1 00:00:02,800 --> 00:00:03,560 Speaker 1: Zone Media. 2 00:00:05,480 --> 00:00:18,079 Speaker 2: Hi, I'm at Zitron and you're listening to Better Offline. 3 00:00:20,079 --> 00:00:23,400 Speaker 2: As a reminder, you can now buy glorious Better Offline merchandise. 4 00:00:23,400 --> 00:00:25,159 Speaker 2: There's a link to it in the episode notes. The 5 00:00:25,200 --> 00:00:28,120 Speaker 2: t shirts, the tumblers, the toe bags, all that shit. 6 00:00:28,200 --> 00:00:31,880 Speaker 2: It's lovely. You're gonna love it, buy it today. But 7 00:00:32,000 --> 00:00:34,360 Speaker 2: if I'm honest today, I'm actually a little bit pissed off. 8 00:00:34,360 --> 00:00:36,239 Speaker 2: And that's why we've got a two part episode this 9 00:00:36,320 --> 00:00:38,880 Speaker 2: week about how fucking stupid the AI boom has become. 10 00:00:39,000 --> 00:00:42,200 Speaker 2: I wrote fascicle in the script, and I'm gonna be honest. 11 00:00:42,240 --> 00:00:44,319 Speaker 2: They need to be a little more pointed because I've 12 00:00:44,320 --> 00:00:46,760 Speaker 2: written tens of thousands of words about this. Now, I 13 00:00:46,880 --> 00:00:49,880 Speaker 2: recorded hours upon hours of podcast and still to this day, 14 00:00:50,240 --> 00:00:52,440 Speaker 2: people are babbling about the AI revolution. Is the sky 15 00:00:52,520 --> 00:00:54,680 Speaker 2: rains blood and crevices open in the fucking earth, dragging 16 00:00:54,680 --> 00:00:58,080 Speaker 2: houses and cars and dovestigated animals into them ores. Things 17 00:00:58,120 --> 00:01:01,960 Speaker 2: are astronomically fucked outside, Yet the tech media continues to 18 00:01:01,960 --> 00:01:03,680 Speaker 2: tell me to get my swimming trunks on and take 19 00:01:03,680 --> 00:01:06,560 Speaker 2: a long, nice dip in the fucking pool. As you 20 00:01:06,600 --> 00:01:08,120 Speaker 2: can tell, this is going to be a little less 21 00:01:08,120 --> 00:01:11,280 Speaker 2: reserved than usual. I've just I'm a little bit frustrated. 22 00:01:12,200 --> 00:01:14,319 Speaker 2: I don't know why I'm the one saying this, and 23 00:01:14,360 --> 00:01:17,080 Speaker 2: I frequently feel with the I a part time blogger 24 00:01:17,120 --> 00:01:21,119 Speaker 2: and podcaster. I'm writing the things that I'm writing. Since 25 00:01:21,160 --> 00:01:23,400 Speaker 2: I've put out the newsletter open ai as the systemic 26 00:01:23,520 --> 00:01:25,360 Speaker 2: risk of the tech industry, and actually it's a couple 27 00:01:25,360 --> 00:01:27,360 Speaker 2: of weeks back, I did the two parter about it too. 28 00:01:27,560 --> 00:01:29,760 Speaker 2: I've heard nothing in response, As was the case with 29 00:01:29,800 --> 00:01:32,040 Speaker 2: how does open ai survive and how open ai is 30 00:01:32,120 --> 00:01:36,480 Speaker 2: bad business. There just seems to be little concern or 31 00:01:36,520 --> 00:01:38,840 Speaker 2: belief that there's any kind of risk at the heart 32 00:01:38,920 --> 00:01:41,200 Speaker 2: of AI and open ai in particular. And there are 33 00:01:41,200 --> 00:01:44,200 Speaker 2: companies that spent nine billion dollars in twenty twenty four 34 00:01:44,280 --> 00:01:47,280 Speaker 2: to lose five billion dollars. Well, I'd love to add 35 00:01:47,280 --> 00:01:50,080 Speaker 2: a because here, if not, because it's important to be 36 00:01:50,080 --> 00:01:52,800 Speaker 2: intellectually honest and represent views that directly can trast my own, 37 00:01:52,840 --> 00:01:56,000 Speaker 2: even if I do in somewhat sarcastic and sardonic fashion. 38 00:01:56,200 --> 00:01:58,680 Speaker 2: Nobody seems to actually have a cogent response to how 39 00:01:58,680 --> 00:02:01,520 Speaker 2: they write this ship hard fork are case in Newton 40 00:02:01,680 --> 00:02:04,440 Speaker 2: throwing a full scale Psycho Tante on a podcast and 41 00:02:04,480 --> 00:02:08,240 Speaker 2: saying I'm wrong because inference costs are coming down. Inference, 42 00:02:08,320 --> 00:02:10,240 Speaker 2: by the way, is when an AI takes an input 43 00:02:10,320 --> 00:02:13,400 Speaker 2: and produces an output. It's the calculations that take place 44 00:02:13,520 --> 00:02:16,880 Speaker 2: right before Google's generativai assistant attributes the Voltaire quote to 45 00:02:17,120 --> 00:02:20,400 Speaker 2: Michael Jackson or says that black tar heroin, when enjoyed 46 00:02:20,400 --> 00:02:23,440 Speaker 2: in moderation, can help you lose weight. Newton is a 47 00:02:23,560 --> 00:02:27,000 Speaker 2: nakedly captured booster that ran an infographic from Anthropic a 48 00:02:27,000 --> 00:02:29,040 Speaker 2: few weeks ago, likes of which I haven't seen since 49 00:02:29,040 --> 00:02:31,200 Speaker 2: twenty thirteen. It was telling you all the ways that 50 00:02:31,240 --> 00:02:33,720 Speaker 2: people use Genera ivai. It looks like some shit from 51 00:02:33,840 --> 00:02:37,000 Speaker 2: I don't know, early day Mashable. No offense Christina, and 52 00:02:37,040 --> 00:02:40,560 Speaker 2: they essentially treat this company propaganda as gospel. But he's 53 00:02:40,600 --> 00:02:43,280 Speaker 2: really far from the only one with a flimsy attachment 54 00:02:43,320 --> 00:02:46,960 Speaker 2: to reality. The information a publication that genuinely does some 55 00:02:47,040 --> 00:02:49,720 Speaker 2: great stuff, which makes it even more heartbreaking to say this. 56 00:02:49,840 --> 00:02:52,520 Speaker 2: Ran a piece in early April that made me even 57 00:02:52,520 --> 00:02:55,760 Speaker 2: more furious than usual, claiming that open ai was forecasting 58 00:02:55,800 --> 00:02:57,840 Speaker 2: revenue topping one hundred and twenty five billion dollars in 59 00:02:57,880 --> 00:03:01,840 Speaker 2: twenty twenty nine, based on selling agents and monetizing free 60 00:03:01,919 --> 00:03:06,000 Speaker 2: users as a driver to higher revenue agents. I should 61 00:03:06,000 --> 00:03:08,720 Speaker 2: add AI systems that can interact with other systems and 62 00:03:08,760 --> 00:03:11,959 Speaker 2: do stuff. So an AI that can order pizza from 63 00:03:12,000 --> 00:03:14,240 Speaker 2: door dash for you is that's an example of an agent. 