: Google claimed, there's a humorous way of explaining that not 134 00:06:28,720 --> 00:06:33,040 Speaker 1: everything is as it seems. But my favorite is, sorry, 135 00:06:33,040 --> 00:06:36,719 Speaker 1: I haven't read this bags. Big winky on the skillet 136 00:06:36,720 --> 00:06:39,039 Speaker 1: bowl is apparently a slang TERWN that refers to a 137 00:06:39,040 --> 00:06:40,440 Speaker 1: piece of bread with an egg in the middle. 138 00:06:40,520 --> 00:06:44,000 Speaker 2: Or very funny. Sure, but is it useful? 139 00:06:44,080 --> 00:06:46,520 Speaker 1: No? And it's also really cool that this company just 140 00:06:46,600 --> 00:06:50,120 Speaker 1: makes billions of dollars off of this. With all its 141 00:06:50,160 --> 00:06:52,080 Speaker 1: data and all its talent, Google has put out the 142 00:06:52,160 --> 00:06:55,600 Speaker 1: laziest version of an LLM on top of its questionably 143 00:06:55,680 --> 00:06:59,040 Speaker 1: functional search product as a means of impressing shareholders. And 144 00:06:59,080 --> 00:07:02,000 Speaker 1: I'd like to say the results speak for themselves. Now, 145 00:07:02,040 --> 00:07:03,440 Speaker 1: it must be clear none of this is to say 146 00:07:03,480 --> 00:07:08,200 Speaker 1: that chat GPT is good like it really isn't. It's 147 00:07:08,240 --> 00:07:12,160 Speaker 1: just better at understanding a user's request than Google Search. Yes, 148 00:07:12,200 --> 00:07:14,720 Speaker 1: I fundamentally believe, by the way, that five hundred million 149 00:07:14,720 --> 00:07:16,960 Speaker 1: people a week could be using chat GPT as some 150 00:07:17,000 --> 00:07:19,240 Speaker 1: sort of search replacement. And no, no, I do not 151 00:07:19,400 --> 00:07:22,240 Speaker 1: believe that's a functional business model, in part because if 152 00:07:22,280 --> 00:07:26,680 Speaker 1: it was chat gpt open AIE, they'd have a functional business. 153 00:07:26,720 --> 00:07:30,000 Speaker 1: And to be clear, that's a load bearing kit because 154 00:07:30,320 --> 00:07:32,840 Speaker 1: I know that five hundred million people aren't using chat 155 00:07:32,840 --> 00:07:34,200 Speaker 1: GPT as a search replacement. 156 00:07:34,280 --> 00:07:35,400 Speaker 2: It's hypothetical. 157 00:07:35,400 --> 00:07:38,160 Speaker 1: We actually don't know, and we cannot trust what comes 158 00:07:38,160 --> 00:07:41,320 Speaker 1: out of them, not that they share. It's not just 159 00:07:41,360 --> 00:07:43,360 Speaker 1: the open aiy as a shambles of a company. 160 00:07:43,480 --> 00:07:44,000 Speaker 2: It appears that. 161 00:07:44,000 --> 00:07:46,280 Speaker 1: Google's ability to turn search into such a big business 162 00:07:46,320 --> 00:07:48,800 Speaker 1: was because it held a monopoly on search, search advertising, 163 00:07:48,880 --> 00:07:51,800 Speaker 1: and pretty much the entire online ads industry. And if 164 00:07:51,800 --> 00:07:54,760 Speaker 1: it was a truly competitive market and Google wasn't allowed 165 00:07:54,800 --> 00:07:57,880 Speaker 1: to be vertically integrated with the entire digital advertising apparatus 166 00:07:57,920 --> 00:08:00,000 Speaker 1: of the Web, it would likely be making much less 167 00:08:00,040 --> 00:08:02,880 Speaker 1: revenue per user. And that's bad if your Google replacement 168 00:08:02,960 --> 00:08:06,200 Speaker 1: costs many, many, many, many many more times more money 169 00:08:06,240 --> 00:08:08,880 Speaker 1: than Google to run as an a side. By the way, 170 00:08:08,880 --> 00:08:11,520 Speaker 1: if you're wondering. No, open Ai cannot just create a 171 00:08:11,560 --> 00:08:15,400 Speaker 1: Google Search competitor. Search GPT will be a significantly more 172 00:08:15,440 --> 00:08:17,960 Speaker 1: expensive product to run at Google scale than Chat GPT, 173 00:08:18,160 --> 00:08:21,240 Speaker 1: both infrastructurally and then the cost of revenue itself, with 174 00:08:21,360 --> 00:08:24,240 Speaker 1: open ai being forced to create this massive advertising arm 175 00:08:24,240 --> 00:08:27,600 Speaker 1: that does not exist. And by the way, another great 176 00:08:27,640 --> 00:08:29,680 Speaker 1: critique I get is, like open Air, I can just 177 00:08:29,800 --> 00:08:32,440 Speaker 1: hire a bunch of ads people. Why do you think 178 00:08:32,520 --> 00:08:36,800 Speaker 1: there hasn't been another competitor to Search? You realize that 179 00:08:36,840 --> 00:08:38,839 Speaker 1: there are lots of other companies with lots of other 180 00:08:38,920 --> 00:08:41,880 Speaker 1: money that could just do this, and indeed they've tried. 181 00:08:42,000 --> 00:08:44,839 Speaker 1: Why do you think that hasn't worked. Is it because 182 00:08:44,880 --> 00:08:47,320 Speaker 1: of the monopoly. It's because of the monopoly. It's because 183 00:08:47,360 --> 00:08:48,000 Speaker 1: of the monopoly. 184 00:08:48,800 --> 00:08:51,040 Speaker 2: Look, people love the chat GPT in todays. 