64 00:03:14,280 --> 00:03:15,840 Speaker 2: And when I say it can order piece of you, 65 00:03:16,280 --> 00:03:18,960 Speaker 2: I am talking entirely theoretically. Is they cannot do this 66 00:03:19,040 --> 00:03:22,239 Speaker 2: right now and may never be able to do so. Indeed, 67 00:03:22,360 --> 00:03:25,480 Speaker 2: the whole agent thing is just what we wish AI 68 00:03:25,760 --> 00:03:28,680 Speaker 2: was and it actually doesn't work. And the piece reported 69 00:03:28,720 --> 00:03:31,760 Speaker 2: based on things and I quote told to some potential 70 00:03:31,800 --> 00:03:35,120 Speaker 2: and current investors, takes great pains to accept literally everything 71 00:03:35,120 --> 00:03:38,200 Speaker 2: that open ai says is perfectly reasonable, if not gospel, 72 00:03:38,240 --> 00:03:42,240 Speaker 2: even if said things make absolutely no goddamn sense. So, 73 00:03:42,600 --> 00:03:46,080 Speaker 2: according to the information's reporting, open ai expects agents and 74 00:03:46,200 --> 00:03:49,240 Speaker 2: new products, and both of those are quotes to contribute 75 00:03:49,280 --> 00:03:52,360 Speaker 2: tens of billions of dollars of revenue, both in the 76 00:03:52,400 --> 00:03:56,080 Speaker 2: near term somehow contributing three billion dollars in revenue this year, 77 00:03:56,120 --> 00:03:57,760 Speaker 2: which I will get to in a little bit, and 78 00:03:57,800 --> 00:04:00,720 Speaker 2: in the long term with an egregious twenty five billion 79 00:04:00,760 --> 00:04:04,360 Speaker 2: dollars in revenue in twenty twenty nine, even projected to 80 00:04:04,400 --> 00:04:08,360 Speaker 2: come from just new products. If you're wondering what those 81 00:04:08,400 --> 00:04:11,160 Speaker 2: new products might be, I am two because the information 82 00:04:11,240 --> 00:04:13,360 Speaker 2: doesn't seem to know and instead of saying open ai 83 00:04:13,520 --> 00:04:15,400 Speaker 2: has no idea what the fuck they're talking about and 84 00:04:15,480 --> 00:04:19,120 Speaker 2: is just saying stuff, the outlet continues to publish things 85 00:04:19,120 --> 00:04:22,080 Speaker 2: with the kind of empty optimism that's indistinguishable from GPT 86 00:04:22,520 --> 00:04:26,200 Speaker 2: generated LinkedIn posts. Must be clear, the information isn't generating 87 00:04:26,200 --> 00:04:30,040 Speaker 2: their articles they're writing and fresh. I want to be really, 88 00:04:30,040 --> 00:04:33,599 Speaker 2: really clear about something we aren't in nearly in May 89 00:04:33,320 --> 00:04:35,520 Speaker 2: twenty five, and indeed one of these will come out 90 00:04:35,720 --> 00:04:39,320 Speaker 2: actually in May. The second part, I see no evidence 91 00:04:39,320 --> 00:04:41,920 Speaker 2: that open ai even has a marketable agent product they 92 00:04:41,920 --> 00:04:44,680 Speaker 2: can sell, let alone one will it will make three 93 00:04:44,720 --> 00:04:49,320 Speaker 2: billion goddamn dollars off of and they definitely are not 94 00:04:49,400 --> 00:04:51,640 Speaker 2: going to do so in the next six or seven months. 95 00:04:51,760 --> 00:04:55,240 Speaker 2: Oh my, for context, let's triple the revenue of open 96 00:04:55,279 --> 00:04:58,520 Speaker 2: ai that they made reportedly at least from selling access 97 00:04:58,560 --> 00:05:01,520 Speaker 2: to their models via at APIs essentially allowing third party 98 00:05:01,520 --> 00:05:04,160 Speaker 2: companies to use GPT in their apps in the entirety 99 00:05:04,200 --> 00:05:07,160 Speaker 2: of twenty twenty four, and those APIs and models actually 100 00:05:07,160 --> 00:05:09,960 Speaker 2: exist in a meaningful sense, as opposed to whatever the 101 00:05:09,960 --> 00:05:13,440 Speaker 2: far copenai is half baked last agents stuff is. In fact, no, no, no, 102 00:05:13,440 --> 00:05:16,320 Speaker 2: no no. I'm not going to be mean, not would 103 00:05:16,320 --> 00:05:19,320 Speaker 2: be calm, be normal. I'm going to explain exactly what 104 00:05:19,360 --> 00:05:22,200 Speaker 2: the information is reporting in an objective way, because writing 105 00:05:22,240 --> 00:05:24,600 Speaker 2: it out really shows how silly it all sounds. I 106 00:05:24,640 --> 00:05:26,800 Speaker 2: am going to rate they believe a lot, because I 107 00:05:26,960 --> 00:05:30,520 Speaker 2: must be clear how stupid this is. Now, according to 108 00:05:30,560 --> 00:05:32,840 Speaker 2: the Information's reporting, they believe the open AI will make 109 00:05:32,880 --> 00:05:35,240 Speaker 2: three billion dollars in twenty twenty five from selling access 110 00:05:35,240 --> 00:05:37,800 Speaker 2: to its agents. This appears to come from soft Bank, 111 00:05:37,839 --> 00:05:39,960 Speaker 2: which has said it will buy three billion dollars worth 112 00:05:40,000 --> 00:05:42,880 Speaker 2: of open ai products annually. Earlier this year, we got 113 00:05:42,920 --> 00:05:45,600 Speaker 2: a bit of extra information about how soft Bank will 114 00:05:45,680 --> 00:05:47,960 Speaker 2: use these products. It plans to create a system called 115 00:05:48,000 --> 00:05:52,599 Speaker 2: Crystal Intelligence that CRISTL and it's one of the most 116 00:05:52,600 --> 00:05:54,400 Speaker 2: generic names I've ever seen, and it will be a 117 00:05:54,480 --> 00:05:57,520 Speaker 2: kind of general purpose AI agent platform for big enterprises. 118 00:05:58,000 --> 00:06:00,960 Speaker 2: The exact specifics of that will shock you, and that 119 00:06:01,000 --> 00:06:03,560 Speaker 2: there are none, but SoftBank intends to use the technology 120 00:06:03,800 --> 00:06:08,040 Speaker 2: internally across its various portfolio companies, as well as market 121 00:06:08,560 --> 00:06:11,680 Speaker 2: it to other large enterprise companies in Japan. I still 122 00:06:11,720 --> 00:06:14,279 Speaker 2: do not know what the fuck this is Crystal Intelligence 123 00:06:14,600 --> 00:06:17,880 Speaker 2: billions of dollars, billions of dollars, and they just don't. 124 00:06:17,920 --> 00:06:20,400 Speaker 2: They can't even describe what it is, just saying yeah, 125 00:06:20,400 --> 00:06:25,640 Speaker 2: it'll be an agent platform that does stuff with your business, 126 00:06:25,800 --> 00:06:28,600 Speaker 2: Like does that sound good? Can I have? I need 127 00:06:28,640 --> 00:06:32,880 Speaker 2: forty billion dollars? I need forty billion dollars, give me okay. 128 00:06:33,960 --> 00:06:35,880 Speaker 2: I also want to add that the Information can't seem 129 00:06:35,880 --> 00:06:38,839 Speaker 2: to keep its story straight on this issue. Back in February, 130 00:06:38,839 --> 00:06:41,120 Speaker 2: they reported that open AI would make three billion dollars 131 00:06:41,120 --> 00:06:44,200 Speaker 2: in revenue only from agents, with a big beautiful chart 132 00:06:44,240 --> 00:06:47,040 Speaker 2: that said three billion dollars would come from it, only 133 00:06:47,080 --> 00:06:49,120 Speaker 2: to add that it would be soft Bank using open 134 00:06:49,160 --> 00:06:52,839 Speaker 2: AIS products across its companies. Based on these numbers, it 135 00:06:52,880 --> 00:06:55,200 Speaker 2: seems like SoftBank will be the only customer for open 136 00:06:55,240 --> 00:06:58,000 Speaker 2: AIS agents. Well, this most likely won't be the case, 137 00:06:58,000 --> 00:07:00,000 Speaker 2: and it isn't because it excludes anyone willing to pay 138 00:07:00,000 --> 00:07:02,200 Speaker 2: a few bucks to test it out. It nonetheless doesn't 139 00:07:02,200 --> 00:07:04,640 Speaker 2: signal good things for agents as a mass market product, 140 00:07:05,320 --> 00:07:08,120 Speaker 2: not that there were any good signals beforehand. Though agents 141 00:07:08,120 --> 00:07:09,920 Speaker 2: do not exist as a product that can be sold 142 00:07:09,960 --> 00:07:13,040 Speaker 2: at scale. Yes, open ai teasted Operator it's first agent 143 00:07:13,080 --> 00:07:15,080 Speaker 2: at the start of the year, but it doesn't seem 144 00:07:15,120 --> 00:07:18,920 Speaker 2: to be able to do anything. The Information's own reporting 145 00:07:18,920 --> 00:07:22,680 Speaker 2: from mid April highlighted how OpenAIS operate a agent struggled 146 00:07:22,720 --> 00:07:25,320 Speaker 2: with comparison shopping on financial products and that's a quote. 147 00:07:25,680 --> 00:07:28,000 Speaker 2: And how Operator or other agents are and I quote 148 00:07:28,000 --> 00:07:30,720 Speaker 2: again tripped by pop ups or logins as well as 149 00:07:30,840 --> 00:07:33,800 Speaker 2: prompts asking for email addresses and phone numbers for marketing 150 00:07:33,960 --> 00:07:40,400 Speaker 2: purpose purposes, which I think accurately describes most websites. And 151 00:07:40,560 --> 00:07:42,760 Speaker 2: just to summarize from everything I've said, the Information is 152 00:07:42,800 --> 00:07:45,080 Speaker 2: saying that the above product will make open ai three 153 00:07:45,160 --> 00:07:48,160 Speaker 2: billion dollars by the end of twenty twenty five. Sounds 154 00:07:48,280 --> 00:07:52,760 Speaker 2: very real to me, sounds extremely real. I love that 155 00:07:52,800 --> 00:07:56,680 Speaker 2: the business media just prints this. I love this. I 156 00:07:56,720 --> 00:08:01,520 Speaker 2: love this so much. I'm having so much fun. Jesus Christ. 157 00:08:01,600 --> 00:08:04,000 Speaker 2: According to The Information's reporting, they believe that open ai 158 00:08:04,080 --> 00:08:06,640 Speaker 2: will basically double revenue every single year for the next 159 00:08:06,720 --> 00:08:09,240 Speaker 2: four years and make thirteen billion dollars in revenue twenty 160 00:08:09,280 --> 00:08:11,640 Speaker 2: twenty five, more than doubling that to twenty nine billion 161 00:08:11,680 --> 00:08:14,480 Speaker 2: dollars in twenty twenty six, nearly doubling that to fifty 162 00:08:14,480 --> 00:08:17,040 Speaker 2: four billion dollars in twenty twenty seven, and nearly doubling 163 00:08:17,120 --> 00:08:19,840 Speaker 2: that again to eighty six billion dollars in twenty twenty eight, 164 00:08:19,840 --> 00:08:22,840 Speaker 2: and eventually leveling out a ridiculous one hundred and twenty 165 00:08:22,840 --> 00:08:25,680 Speaker 2: five billion dollars of revenue in twenty twenty nine. Said 166 00:08:25,720 --> 00:08:28,520 Speaker 2: revenue estimates as of twenty twenty six includes billions of 167 00:08:28,640 --> 00:08:32,680 Speaker 2: dollars of new products that include free monetization free user 168 00:08:32,720 --> 00:08:35,560 Speaker 2: monetization either and if you're wondering what that means, I 169 00:08:36,080 --> 00:08:40,079 Speaker 2: also am. The information does not explain JESSICLLESSI must have 170 00:08:40,120 --> 00:08:42,640 Speaker 2: been busy being horrible to people that work for her. 171 00:08:42,960 --> 00:08:45,080 Speaker 2: They do, however, say that open ai will start, and 172 00:08:45,080 --> 00:08:48,160 Speaker 2: I'm quoting this won't start generating much revenue from free 173 00:08:48,240 --> 00:08:51,000 Speaker 2: users and other products until next year. That's twenty twenty six. 174 00:08:51,400 --> 00:08:54,559 Speaker 2: In twenty nine, and I'm still quoting. However, it projects 175 00:08:54,600 --> 00:08:57,240 Speaker 2: revenue from free users and other products will reach twenty 176 00:08:57,240 --> 00:09:00,120 Speaker 2: five billion dollars, a one fifth of all revenue, and 177 00:09:00,120 --> 00:09:05,679 Speaker 2: then adds that shopping is another potential avenue. You still 178 00:09:05,720 --> 00:09:08,600 Speaker 2: probably don't know what they're doing, and neither do iron 179 00:09:08,960 --> 00:09:13,040 Speaker 2: I have driven myself insane reading about this. I really 180 00:09:13,080 --> 00:09:16,360 Speaker 2: cannot express my disgust about how willing publications are to 181 00:09:16,400 --> 00:09:19,800 Speaker 2: blindly published projections like these, especially when they're all so stupid. 