185 00:08:51,120 --> 00:08:52,559 Speaker 1: It's the box where you can type in one thing, 186 00:08:52,559 --> 00:08:54,800 Speaker 1: get another thing out because it resembles how everybody has 187 00:08:54,840 --> 00:08:56,400 Speaker 1: always wanted Google Search to work. 188 00:08:56,840 --> 00:08:57,680 Speaker 2: Does it work well? 189 00:08:57,720 --> 00:08:59,959 Speaker 1: Who knows, but people feel like they're getting more out. 190 00:09:10,679 --> 00:09:12,880 Speaker 1: But before I go any further, I'd like to talk 191 00:09:12,920 --> 00:09:18,760 Speaker 1: to you about HEGI Artificial general Intelligence art official bloody 192 00:09:18,840 --> 00:09:21,800 Speaker 1: general intelligence, the conscious computer that does not exist and 193 00:09:21,920 --> 00:09:24,719 Speaker 1: is entirely fictional. And this two part of by the Way, 194 00:09:24,720 --> 00:09:28,480 Speaker 1: has been a break from my usual onerous, tortured, punished 195 00:09:28,679 --> 00:09:31,800 Speaker 1: and analysis. And it's because I needed to have a 196 00:09:31,840 --> 00:09:33,200 Speaker 1: little fun. And I think you've kind of heard that 197 00:09:33,200 --> 00:09:35,679 Speaker 1: in the episode. I'm like a little flexible. I'm more 198 00:09:36,120 --> 00:09:38,800 Speaker 1: all like limber and excited, and it also comes from 199 00:09:38,800 --> 00:09:42,319 Speaker 1: a place of frustration. None of this AI stuff has 200 00:09:42,320 --> 00:09:45,080 Speaker 1: ever felt substantive or real because the actual things that 201 00:09:45,120 --> 00:09:47,400 Speaker 1: you can do with Generator IVAI never seemed to come 202 00:09:47,440 --> 00:09:49,360 Speaker 1: close to the things that people are Sam Ortman and 203 00:09:49,360 --> 00:09:51,760 Speaker 1: Warrio Aama Day seem to be promising, nor do they 204 00:09:51,760 --> 00:09:54,559 Speaker 1: come close to the bullshit like people Casey Newton and 205 00:09:54,640 --> 00:09:58,679 Speaker 1: Kevin Ruster peddling. None of this ever resembled artificial general intelligence, 206 00:09:58,679 --> 00:10:00,760 Speaker 1: and if I'm honest, very little it seems to even 207 00:10:00,800 --> 00:10:04,319 Speaker 1: suggest it's a functional industry. And when cynical plants like 208 00:10:04,400 --> 00:10:07,560 Speaker 1: Kevin rus of the Times bumble around asking theoretical questions 209 00:10:07,600 --> 00:10:10,000 Speaker 1: such as, do you think that there's a fifty percent 210 00:10:10,120 --> 00:10:12,560 Speaker 1: chance or higher that AGI to find this an AI 211 00:10:12,640 --> 00:10:15,800 Speaker 1: system that outperforms human experts, Eventually all cognitive tasks will 212 00:10:15,840 --> 00:10:19,760 Speaker 1: be built before twenty thirty, I think when we're reading 213 00:10:20,040 --> 00:10:22,160 Speaker 1: people saying that, and that was Kevin rus at like 214 00:10:22,200 --> 00:10:25,720 Speaker 1: some little AI panel, things surrounded, but just like venture 215 00:10:25,720 --> 00:10:30,080 Speaker 1: capitalists and the senior officials of anthropic, when you've got 216 00:10:30,120 --> 00:10:32,400 Speaker 1: like the lead columnist of the time saying this, we 217 00:10:32,440 --> 00:10:34,480 Speaker 1: should all be terrified. We should be terrified, but not 218 00:10:34,520 --> 00:10:37,520 Speaker 1: of AGI, but that the lead technic columnist that The 219 00:10:37,520 --> 00:10:41,400 Speaker 1: New York Times appears to have an undiagnosed concussion. Rus's logic, 220 00:10:41,440 --> 00:10:43,400 Speaker 1: as with Kate sin Newton's, is based on the idea 221 00:10:43,440 --> 00:10:46,160 Speaker 1: that he's talked to a bunch of people that say, yeah, yeah, 222 00:10:46,160 --> 00:10:48,680 Speaker 1: AGI is right around the corner, rather than any kind 223 00:10:48,679 --> 00:10:52,840 Speaker 1: of like tangible proof for evidence. Just line going up. 224 00:10:53,240 --> 00:10:53,720 Speaker 2: Do you see it? 225 00:10:53,720 --> 00:10:55,920 Speaker 1: The line's going up? Which line is it? Get out 226 00:10:55,960 --> 00:11:01,079 Speaker 1: of my house. Look, I will say I've been a 227 00:11:01,080 --> 00:11:03,080 Speaker 1: little mean on Kevin Ruse. I've been saying that he's 228 00:11:03,120 --> 00:11:06,920 Speaker 1: written the worst shit about artificial intelligence, that he is 229 00:11:07,000 --> 00:11:10,920 Speaker 1: a credulous buffoon or at worse, c cynical plant, that 230 00:11:11,200 --> 00:11:13,640 Speaker 1: Kevin Ruse at The New York Times is regularly pushing 231 00:11:13,679 --> 00:11:17,559 Speaker 1: things like NFTs and crypto and getting basic things wrong, 232 00:11:17,640 --> 00:11:20,319 Speaker 1: like the Helium piece in which Helium lied. 233 00:11:20,160 --> 00:11:22,200 Speaker 2: About their customers. 234 00:11:23,080 --> 00:11:26,400 Speaker 1: I've said that Kevin Ruse regularly says things that don't 235 00:11:26,440 --> 00:11:28,640 Speaker 1: make any sense, and he has indeed written the worst 236 00:11:28,679 --> 00:11:31,959 Speaker 1: tech criticism out there, criticism being the wrong word, take 237 00:11:32,000 --> 00:11:34,080 Speaker 1: analysis out there. And I want to just say that 238 00:11:34,120 --> 00:11:35,960 Speaker 1: I've been wrong. I've been wrong the entire time. What 239 00:11:36,040 --> 00:11:37,800 Speaker 1: I should have said is that he's written the worst 240 00:11:37,840 --> 00:11:41,439 Speaker 1: stuff yet. And I should have said that because he's 241 00:11:41,480 --> 00:11:44,199 Speaker 1: recently published one of the dumbest fucking things I've read 242 00:11:44,200 --> 00:11:45,520 Speaker 1: in my life, and I can't wait to tell you 243 00:11:45,520 --> 00:11:48,760 Speaker 1: about it. His most egregious example of credulousness came in 244 00:11:48,840 --> 00:11:51,840 Speaker 1: late April when he published this thinly veiled puff piece 245 00:11:52,040 --> 00:11:55,080 Speaker 1: about what to do if AI models become conscious in 246 00:11:55,120 --> 00:11:58,880 Speaker 1: the future. Now, this piece, this piece is possibly the 247 00:11:58,960 --> 00:12:01,439 Speaker 1: dumbest tech piece I've read in my life. I really 248 00:12:01,480 --> 00:12:03,680 Speaker 1: mean this. I've read some dog share, I've gone on 249 00:12:03,720 --> 00:12:06,839 Speaker 1: telegram groups for the smallest icos you've seen in your life. 250 00:12:06,840 --> 00:12:09,319 Speaker 1: From then the rutger House speech from a blade runner 251 00:12:09,360 --> 00:12:11,959 Speaker 1: right now, the things I've seen with your eyes different speech. 252 00:12:12,000 --> 00:12:14,400 Speaker 1: I realize this piece from the New York Times is 253 00:12:14,400 --> 00:12:17,640 Speaker 1: called if AI Systems become conscious, should they have rights? 254 00:12:18,200 --> 00:12:21,480 Speaker 1: And the subhead is as artificial intelligence systems become smarter, 255 00:12:21,559 --> 00:12:23,840 Speaker 1: one AI company is trying to figure out what to 256 00:12:23,840 --> 00:12:26,520 Speaker 1: do if they become conscious. I'm going to give you 257 00:12:26,920 --> 00:12:30,360 Speaker 1: exactly one guess what company he's talking about. And here's 258 00:12:30,360 --> 00:12:35,160 Speaker 1: a clue. It's the company Casey Newton's boyfriend WORKSAM. That's right, 259 00:12:35,200 --> 00:12:39,080 Speaker 1: it's Anthropic. Kevin Ruse interviewed two people, both employed by Anthropic, 260 00:12:39,120 --> 00:12:42,440 Speaker 1: with one holding the genuinely hilarious job description of AI 261 00:12:42,559 --> 00:12:47,040 Speaker 1: welfare researcher, who said nakedly insane things like there's only 262 00:12:47,120 --> 00:12:50,080 Speaker 1: a small chance, maybe fifteen percent or so, that Claude 263 00:12:50,160 --> 00:12:53,400 Speaker 1: or another AI system is conscious. And it seems that 264 00:12:53,480 --> 00:12:55,800 Speaker 1: if you find yourself in a situation of bringing some 265 00:12:55,920 --> 00:12:58,240 Speaker 1: new class of being into existence dot dot dart, then 266 00:12:58,280 --> 00:13:01,000 Speaker 1: it seems quite prudent to at least be asking questions 267 00:13:01,000 --> 00:13:04,720 Speaker 1: about whether that system might have its own kinds of experiences. Well, 268 00:13:04,760 --> 00:13:09,199 Speaker 1: I'm experiencing brain damage reading this article out. And what 269 00:13:09,280 --> 00:13:12,240 Speaker 1: makes this article so appalling is that Ruth acknowledges that 270 00:13:12,280 --> 00:13:14,679 Speaker 1: this shit is seen by most level headed people as 271 00:13:14,760 --> 00:13:18,800 Speaker 1: utterly fantastical. He describes the concept of AI consciousness as 272 00:13:18,800 --> 00:13:21,040 Speaker 1: a taboo subject and that many critics will see this 273 00:13:21,120 --> 00:13:23,680 Speaker 1: as crazy talk, but doesn't bother to speak to any 274 00:13:23,720 --> 00:13:26,559 Speaker 1: actual critics. He does, however, speculate on the motives that 275 00:13:26,640 --> 00:13:29,400 Speaker 1: said critics, saying that they might object to an AI 276 00:13:29,440 --> 00:13:32,800 Speaker 1: company's studying consciousness in the first place because it might 277 00:13:32,840 --> 00:13:35,600 Speaker 1: create incentives to train their systems to act more sentient 278 00:13:35,800 --> 00:13:38,360 Speaker 1: than they actually are, Eh, Kevin, Wouldn't it be really 279 00:13:38,440 --> 00:13:41,040 Speaker 1: bad if a generative AI company deliberately did something to 280 00:13:41,040 --> 00:13:43,840 Speaker 1: convince someone that their systems were sentient. You know, someone 281 00:13:43,880 --> 00:13:47,360 Speaker 1: really credulous would accept whatever narrative said AI company was pushing, 282 00:13:47,520 --> 00:13:49,840 Speaker 1: such as that they created a special job just in 283 00:13:49,880 --> 00:13:53,920 Speaker 1: case AI was sentient. Genuine question, Kevin, when you walk 284 00:13:53,960 --> 00:13:55,960 Speaker 1: past the mirror, do you start barking at it because 285 00:13:55,960 --> 00:14:00,240 Speaker 1: you see another guy? Anyway, nothing about anything that open 286 00:14:00,280 --> 00:14:02,640 Speaker 1: AI or anthropic is building or shipping suggests that we 287 00:14:02,679 --> 00:14:05,920 Speaker 1: are anywhere near any kind of autonomous computing. They've used 288 00:14:05,920 --> 00:14:08,840 Speaker 1: the concept of AI safety and now AI welfare as 289 00:14:08,880 --> 00:14:11,960 Speaker 1: a marketing term to convince people that they're expensive, wasteful 290 00:14:12,000 --> 00:14:15,440 Speaker 1: software will somehow become conscious because they're having discussions about 291 00:14:15,440 --> 00:14:18,120 Speaker 1: what to do if it does so, and anyone literally 292 00:14:18,160 --> 00:14:20,360 Speaker 1: any report are accepting this that face Value is doing. 