182 00:09:20,160 --> 00:09:23,280 Speaker 2: Let me just read this to you, all right, and 183 00:09:23,360 --> 00:09:26,720 Speaker 2: I quote. Open ai has already begun experimenting with launching 184 00:09:26,720 --> 00:09:29,800 Speaker 2: software features for shopping. Starting in January, some users can 185 00:09:29,840 --> 00:09:32,920 Speaker 2: access web browsing agent Operator as part of their Prochat 186 00:09:32,960 --> 00:09:36,440 Speaker 2: GBT subscription to here to order groceries from Instacart and 187 00:09:36,480 --> 00:09:40,600 Speaker 2: make restaurant reservations on open table. Just want to be clear. 188 00:09:40,640 --> 00:09:43,600 Speaker 2: This is a few episodes ago I mentioned Casey Newton 189 00:09:43,679 --> 00:09:47,480 Speaker 2: not even being able to say this worked. I just 190 00:09:47,480 --> 00:09:49,400 Speaker 2: want to be really clear as well what the information 191 00:09:49,440 --> 00:09:52,680 Speaker 2: is saying. So they're saying that this experimental software launched 192 00:09:52,679 --> 00:09:55,199 Speaker 2: to an indeterminate amount of people that barely works, is 193 00:09:55,240 --> 00:09:58,199 Speaker 2: going to make open Ai three billion dollars in twenty 194 00:09:58,240 --> 00:10:01,000 Speaker 2: twenty five, and then somehow this is going to lead 195 00:10:01,040 --> 00:10:03,160 Speaker 2: to open ai making twenty nine billion dollars in twenty 196 00:10:03,200 --> 00:10:05,240 Speaker 2: twenty six, and then they're can eventually be up to 197 00:10:05,280 --> 00:10:08,600 Speaker 2: one hunch and twenty five billion dollars. What the fuck? How? How? 198 00:10:08,760 --> 00:10:11,360 Speaker 2: What fucking universe are we all living in? There's no 199 00:10:11,520 --> 00:10:13,480 Speaker 2: proof that open ai can do this other than the 200 00:10:13,480 --> 00:10:15,080 Speaker 2: fact it has a lot of users and a lot 201 00:10:15,120 --> 00:10:27,400 Speaker 2: of venture capital. In fact, I think we have real 202 00:10:27,480 --> 00:10:30,120 Speaker 2: reason to worry about whether open ai even makes its 203 00:10:30,160 --> 00:10:33,520 Speaker 2: current projections. In my last multi part episode and then 204 00:10:33,559 --> 00:10:35,800 Speaker 2: the newsletter open ai as a systemic risk for those 205 00:10:35,840 --> 00:10:37,480 Speaker 2: of you who are like to read while listening to 206 00:10:37,520 --> 00:10:40,599 Speaker 2: my fucking podcast, I wrote the Bloomberg had estimated that 207 00:10:40,640 --> 00:10:43,440 Speaker 2: open ai would triple revenue to twelve point seven billion 208 00:10:43,440 --> 00:10:45,400 Speaker 2: dollars in twenty twenty five, and based on its current 209 00:10:45,440 --> 00:10:48,360 Speaker 2: subscriber base, open ai would effectively have to double its 210 00:10:48,360 --> 00:10:51,840 Speaker 2: current subscription revenue and massively increase its API revenue to 211 00:10:51,920 --> 00:10:55,960 Speaker 2: hit these targets. These projections rely on one entity, SoftBank, 212 00:10:56,000 --> 00:10:59,600 Speaker 2: spending three billion dollars specifically on open AI's services really 213 00:10:59,600 --> 00:11:02,160 Speaker 2: shouldn't said specifically, because they keep changing what it means, 214 00:11:02,600 --> 00:11:05,360 Speaker 2: meaning that they'd have to make enough on API course, 215 00:11:05,400 --> 00:11:07,640 Speaker 2: so people plug in the models into their products to 216 00:11:07,679 --> 00:11:10,160 Speaker 2: generate more revenue. In the open Ai made in subscriptions 217 00:11:10,160 --> 00:11:12,960 Speaker 2: in the entirety of twenty twenty four and something else 218 00:11:13,000 --> 00:11:16,080 Speaker 2: that I can only describe as an act of God, 219 00:11:16,800 --> 00:11:18,880 Speaker 2: and that I admit. It seems that soft bank spending 220 00:11:18,880 --> 00:11:21,160 Speaker 2: commitment is based on usage and not like a flat 221 00:11:21,160 --> 00:11:23,640 Speaker 2: fee where Softpak just hands them three billion dollars and 222 00:11:23,679 --> 00:11:28,280 Speaker 2: gets infinite levels of access. Assuming it's the former, I'd 223 00:11:28,280 --> 00:11:31,120 Speaker 2: be stunned if soft Bank's consumption hits three billion dollars 224 00:11:31,160 --> 00:11:33,200 Speaker 2: in twenty twenty five, even with the massive cost of 225 00:11:33,240 --> 00:11:36,120 Speaker 2: the reasoning models that Crystal Intelligence will maybe be based 226 00:11:36,120 --> 00:11:39,559 Speaker 2: off of. Again, we don't know, and soft Bank announced 227 00:11:39,600 --> 00:11:42,920 Speaker 2: this deal with open Ai in February. Crystal Intelligence, if 228 00:11:42,920 --> 00:11:45,319 Speaker 2: it works, and that is possibly the most load bearing 229 00:11:45,480 --> 00:11:48,280 Speaker 2: IF of all time, will be a massive, complicated and 230 00:11:48,320 --> 00:11:51,680 Speaker 2: ambitious product. Details are vague, but from what I understand, 231 00:11:51,720 --> 00:11:53,640 Speaker 2: soft Bank wants to create an AI that handles a 232 00:11:53,640 --> 00:11:56,320 Speaker 2: bunch of varied tasks that knowledge workers do. I mean, 233 00:11:56,360 --> 00:11:59,880 Speaker 2: it's just the same marketing bullshit. It's the same thing. 