293 00:14:20,440 --> 00:14:23,840 Speaker 1: Their readers are disservice and embarrassing themselves in the process. 294 00:14:24,400 --> 00:14:27,240 Speaker 1: If AI safety advocates cared about I don't know safety 295 00:14:27,320 --> 00:14:29,800 Speaker 1: or AI, they'd have cared about the environmental impact or 296 00:14:29,800 --> 00:14:32,480 Speaker 1: the fact that these models train using stolen material, or 297 00:14:32,480 --> 00:14:36,120 Speaker 1: the fact that these models don't really deliver on their promises. 298 00:14:36,240 --> 00:14:39,160 Speaker 1: And if they did, it would shock the labor market 299 00:14:39,480 --> 00:14:41,560 Speaker 1: and it would hurt millions, if not billions of people. 300 00:14:41,600 --> 00:14:43,760 Speaker 1: And we don't have anywhere near the social safety network 301 00:14:43,800 --> 00:14:45,760 Speaker 1: to support the ones we have right now, let alone 302 00:14:45,800 --> 00:14:49,480 Speaker 1: in this completely fictional future. These companies do not care 303 00:14:49,520 --> 00:14:51,960 Speaker 1: about your safety, and they don't have any way to 304 00:14:52,000 --> 00:14:54,480 Speaker 1: get to AGI. They're full of shit, and it's time 305 00:14:54,520 --> 00:14:56,400 Speaker 1: to stop being honest that you don't have any proof 306 00:14:56,400 --> 00:14:59,880 Speaker 1: that they do anything that they say they will. Also, 307 00:15:00,080 --> 00:15:03,560 Speaker 1: by the way, I think the bubble might actually be bursting. Yeah, 308 00:15:03,880 --> 00:15:06,000 Speaker 1: remember in August of last year I was talking about 309 00:15:06,040 --> 00:15:08,680 Speaker 1: the pale horses of the AI apocalypse. Well, one of 310 00:15:08,680 --> 00:15:11,760 Speaker 1: the major warning signs that the bubble was bursting was 311 00:15:11,840 --> 00:15:14,680 Speaker 1: big tech firms would be reducing their capital expenditures and 312 00:15:14,720 --> 00:15:16,880 Speaker 1: This is a call I've made quite a few times, 313 00:15:16,880 --> 00:15:19,480 Speaker 1: and there was one in alarming Clarity from April fourth, 314 00:15:19,520 --> 00:15:22,080 Speaker 1: twenty twenty four. And I'm going to quote myself here. Well, 315 00:15:22,120 --> 00:15:24,720 Speaker 1: I hope no, that's the wrong voice. Well, I hope 316 00:15:24,760 --> 00:15:26,720 Speaker 1: I'm wrong. The calamity I fear is one where the 317 00:15:26,760 --> 00:15:29,200 Speaker 1: massive overinvestment in data centers is met with a lack 318 00:15:29,240 --> 00:15:31,800 Speaker 1: of meaningful growth or profit, leading to the markets turning 319 00:15:31,840 --> 00:15:33,840 Speaker 1: on the major cloud players that state their future and 320 00:15:33,920 --> 00:15:38,640 Speaker 1: unproven GENERATIVEAI. If businesses don't adopt AI at scale, not experimentally, 321 00:15:38,640 --> 00:15:40,680 Speaker 1: but at the core of their operations, the revenue is 322 00:15:40,720 --> 00:15:42,760 Speaker 1: simply not there to sustain the hype, and once the 323 00:15:42,760 --> 00:15:45,480 Speaker 1: market turns, it will turn hard, demanding efficiency and cutbacks 324 00:15:45,520 --> 00:15:48,840 Speaker 1: that will lead to tens of thousands of job carts. Well, 325 00:15:48,880 --> 00:15:51,080 Speaker 1: we're about to find out if I'm right. On April 326 00:15:51,080 --> 00:15:53,800 Speaker 1: twenty first year, who Finance reported that analyst Josh Beck 327 00:15:54,040 --> 00:15:57,240 Speaker 1: said that Amazon's generative AI revenue for Amazon Web services 328 00:15:57,280 --> 00:16:02,160 Speaker 1: would be five billion dollars five billion dollars this year, 329 00:16:02,840 --> 00:16:05,880 Speaker 1: a remarkably small sum that is not profit, by the way, 330 00:16:05,920 --> 00:16:08,400 Speaker 1: and also are dropping the bucket compared to Amazon's projected 331 00:16:08,400 --> 00:16:10,840 Speaker 1: one hundred and five billion dollars in capital expenditures in 332 00:16:10,880 --> 00:16:13,600 Speaker 1: twenty twenty five, it's seventy eight point two billion dollars 333 00:16:13,640 --> 00:16:16,080 Speaker 1: in expenditures in twenty twenty four, or it's forty eight 334 00:16:16,080 --> 00:16:17,880 Speaker 1: point four billion dollars in twenty twenty three. 335 00:16:17,880 --> 00:16:20,360 Speaker 2: And is that really fucking Are you fucking kidding me? 336 00:16:21,360 --> 00:16:24,360 Speaker 1: Are you fucking kidding me? I have had Jack Coles 337 00:16:24,360 --> 00:16:26,520 Speaker 1: for years telling me this was the next big money driver. 338 00:16:26,720 --> 00:16:28,560 Speaker 1: Five billion dollars, five billion. 339 00:16:28,600 --> 00:16:29,200 Speaker 2: Goddamn dog? 340 00:16:29,280 --> 00:16:31,760 Speaker 1: Are you fucking kid? You'd make more money auctioning dogs. 341 00:16:31,800 --> 00:16:34,040 Speaker 1: This is a disgrace. And if you're wondering, yes, all 342 00:16:34,080 --> 00:16:37,560 Speaker 1: of this is for AI, all of that capex from 343 00:16:37,600 --> 00:16:40,760 Speaker 1: the Yahoo Finance article, the one for Laura Bratton wrote, well, 344 00:16:40,800 --> 00:16:42,480 Speaker 1: this is a quote, by the way, from the article. 