234 00:12:00,080 --> 00:12:02,240 Speaker 2: It's the thing they've been lying about before. And to 235 00:12:02,240 --> 00:12:05,520 Speaker 2: be clear, open AI's agents cannot consistently do well anything 236 00:12:05,640 --> 00:12:08,679 Speaker 2: right now, what I believe is happening is that reporters 237 00:12:08,720 --> 00:12:11,240 Speaker 2: are taking open AI's rapid growth in revenue from twenty 238 00:12:11,240 --> 00:12:13,240 Speaker 2: twenty three to twenty twenty four, when they went from 239 00:12:13,280 --> 00:12:15,600 Speaker 2: like tens of millions of dollars a month in the 240 00:12:15,600 --> 00:12:17,960 Speaker 2: beginning of the twenty twenty three to three hundred million 241 00:12:18,000 --> 00:12:20,760 Speaker 2: in August twenty twenty four, genuinely a big leap. They've 242 00:12:20,760 --> 00:12:23,040 Speaker 2: taken this to mean that the company will always effectively 243 00:12:23,040 --> 00:12:25,920 Speaker 2: double or triple revenue every single year forever, with their 244 00:12:25,960 --> 00:12:28,720 Speaker 2: evidence being open ai has said that this will happen 245 00:12:28,720 --> 00:12:32,960 Speaker 2: in projections. It's bullshit. I'm sorry, it's bullshit. It's bullshit. 246 00:12:34,000 --> 00:12:36,880 Speaker 2: As I wrote before in a newsletter, it's called there's 247 00:12:36,920 --> 00:12:40,000 Speaker 2: No Ai Revolution and the accompanying episodes. At the time, 248 00:12:40,080 --> 00:12:43,000 Speaker 2: open Ai effectively is the generative AI industry, and nothing 249 00:12:43,040 --> 00:12:45,720 Speaker 2: about the rest of the generative AI industry suggests that 250 00:12:45,760 --> 00:12:49,520 Speaker 2: the revenue exists to sustain these ridiculous obscenes and frankly 251 00:12:49,559 --> 00:12:53,640 Speaker 2: fucking stupid valuations and projections. What do I mean by that? 252 00:12:53,679 --> 00:12:56,360 Speaker 2: By the way, okay, let me get into it. Chat 253 00:12:56,400 --> 00:12:58,840 Speaker 2: GPT is the only real generative AI product with any 254 00:12:58,920 --> 00:13:02,240 Speaker 2: significant usage, or rather their nearest rivals or a fraction 255 00:13:02,320 --> 00:13:04,480 Speaker 2: of said user base, or maybe I need to be 256 00:13:04,520 --> 00:13:07,960 Speaker 2: a little bit blunter. If anyone held at Google Gemini 257 00:13:08,120 --> 00:13:11,000 Speaker 2: user conference, all the attendees could probably share a cab. 258 00:13:11,960 --> 00:13:14,760 Speaker 2: Believing the open AI growth myth and yes, reporting it 259 00:13:14,800 --> 00:13:18,040 Speaker 2: objectively is both endorsing and believing these numbers is engaging 260 00:13:18,040 --> 00:13:20,840 Speaker 2: in childlike logic where you take one event, which is 261 00:13:20,880 --> 00:13:24,040 Speaker 2: open AI's revenue grew seventeen hundred percent from twenty twenty 262 00:13:24,040 --> 00:13:27,480 Speaker 2: three to twenty twenty four. Wow, to mean another will 263 00:13:27,480 --> 00:13:29,480 Speaker 2: take place, which is the open a I will continue 264 00:13:29,520 --> 00:13:32,920 Speaker 2: to double revenue literally every other year. Another insane thing 265 00:13:32,960 --> 00:13:36,520 Speaker 2: to believe, and you're consciously ignoring difficult questions such as 266 00:13:36,800 --> 00:13:39,439 Speaker 2: how will they do this? And what's the total addressable 267 00:13:39,480 --> 00:13:42,800 Speaker 2: market of large language models and their associated subscriptions exactly? 268 00:13:42,800 --> 00:13:45,360 Speaker 2: And how does this company even survive when it expects 269 00:13:45,400 --> 00:13:47,360 Speaker 2: the costs of inference the triple this year to six 270 00:13:47,400 --> 00:13:52,440 Speaker 2: billion dollars alone. Wait wait, wait, sorry, sorry, I really 271 00:13:52,480 --> 00:13:54,200 Speaker 2: need to be clear with that last one, because it's 272 00:13:54,200 --> 00:13:57,320 Speaker 2: a direct quote from the information hm M. The company 273 00:13:57,320 --> 00:13:59,880 Speaker 2: also expects growth and inference costs the costs of running 274 00:13:59,880 --> 00:14:02,640 Speaker 2: A products such as chat GPT in their underlying models 275 00:14:02,679 --> 00:14:05,320 Speaker 2: to moderate over the next half decade. These costs will 276 00:14:05,320 --> 00:14:07,760 Speaker 2: triple this year referring to twenty twenty five to six 277 00:14:07,800 --> 00:14:11,600 Speaker 2: billion dollars, and rise to nearly forty seven billion dollars 278 00:14:11,679 --> 00:14:14,800 Speaker 2: in twenty thirty. Still, the annual growth rate will fall 279 00:14:14,800 --> 00:14:18,120 Speaker 2: to about thirty percent. Then, okay, thanks Also, are you 280 00:14:18,200 --> 00:14:21,960 Speaker 2: fucking kidding me? Six billion fucking dollars for fucking influence? Hey, 281 00:14:22,000 --> 00:14:24,440 Speaker 2: ca C Newton, I thought those costs were coming down. 282 00:14:24,600 --> 00:14:32,560 Speaker 2: KC case, KC. Wow, he's not here. He's not here anyway. 283 00:14:32,640 --> 00:14:35,480 Speaker 2: That's not really great at all. That's actually really bad. 284 00:14:35,720 --> 00:14:38,200 Speaker 2: The Information reports that open ai will make about eight 285 00:14:38,200 --> 00:14:41,560 Speaker 2: billion dollars some subscriptions to chat GPT in twenty twenty five, 286 00:14:41,760 --> 00:14:44,680 Speaker 2: meaning that seventy five percent of open AI's largest revenue 287 00:14:44,720 --> 00:14:46,720 Speaker 2: source is eaten up by the price of providing it. 288 00:14:47,520 --> 00:14:49,480 Speaker 2: This is meant to be the cheap part. This is 289 00:14:49,520 --> 00:14:52,000 Speaker 2: the one fucking thing people say to me is meant 290 00:14:52,000 --> 00:14:54,080 Speaker 2: to come down in price. I've had assholes saying to 291 00:14:54,120 --> 00:14:56,440 Speaker 2: me for the last year. Custom influence is coming down? 292 00:14:56,800 --> 00:14:59,680 Speaker 2: Is it? Are we living in different dimensions? Are there 293 00:14:59,760 --> 00:15:01,560 Speaker 2: large as parts of the tech media that have fucking 294 00:15:01,600 --> 00:15:04,640 Speaker 2: gas leaks? What am I missing? Tell me what I 295 00:15:04,680 --> 00:15:06,960 Speaker 2: am missing near ed. You haven't take to people to 296 00:15:07,040 --> 00:15:09,200 Speaker 2: Billy Sea. You don't need it it. Shut the fuck up. 297 00:15:09,240 --> 00:15:10,800 Speaker 2: If you are one of these people who says I 298 00:15:10,880 --> 00:15:13,720 Speaker 2: need to in Casey, you're included, man. Fuck like, I'm 299 00:15:13,760 --> 00:15:15,960 Speaker 2: so sick of this, Oh you don't talk to people 300 00:15:16,080 --> 00:15:19,040 Speaker 2: running these things. I am sick of people like Casey, 301 00:15:19,120 --> 00:15:21,880 Speaker 2: you and and others too saying, Oh, you don't talk 302 00:15:21,920 --> 00:15:24,600 Speaker 2: to enough AI people. You haven't listened to them. You mean, 303 00:15:24,680 --> 00:15:26,760 Speaker 2: I haven't listened to the problem of the people that 304 00:15:26,840 --> 00:15:30,320 Speaker 2: make money off of lying about this dog shit? Are 305 00:15:30,360 --> 00:15:33,720 Speaker 2: you really thinking you think that's what's missing from my analysis, 306 00:15:33,840 --> 00:15:36,160 Speaker 2: interviewing people who work at these companies and understanding how 307 00:15:36,200 --> 00:15:38,800 Speaker 2: the technologies work. I know other technologies work. I don't 308 00:15:38,800 --> 00:15:41,200 Speaker 2: need to talk to these fucking people. There are people 309 00:15:41,200 --> 00:15:43,520 Speaker 2: out there like Simon Wilson and Max Wolf who know 310 00:15:43,560 --> 00:15:45,760 Speaker 2: how these things work that I talk to fairly regularly, 311 00:15:45,800 --> 00:15:47,520 Speaker 2: and both of them push back on me because they 312 00:15:47,760 --> 00:15:51,200 Speaker 2: know how large language models work. Those people matter. What 313 00:15:51,360 --> 00:15:54,120 Speaker 2: doesn't matter to me, what will never matter to me 314 00:15:54,360 --> 00:15:56,800 Speaker 2: is what Dario, Emma, Dave, Jack Clark and all the 315 00:15:56,800 --> 00:15:59,760 Speaker 2: other fucking people anthropic thing and I think it's the 316 00:15:59,800 --> 00:16:05,000 Speaker 2: tech and actively honestly malpractice in journalism to pretend that 317 00:16:05,040 --> 00:16:08,720 Speaker 2: there's something ethical about speaking to these people and listening 318 00:16:08,800 --> 00:16:14,200 Speaker 2: and taking in their marketing spiel. It's actually a little 319 00:16:14,240 --> 00:16:18,840 Speaker 2: bit disgusting that this is even a critique leveled at anyone. 320 00:16:20,440 --> 00:16:21,720 Speaker 2: But you can have to forgive me. I'm gonna be 321 00:16:21,760 --> 00:16:23,400 Speaker 2: a little rude, and I know that seemed like it, 322 00:16:23,480 --> 00:16:25,920 Speaker 2: but I'm not even getting excited. In fact, you know what, 323 00:16:27,640 --> 00:16:31,560 Speaker 2: I think it's time, Okay, everyone, I think it's time 324 00:16:32,160 --> 00:16:34,640 Speaker 2: that I go through the most common critiques in AI. 325 00:16:34,720 --> 00:16:37,400 Speaker 2: It's time for me to really sit down. And I'm 326 00:16:37,400 --> 00:16:39,120 Speaker 2: going to do my Kevin Rooth's voice. And I know 327 00:16:39,280 --> 00:16:41,360 Speaker 2: a lot of you like my Kevin Ruth's voice, and 328 00:16:41,480 --> 00:16:43,680 Speaker 2: some of you, not a lot of you. I'm going 329 00:16:43,720 --> 00:16:45,160 Speaker 2: to say I'm being rude to these people, and it 330 00:16:45,160 --> 00:16:47,600 Speaker 2: weakens my analysis, to which I say, kiss my ass, 331 00:16:48,440 --> 00:16:50,480 Speaker 2: I will I will turn you, I will cube you 332 00:16:50,920 --> 00:16:55,200 Speaker 2: like a car in a garbage dump. But let's start, 333 00:16:55,320 --> 00:16:58,400 Speaker 2: shall we. The cost of inference coming down? That's one argument. Okay, 334 00:16:58,600 --> 00:17:02,840 Speaker 2: source sauce, where is your source? If you are someone 335 00:17:02,960 --> 00:17:05,760 Speaker 2: saying to me that the costs of inference are coming down. 336 00:17:06,320 --> 00:17:08,159 Speaker 2: I want your source. I want you to show me 337 00:17:08,240 --> 00:17:09,840 Speaker 2: the costs. I want you to show me the costs 338 00:17:09,920 --> 00:17:12,600 Speaker 2: at scale, because it sure seems like they're increasing for 339 00:17:12,680 --> 00:17:15,240 Speaker 2: open ai, and they're effectively the entire user base of 340 00:17:15,280 --> 00:17:18,560 Speaker 2: the entire generative AI industry. But Edward about deep U, 341 00:17:18,760 --> 00:17:21,359 Speaker 2: sweet idiot child. Deep Seek is not open Ai, and 342 00:17:21,400 --> 00:17:23,920 Speaker 2: open AI's latest models only seem to be getting more 343 00:17:23,920 --> 00:17:26,920 Speaker 2: expensive as time drags on. GPT four point five costs 344 00:17:26,920 --> 00:17:29,639 Speaker 2: seventy five dollars per million input tokens and one hundred 345 00:17:29,680 --> 00:17:32,080 Speaker 2: and fifty dollars per million output tokens, and that the 346 00:17:32,160 --> 00:17:34,880 Speaker 2: risk of repeating myself. Open ai is effectively the entire 347 00:17:34,960 --> 00:17:37,800 Speaker 2: generitive AI industry, at least for the world outside of China. 