345 00:16:42,720 --> 00:16:45,240 Speaker 1: CEO Andy Jassey said in February that the vast majority 346 00:16:45,240 --> 00:16:47,640 Speaker 1: of it's this year's one hundred billion dollars in capital 347 00:16:47,960 --> 00:16:50,240 Speaker 1: investments from the tech giant will go towards building out 348 00:16:50,280 --> 00:16:55,680 Speaker 1: artificial intelligence capacity for its cloud segment Amazon Web Services AWS. Well, shit, 349 00:16:55,760 --> 00:16:57,680 Speaker 1: I bet investors are going to love this. Better save 350 00:16:57,760 --> 00:16:59,280 Speaker 1: some money, Andy, what's that? 351 00:17:00,400 --> 00:17:01,040 Speaker 2: What's that? Handy? 352 00:17:01,920 --> 00:17:06,400 Speaker 1: What Oh, you're saving money already? How oh shit? 353 00:17:06,600 --> 00:17:07,600 Speaker 2: Oh uh oh. 354 00:17:07,720 --> 00:17:09,960 Speaker 1: A report from Wells Fargo analysts published at the end 355 00:17:10,000 --> 00:17:13,240 Speaker 1: of April, called Data Centers Aws goes on Pause says 356 00:17:13,280 --> 00:17:15,720 Speaker 1: that Amazon has paused a portion of its leasing discussions 357 00:17:15,720 --> 00:17:17,960 Speaker 1: on the co location side, and while it's not clear 358 00:17:18,000 --> 00:17:20,119 Speaker 1: the magnitude of the pause, the positioning is similar to 359 00:17:20,160 --> 00:17:23,480 Speaker 1: what analysts have heard recently from Microsoft that they are 360 00:17:23,520 --> 00:17:26,840 Speaker 1: digesting aggressive recent lease ubdeals, pulling back from a pipeline 361 00:17:26,840 --> 00:17:30,359 Speaker 1: of loys or sqq's. Now, some asshole is going to 362 00:17:30,400 --> 00:17:32,600 Speaker 1: say lwys and sqqs aren't a big deal, But they are. 363 00:17:33,160 --> 00:17:36,040 Speaker 1: They are. Go listen to the Power Cut episodes. Go listen, 364 00:17:36,080 --> 00:17:38,040 Speaker 1: Go back and listen to them right now, Get off 365 00:17:38,080 --> 00:17:39,639 Speaker 1: my how'd you get in my house? 366 00:17:39,960 --> 00:17:40,360 Speaker 2: Anyway? 367 00:17:40,840 --> 00:17:43,480 Speaker 1: Digesting in this case refers to when hyperscalers sit with 368 00:17:43,480 --> 00:17:45,560 Speaker 1: their current capacity for a minute, and Wells Fargo adds 369 00:17:45,600 --> 00:17:47,960 Speaker 1: that these periods typically last six to twelve months, that 370 00:17:48,040 --> 00:17:50,560 Speaker 1: they can be much shorter. It's not obvious how much 371 00:17:50,560 --> 00:17:53,000 Speaker 1: capacity Amazon is walking away from, but they're walking away 372 00:17:53,040 --> 00:18:10,560 Speaker 1: from capacity. It's happening. But what if it wasn't just Amazon. 373 00:18:11,640 --> 00:18:13,760 Speaker 1: Another report from a friend of the podcast read people 374 00:18:13,760 --> 00:18:16,560 Speaker 1: I email, occasionally asking for a PDF, of course, referring 375 00:18:16,560 --> 00:18:19,240 Speaker 1: to analyst td CO and put our report last week that, 376 00:18:19,359 --> 00:18:21,240 Speaker 1: while titled in a way that suggested there wasn't a 377 00:18:21,240 --> 00:18:23,320 Speaker 1: pullback at all, actually said that there was. It was 378 00:18:23,320 --> 00:18:25,000 Speaker 1: called like hyperscaled demand is up. 379 00:18:25,280 --> 00:18:25,880 Speaker 2: Let's take a. 380 00:18:25,840 --> 00:18:29,600 Speaker 1: Look at one damning quote a relative to the hyperscalar 381 00:18:29,680 --> 00:18:32,480 Speaker 1: demand backdrop at PCC, which is a conference hyperscal of 382 00:18:32,520 --> 00:18:35,120 Speaker 1: demand has moderated a bit, driven by the Microsoft pullback 383 00:18:35,160 --> 00:18:39,080 Speaker 1: and to a lesser extent Amazon discussed later, particularly in Europe, 384 00:18:39,119 --> 00:18:41,240 Speaker 1: there has been a broader moderation in the urgency and 385 00:18:41,320 --> 00:18:44,320 Speaker 1: speed with which the hyperscalers are looking to take down capacity, 386 00:18:44,680 --> 00:18:47,520 Speaker 1: and the number of large deals a four hundred megwat 387 00:18:47,520 --> 00:18:50,440 Speaker 1: deals in the market appears to have moderated. In plain English, 388 00:18:50,520 --> 00:18:52,720 Speaker 1: this means demand has come down, there's less urgency to 389 00:18:52,720 --> 00:18:56,000 Speaker 1: build stuff, and the market is also slowing down. Cohen 390 00:18:56,040 --> 00:18:59,080 Speaker 1: also added that it observed a moderation in the exuberance 391 00:18:59,280 --> 00:19:02,480 Speaker 1: around the outlook for hyperscale at demand, which characterized the 392 00:19:02,520 --> 00:19:07,439 Speaker 1: market this time last year. Brother, my brother in Christ, 393 00:19:07,520 --> 00:19:09,000 Speaker 1: isn't this meant to be the next big thing? 