348 00:17:37,840 --> 00:17:41,760 Speaker 2: On top of that, we actually don't know whether deep 349 00:17:41,760 --> 00:17:44,560 Speaker 2: seak is even profitable to run at scale. It is 350 00:17:44,640 --> 00:17:48,240 Speaker 2: definitely cheaper to run, but we don't know if it's 351 00:17:48,240 --> 00:17:51,320 Speaker 2: actually profitable. Indeed, I don't know, even though how you 352 00:17:51,520 --> 00:17:54,520 Speaker 2: calculate this, because running a deep seek model is just 353 00:17:54,600 --> 00:17:57,040 Speaker 2: one person is one thing. The question is whether you 354 00:17:57,080 --> 00:17:59,840 Speaker 2: could scale it up like open ai. We don't know. 355 00:18:00,000 --> 00:18:02,399 Speaker 2: But let's get up back to the other critiques. This 356 00:18:02,520 --> 00:18:04,680 Speaker 2: is a company, it's growth stage. They can just hit 357 00:18:04,720 --> 00:18:07,240 Speaker 2: the button, it all be profitable. You have the mind 358 00:18:07,240 --> 00:18:09,560 Speaker 2: of a child. If this was the case, why would 359 00:18:09,560 --> 00:18:12,159 Speaker 2: both anthropic and open AI be losing so much money? 360 00:18:12,280 --> 00:18:14,800 Speaker 2: Why are none of the hyper scalers making profit on AI? 361 00:18:14,920 --> 00:18:17,480 Speaker 2: Why does nobody want to talk about the underlying economics 362 00:18:17,680 --> 00:18:20,240 Speaker 2: if they're at the growth stage? And also a little 363 00:18:20,280 --> 00:18:22,200 Speaker 2: side point as well, why have we been at the 364 00:18:22,200 --> 00:18:24,880 Speaker 2: growth stage for years? And why are hyper scalers at 365 00:18:24,880 --> 00:18:28,720 Speaker 2: the growth stage? They're not startups anyway, on to another one, though, 366 00:18:28,760 --> 00:18:30,840 Speaker 2: these are the early days of AI. It's just that 367 00:18:30,920 --> 00:18:34,680 Speaker 2: the early days wrong. Wrong. We have all the king's 368 00:18:34,720 --> 00:18:36,960 Speaker 2: horses and all the king's men, the entire tech industry, 369 00:18:37,119 --> 00:18:39,600 Speaker 2: and more money that has ever been invested into anything 370 00:18:39,640 --> 00:18:42,960 Speaker 2: piled into generative AI, and the result has been utterly mediocre. 371 00:18:43,359 --> 00:18:46,959 Speaker 2: Nobody's making on money on AI other than Nvidia and 372 00:18:47,000 --> 00:18:51,080 Speaker 2: maybe during a consultancy, but ed they're already showing signs 373 00:18:51,119 --> 00:18:54,560 Speaker 2: that the AI is going to be powerful. No it's not. No, 374 00:18:54,680 --> 00:18:57,320 Speaker 2: it's not like I'm If anyone brings these critiques, you 375 00:18:57,440 --> 00:19:00,399 Speaker 2: just say no, no, they're not. Show me, show me, 376 00:19:01,000 --> 00:19:04,240 Speaker 2: show me, why is it the only people I'm giving 377 00:19:04,240 --> 00:19:06,640 Speaker 2: Simon Wilson credit here. He's one of the only people 378 00:19:06,680 --> 00:19:08,960 Speaker 2: who'll show you anything cool. And it's cloud compute stuff. 379 00:19:09,000 --> 00:19:11,879 Speaker 2: It's like relatively boring enterprise stuff. It's exciting for the 380 00:19:12,000 --> 00:19:16,480 Speaker 2: niche cases, like software generally is, but it's really not 381 00:19:16,680 --> 00:19:19,680 Speaker 2: showing any power. We talk about this powerful AI thing, 382 00:19:19,840 --> 00:19:22,840 Speaker 2: is it in the room? Like? Where is this? Where 383 00:19:22,880 --> 00:19:25,719 Speaker 2: is this powerful AI? But then I have actually had 384 00:19:25,720 --> 00:19:28,280 Speaker 2: a few emails saying, ed ed look at Open Eyes 385 00:19:28,480 --> 00:19:31,040 Speaker 2: three model, and I just want to be clear that 386 00:19:31,119 --> 00:19:35,760 Speaker 2: this new and extremely expensive reasoning model also hallucinates more. 387 00:19:36,800 --> 00:19:40,399 Speaker 2: Is that AGI? By the way, is this AGI? Is 388 00:19:40,480 --> 00:19:43,040 Speaker 2: the AGI in the room with us? Did the AGI 389 00:19:43,200 --> 00:19:45,119 Speaker 2: tell you it loved you? Did it tell you to 390 00:19:45,200 --> 00:19:48,760 Speaker 2: leave your wife? Did it offer you sex? I hope you're. 391 00:19:48,640 --> 00:20:07,119 Speaker 1: Okay, But ed Edie really is the early days though 392 00:20:07,119 --> 00:20:08,760 Speaker 1: it's just like this in the early days of the Internet. 393 00:20:08,840 --> 00:20:13,880 Speaker 2: No, it was not, and you're a buffoon for suggesting otherwise. 394 00:20:14,280 --> 00:20:16,439 Speaker 2: Jim Cavello of Goldman Sachs wrote in a note from 395 00:20:16,520 --> 00:20:18,640 Speaker 2: last year and the episode pop Culture, you can listen 396 00:20:18,680 --> 00:20:21,600 Speaker 2: to it, which the early days of the Internet were 397 00:20:21,640 --> 00:20:24,520 Speaker 2: nothing like this. Nothing at all. Nothing. There were these 398 00:20:24,560 --> 00:20:28,240 Speaker 2: sixty two thousand and sixty four thousand dollars some microsystems, servers, yes, 399 00:20:28,280 --> 00:20:32,000 Speaker 2: but there were so so many few of them. But 400 00:20:32,280 --> 00:20:35,199 Speaker 2: ed smartphones, I've got you. I finally have me and 401 00:20:35,680 --> 00:20:38,040 Speaker 2: my sex. People doubted those two, they didn't. I will 402 00:20:38,119 --> 00:20:41,639 Speaker 2: drown you in an icy lake if another person comes 403 00:20:41,680 --> 00:20:46,040 Speaker 2: to me and says, hey, ed smartphones. People doubted smartphones. 