394 00:19:09,160 --> 00:19:09,520 Speaker 2: We need? 395 00:19:09,560 --> 00:19:12,800 Speaker 1: More exuberants, not less. We're still a Microsoft appears to 396 00:19:12,840 --> 00:19:15,280 Speaker 1: have pulled back even further, with TD Cohen noting that 397 00:19:15,359 --> 00:19:18,040 Speaker 1: there has been a slow in demand, slow down in demand, 398 00:19:18,200 --> 00:19:20,359 Speaker 1: and that it saw very little third party leasing from 399 00:19:20,400 --> 00:19:23,359 Speaker 1: Microsoft this quarter and most damningly in our boat and 400 00:19:23,400 --> 00:19:25,240 Speaker 1: I'll be can't I really want you to take this? 401 00:19:25,320 --> 00:19:28,560 Speaker 1: Let it ring through you? These deals in totality suggest 402 00:19:28,600 --> 00:19:32,960 Speaker 1: that Microsoft's run rate demand has accelerated materially. For those 403 00:19:33,000 --> 00:19:35,040 Speaker 1: of you wondering, it means that they're not getting any 404 00:19:35,080 --> 00:19:38,520 Speaker 1: fucking revenue demand for generative AI. Well, at least Meta 405 00:19:38,560 --> 00:19:44,359 Speaker 1: an Oracle aren't slowing down right well. TD Cohen also 406 00:19:44,359 --> 00:19:47,560 Speaker 1: reported that it received reverse inquiries from industry participants around 407 00:19:47,600 --> 00:19:50,480 Speaker 1: the potential slow down and demand from Oracle, leading to 408 00:19:50,480 --> 00:19:52,360 Speaker 1: the analyst to ask around and find that there had 409 00:19:52,400 --> 00:19:54,840 Speaker 1: been a near term slow down in decision making, admit 410 00:19:55,160 --> 00:19:58,119 Speaker 1: organizational changes at Oracle. That it adds that this might 411 00:19:58,200 --> 00:20:00,200 Speaker 1: not mean this is changing its needs of the beat 412 00:20:00,200 --> 00:20:02,719 Speaker 1: it which it secures capacity. And if you're wondering what 413 00:20:02,760 --> 00:20:04,600 Speaker 1: else this could mean, you were correct in doing so, 414 00:20:04,640 --> 00:20:10,639 Speaker 1: because slowing down traditionally refers to a change in speed. Yeah, 415 00:20:10,880 --> 00:20:12,920 Speaker 1: even the analysts are kind of like trying to dance 416 00:20:12,960 --> 00:20:15,800 Speaker 1: around this. It's kind of sad. TD Cohen also adds 417 00:20:15,840 --> 00:20:18,040 Speaker 1: that Meta has continued demand, which means that they're buying 418 00:20:18,040 --> 00:20:21,440 Speaker 1: more stuff, albeit with less volume of megawat signings quarter 419 00:20:21,480 --> 00:20:24,359 Speaker 1: over quarter, then adding that Meta's data center activity is 420 00:20:24,440 --> 00:20:27,480 Speaker 1: historically being characterized by short periods of strong activity followed 421 00:20:27,520 --> 00:20:30,879 Speaker 1: by digestion. In essence, Meta is signing less megawats of 422 00:20:30,920 --> 00:20:33,560 Speaker 1: compute and has in the past followed these kind of 423 00:20:33,560 --> 00:20:39,360 Speaker 1: aggressive buildouts with fewer buildouts. It doesn't sound good, does it. 424 00:20:39,359 --> 00:20:42,639 Speaker 1: It doesn't sound like we're cooking and now we're going 425 00:20:42,680 --> 00:20:44,720 Speaker 1: to get to and the whole thing's been kind of defensive. 426 00:20:44,720 --> 00:20:45,240 Speaker 2: I don't care. 427 00:20:45,840 --> 00:20:49,000 Speaker 1: I mean, I've yet to receive an email that meaningfully 428 00:20:49,080 --> 00:20:51,960 Speaker 1: changes my opinion on any of this stuff. But a 429 00:20:52,000 --> 00:20:54,199 Speaker 1: little bit of defensiveness in this. It's not because I 430 00:20:54,240 --> 00:20:56,480 Speaker 1: feel self conscious. It's because I'm tired of hearing the 431 00:20:56,480 --> 00:20:59,639 Speaker 1: same fucking questions. But I'll ask you, if you're a 432 00:20:59,640 --> 00:21:02,040 Speaker 1: cruitter or a a hater, or one of the many 433 00:21:02,080 --> 00:21:03,920 Speaker 1: people who would like to see be punished, put me 434 00:21:03,960 --> 00:21:05,760 Speaker 1: on the cross, set me on fire. It's what you 435 00:21:05,800 --> 00:21:07,479 Speaker 1: would love wouldn't it, But you'd love it if I 436 00:21:07,520 --> 00:21:10,199 Speaker 1: was dead putting that joke aside, I will ask my 437 00:21:10,240 --> 00:21:13,440 Speaker 1: haters a question if I'm wrong. How exactly am I wrong? 438 00:21:14,600 --> 00:21:16,960 Speaker 1: I don't know, man, All of this seems all this 439 00:21:17,080 --> 00:21:19,399 Speaker 1: stuff sure seems like the hyper scalers are reducing their 440 00:21:19,440 --> 00:21:22,880 Speaker 1: capital expenditures at a time when tariffs and economic uncertainty 441 00:21:22,920 --> 00:21:26,440 Speaker 1: are making investors far more critical of revenues. It sure 442 00:21:26,560 --> 00:21:29,000 Speaker 1: seems that nobody outside of open AI is making any 443 00:21:29,040 --> 00:21:31,639 Speaker 1: real revenue on generative AI, and they're certainly not making 444 00:21:31,640 --> 00:21:34,119 Speaker 1: a profit. It also seems at this point it's pretty 445 00:21:34,119 --> 00:21:36,320 Speaker 1: obvious that generative AI isn't going to do much more 446 00:21:36,320 --> 00:21:39,000 Speaker 1: than it does today. If Amazon is only making five 447 00:21:39,080 --> 00:21:41,680 Speaker 1: billion dollars in revenue from the literal only shiny new 448 00:21:41,680 --> 00:21:44,760 Speaker 1: thing it has sold on the world's premiere cloud platform 449 00:21:44,880 --> 00:21:46,960 Speaker 1: at a time when businesses are hungry and desperate to 450 00:21:46,960 --> 00:21:49,960 Speaker 1: integrate AI, then there's little chance this suddenly turns into 451 00:21:50,040 --> 00:21:53,960 Speaker 1: a remarkable revenue driver. Amazon made one hundred and eighty 452 00:21:53,960 --> 00:21:57,200 Speaker 1: seven point seven nine billion dollars in their last quarterly earnings, 453 00:21:57,240 --> 00:21:59,880 Speaker 1: and if only five billion dollars of that is coming 454 00:22:00,119 --> 00:22:01,000 Speaker 1: from AI. 455 00:22:00,800 --> 00:22:01,800 Speaker 2: At the height of the bubble. 456 00:22:01,840 --> 00:22:03,879 Speaker 1: It heavily suggests that they may not actually be that 457 00:22:04,000 --> 00:22:06,440 Speaker 1: much money to make either, because it's too expensive to 458 00:22:06,480 --> 00:22:09,960 Speaker 1: run these services, because well, these services don't have the 459 00:22:10,040 --> 00:22:15,080 Speaker 1: kind of addressable market that AWS generally has. Microsoft has 460 00:22:15,119 --> 00:22:17,760 Speaker 1: reported that it was making a paltry thirteen billion dollars 461 00:22:17,800 --> 00:22:19,760 Speaker 1: a year in annualized revenue, so the equivalent of three 462 00:22:19,760 --> 00:22:23,280 Speaker 1: point twenty five billion dollars a quarter selling GENERATIVEAI services 463 00:22:23,280 --> 00:22:26,880 Speaker 1: and model access. The Information reported that Salesforces Agent Force 464 00:22:26,920 --> 00:22:29,680 Speaker 1: bullshit isn't even going to boast sales growth in twenty 465 00:22:29,720 --> 00:22:31,800 Speaker 1: twenty five, in part because it's pitching it as a 466 00:22:31,840 --> 00:22:34,600 Speaker 1: digital labor that can essentially replace humans with tasks, and 467 00:22:34,640 --> 00:22:36,240 Speaker 1: it turns out that it doesn't do that very well 468 00:22:36,280 --> 00:22:39,120 Speaker 1: at all, cost two dollars a conversation and requires paying 469 00:22:39,160 --> 00:22:43,119 Speaker 1: Salesforce to use its data cloud product. What if anything 470 00:22:43,280 --> 00:22:46,160 Speaker 1: suggests that I'm wrong here? The things have worked out 471 00:22:46,200 --> 00:22:48,920 Speaker 1: in the past with things like the Internet and smartphones, 472 00:22:48,960 --> 00:22:51,800 Speaker 1: and that surely that's going to happen with GENERATIVEAI and 473 00:22:51,840 --> 00:22:54,960 Speaker 1: by extension, open AI. That companies like Uber lost money 474 00:22:54,960 --> 00:22:57,480 Speaker 1: and eventually worked out. And I've actually written a thing 475 00:22:57,480 --> 00:22:59,919 Speaker 1: I'll link to about that. You can get it in 476 00:23:00,080 --> 00:23:03,240 Speaker 1: episode notes. Is it that Open Eye is growing fast 477 00:23:03,280 --> 00:23:05,639 Speaker 1: and that somehow discounts the fact that it burns billions 478 00:23:05,640 --> 00:23:07,480 Speaker 1: of dollars and does not appear to have any part 479 00:23:07,480 --> 00:23:10,320 Speaker 1: of the profitability? Do you think that agents will suddenly 480 00:23:10,359 --> 00:23:13,560 Speaker 1: start working and everything will be fine. It's a fucking joke. 481 00:23:13,680 --> 00:23:19,080 Speaker 1: I'm tired of it. I'm really, I'm just I've tried 482 00:23:19,119 --> 00:23:22,240 Speaker 1: to be like intellectual, I've tried to explain it nicely, 483 00:23:22,320 --> 00:23:24,679 Speaker 1: and I'll probably still continue, but I'm just kind of 484 00:23:24,680 --> 00:23:26,560 Speaker 1: fucking tired of it. I'm not going to stop doing this, 485 00:23:26,600 --> 00:23:30,200 Speaker 1: by the way. If anything, I'm more galvanized by my rage. 486 00:23:30,200 --> 00:23:31,800 Speaker 1: My aura ring is going to claim I did a 487 00:23:31,840 --> 00:23:35,400 Speaker 1: workout these episodes. It's great, I'm burning calories in real time. 488 00:23:36,000 --> 00:23:39,480 Speaker 1: But large language models and their associated businesses, in my opinion, 489 00:23:40,080 --> 00:23:43,040 Speaker 1: are a fifty billion dollar industry masquerading as a trillion 490 00:23:43,080 --> 00:23:45,240 Speaker 1: dollar solution for a tech industry. 491 00:23:44,880 --> 00:23:46,320 Speaker 2: That's lost the fucking plant. 492 00:23:46,560 --> 00:23:49,520 Speaker 1: Silicon Valley is dominated by management consults that no longer 493 00:23:49,600 --> 00:23:51,960 Speaker 1: know what innovation looks like, and they've been tricked by 494 00:23:51,960 --> 00:23:54,000 Speaker 1: Sam Worltman, who's kind of a savvy con artist that 495 00:23:54,040 --> 00:23:57,520 Speaker 1: took advantage of their desperation for growth. Generative AI is 496 00:23:57,560 --> 00:24:01,320 Speaker 1: the perfect nihilistic form of tech, a way for people 497 00:24:01,359 --> 00:24:03,040 Speaker 1: to spend a lot of money on power on cloud 498 00:24:03,040 --> 00:24:05,920 Speaker 1: compute because they don't have anything else to do. Large 499 00:24:05,960 --> 00:24:09,600 Speaker 1: language models are boring, unprofitable, cloud software stretched to the limits, 500 00:24:09,640 --> 00:24:13,200 Speaker 1: both ethically and technologically speaking as a means of kind 501 00:24:13,280 --> 00:24:16,640 Speaker 1: of keeping text collapsing growth eer going, and open AI's 502 00:24:16,760 --> 00:24:19,520 Speaker 1: nonprofit mission has been fattened up to make fois graph 503 00:24:19,560 --> 00:24:22,439 Speaker 1: for SaaS companies to upsell their clients and cloud compute 504 00:24:22,440 --> 00:24:25,560 Speaker 1: companies to sell GPUs at an hourly rate. The rot 505 00:24:25,600 --> 00:24:30,359 Speaker 1: economy has consumed the tech industry. Every American tech firm 506 00:24:30,600 --> 00:24:33,160 Speaker 1: has been corrupted by the growth of or cost mindset, 507 00:24:33,280 --> 00:24:35,200 Speaker 1: and thus they no longer know how to make sustainable 508 00:24:35,240 --> 00:24:37,760 Speaker 1: businesses that solve real problems, largely because the people that 509 00:24:37,840 --> 00:24:41,440 Speaker 1: run them haven't experienced them for decades. As a result, 510 00:24:41,520 --> 00:24:43,560 Speaker 1: none of them were ready for when Sam Moultman tricked 511 00:24:43,560 --> 00:24:47,320 Speaker 1: them into believing he was their savior. Generative AI isn't 512 00:24:47,359 --> 00:24:49,400 Speaker 1: about helping you or me to do things, or even 513 00:24:49,440 --> 00:24:53,639 Speaker 1: about replacing workers. It's about making uskus new monthly subscription 514 00:24:53,800 --> 00:24:56,880 Speaker 1: costs for consumers and enterprises new ways to convince people 515 00:24:56,880 --> 00:24:59,639 Speaker 1: to pay more for the things they already have used 516 00:25:00,040 --> 00:25:03,359 Speaker 1: in slightly different ways that oftentimes end up being worse. 517 00:25:04,160 --> 00:25:06,879 Speaker 1: Only an industry out of options would choose this bubble, 518 00:25:07,000 --> 00:25:09,560 Speaker 1: and the punishment for them doing so will be grim. 519 00:25:09,800 --> 00:25:11,159 Speaker 1: I don't know if you think I'm right or not. 520 00:25:11,200 --> 00:25:13,200 Speaker 1: I don't know if you think I'm insane for the 521 00:25:13,240 --> 00:25:15,840 Speaker 1: way I communicate about this industry. Even if you think 522 00:25:15,880 --> 00:25:17,919 Speaker 1: I am, think long and hard about why it is 523 00:25:17,960 --> 00:25:20,520 Speaker 1: you disagree with me, and the consequences of me being 524 00:25:20,560 --> 00:25:25,439 Speaker 1: wrong or right. There's nothing else left after generative AI. 525 00:25:25,960 --> 00:25:29,359 Speaker 1: There are no hypergrowth markets left in tech SaaS Companies 526 00:25:29,359 --> 00:25:32,240 Speaker 1: are out of things to upsell. Google, Microsoft, Amazon, Andmeta 527 00:25:32,320 --> 00:25:35,000 Speaker 1: do not have any other ways to continue your own growth, 528 00:25:35,040 --> 00:25:36,840 Speaker 1: And when the market works that out, there will be 529 00:25:36,880 --> 00:25:40,119 Speaker 1: hell to pay. Hell that will reverberate through the valuations 530 00:25:40,160 --> 00:25:43,280 Speaker 1: of at the very least every public software company and 531 00:25:43,320 --> 00:25:46,040 Speaker 1: many of the hardware wants to I have no idea 532 00:25:46,040 --> 00:25:48,760 Speaker 1: how this ends, and I can't see how any of 533 00:25:48,800 --> 00:25:51,600 Speaker 1: it works out well for anyone. In any case, I'll 534 00:25:51,600 --> 00:25:53,439 Speaker 1: be here to tell you whether it does or not. 535 00:25:53,680 --> 00:25:57,640 Speaker 1: And I must say I love doing the show so much, 536 00:25:57,880 --> 00:26:01,000 Speaker 1: and thank you for humoring me when I experience brain 537 00:26:01,080 --> 00:26:02,200 Speaker 1: madness every week. 538 00:26:02,520 --> 00:26:03,520 Speaker 2: I'm not gonna stop. 539 00:26:03,640 --> 00:26:06,040 Speaker 1: I really enjoyed doing this, and the emails I got 540 00:26:06,040 --> 00:26:09,600 Speaker 1: around the web were genuinely wonderful, the people on the reddit, 541 00:26:09,640 --> 00:26:12,639 Speaker 1: people over email messages, just you're all so wonderful. 542 00:26:13,000 --> 00:26:15,840 Speaker 2: I'm gonna keep doing this. I mean, I'm. 543 00:26:15,760 --> 00:26:18,320 Speaker 1: Contractually agreed to as well, but I love doing this 544 00:26:18,440 --> 00:26:20,399 Speaker 1: and I genuinely feel very. 545 00:26:20,280 --> 00:26:30,480 Speaker 2: Lucky to be able to thank you for listening to 546 00:26:30,520 --> 00:26:31,359 Speaker 2: Better Offline. 547 00:26:31,480 --> 00:26:33,920 Speaker 1: The editor and composer of the Better Offline theme song 548 00:26:34,000 --> 00:26:36,640 Speaker 1: is Matasowski. You can check out more of his music 549 00:26:36,640 --> 00:26:40,320 Speaker 1: and audio projects at Matasowski dot com, M A T 550 00:26:40,320 --> 00:26:44,760 Speaker 1: T O S O W s ki dot com. You 551 00:26:44,800 --> 00:26:47,280 Speaker 1: can email me at easy at Better offline dot com, 552 00:26:47,400 --> 00:26:49,680 Speaker 1: or visit Better offline dot com to find more podcast 553 00:26:49,760 --> 00:26:53,080 Speaker 1: links and of course, my newsletter. I also really recommend 554 00:26:53,119 --> 00:26:55,040 Speaker 1: you go to chat dot Where's youreed dot at to 555 00:26:55,119 --> 00:26:57,920 Speaker 1: visit the discord, and go to our slash Better Offline 556 00:26:57,960 --> 00:27:01,199 Speaker 1: to check out our reddit. Thank you so much for listening. 557 00:27:02,040 --> 00:27:04,520 Speaker 2: Better Offline is a production of Cool Zone Media. 558 00:27:04,640 --> 00:27:08,000 Speaker 1: For more from Cool Zone Media, visit our website coolzonemedia 559 00:27:08,080 --> 00:27:10,880 Speaker 1: dot com, or check us out on the iHeartRadio app, 560 00:27:10,960 --> 00:27:13,399 Speaker 1: Apple Podcasts, or wherever you get your podcasts.