404 00:20:46,119 --> 00:20:48,720 Speaker 2: Nobody darted smartphones. Why do people get I've read this 405 00:20:48,760 --> 00:20:50,679 Speaker 2: point so many times, but no one seems to have 406 00:20:50,680 --> 00:20:54,960 Speaker 2: a fucking hyperlink because they're lying. They goddamn lying. Cavello 407 00:20:55,040 --> 00:20:58,320 Speaker 2: of Goldman Sachs also noted and including an entire thing 408 00:20:58,320 --> 00:21:01,040 Speaker 2: about how smartphones were fully telegram after analysts and Advance 409 00:21:01,080 --> 00:21:04,960 Speaker 2: with hundreds of presentations that accurately fit house smartphones rolled out, 410 00:21:05,359 --> 00:21:09,400 Speaker 2: and that said that no roadmap exists for AI. It's 411 00:21:09,520 --> 00:21:13,760 Speaker 2: just we're years into this and I'm still repeating the 412 00:21:13,760 --> 00:21:16,359 Speaker 2: same points, and I still don't have much in return 413 00:21:16,440 --> 00:21:18,480 Speaker 2: other than the cost of inference that going down. But 414 00:21:19,440 --> 00:21:22,040 Speaker 2: here's another point of people like it. They go out ed, 415 00:21:22,080 --> 00:21:23,800 Speaker 2: you're so bone and check out this article, and some 416 00:21:23,880 --> 00:21:26,000 Speaker 2: of you love to emailed me this fucking thing. Not 417 00:21:26,080 --> 00:21:28,640 Speaker 2: many of you, I must be clear. The listeners. You're wonderful. 418 00:21:28,680 --> 00:21:30,240 Speaker 2: I love you so much, but there's one or two 419 00:21:30,280 --> 00:21:32,760 Speaker 2: of you out there. Really you're very attached to your 420 00:21:32,800 --> 00:21:34,760 Speaker 2: generative ais and I'm never gonna like it. But some 421 00:21:34,840 --> 00:21:37,320 Speaker 2: of you like to send me this article from Newsweek 422 00:21:37,520 --> 00:21:39,919 Speaker 2: in nineteen ninety five from a guy who said that 423 00:21:39,960 --> 00:21:42,160 Speaker 2: the web would not be a big business Clifford style. 424 00:21:42,200 --> 00:21:46,600 Speaker 2: He said, why the Web won't be Nirvana? And this piece, 425 00:21:46,680 --> 00:21:48,919 Speaker 2: by the way, is quite detailed. You should read it. 426 00:21:48,960 --> 00:21:51,080 Speaker 2: I'm going to have it in the episode notes. But 427 00:21:51,560 --> 00:21:55,760 Speaker 2: they think that sending me this that one guy one 428 00:21:55,800 --> 00:22:00,200 Speaker 2: guy was wrong. Once one guy he said that the 429 00:22:00,280 --> 00:22:03,240 Speaker 2: Internet wonn't be big, And this proves that I add 430 00:22:03,359 --> 00:22:07,320 Speaker 2: ze Tron what's that ninety nine us like twenty years 431 00:22:07,400 --> 00:22:12,880 Speaker 2: later because one guy said that the Internet won't be big, 432 00:22:13,000 --> 00:22:15,480 Speaker 2: that I am wrong? Summer motherfucker? Have you read the piece? 433 00:22:15,680 --> 00:22:17,679 Speaker 2: That's actually the thing. All of these are things that 434 00:22:17,720 --> 00:22:19,440 Speaker 2: you can box up and use. Some people who use 435 00:22:19,480 --> 00:22:22,320 Speaker 2: this half fast bullshit Clifford Stole basically says that the 436 00:22:22,440 --> 00:22:24,600 Speaker 2: Internet at the time was pretty limited, and yes, he 437 00:22:24,600 --> 00:22:26,320 Speaker 2: conflated that with the idea that he won't be big 438 00:22:26,320 --> 00:22:29,640 Speaker 2: in the future. However, Stole's piece, also, as Michael Hiltzier 439 00:22:29,720 --> 00:22:32,880 Speaker 2: wrote for The La Times, was alarmingly accurate about misinformation 440 00:22:33,160 --> 00:22:36,920 Speaker 2: and sleazy companies selling computerized replacements for education. In any case, 441 00:22:36,960 --> 00:22:38,800 Speaker 2: one guy saying that the Internet won't be big doesn't 442 00:22:38,840 --> 00:22:40,879 Speaker 2: make the fucking thing about Jennity, I v ai and 443 00:22:40,880 --> 00:22:43,320 Speaker 2: you were a simpleton if you think it does. One 444 00:22:43,320 --> 00:22:45,440 Speaker 2: guy being wrong in some way is not a response 445 00:22:45,440 --> 00:22:48,439 Speaker 2: to a criticism. I will crush you like a bug. 446 00:22:48,480 --> 00:22:51,520 Speaker 2: If this is your logic, I will eat you. I 447 00:22:51,600 --> 00:22:53,480 Speaker 2: will put you in my mouth like Kirby, and I'll 448 00:22:53,480 --> 00:22:55,000 Speaker 2: shit you out, and I will have the powers of 449 00:22:55,000 --> 00:22:58,560 Speaker 2: a dunce. Stole's analysis also isn't based on hundreds of 450 00:22:58,640 --> 00:23:01,520 Speaker 2: hours of research and endless reporting. Mine is I will 451 00:23:01,520 --> 00:23:04,439 Speaker 2: grab you from the ceiling like the war Master from Zelda, 452 00:23:04,720 --> 00:23:09,840 Speaker 2: and you will never be heard from again. Anyway. Another argument, 453 00:23:10,680 --> 00:23:12,639 Speaker 2: another argument that people are to give me is the 454 00:23:12,640 --> 00:23:15,639 Speaker 2: open ai and anthropic of research entities not business, and 455 00:23:15,640 --> 00:23:18,560 Speaker 2: that they are not focused on profit. Okay, so just 456 00:23:18,600 --> 00:23:20,399 Speaker 2: so we're clear that if that's the case, they're just 457 00:23:20,400 --> 00:23:23,600 Speaker 2: going to burn money forever. Is that the case or 458 00:23:23,600 --> 00:23:26,400 Speaker 2: are they going to hit like the be profitable button sometime. Also, 459 00:23:26,440 --> 00:23:28,560 Speaker 2: if open ai was a research entity, why does it 460 00:23:28,600 --> 00:23:30,640 Speaker 2: need forty billion dollars some soft bank or to change 461 00:23:30,640 --> 00:23:33,920 Speaker 2: its weird corporate structure to become a full profit Actually wait, 462 00:23:34,040 --> 00:23:37,560 Speaker 2: wait a second, that just occurred to me. Open ai 463 00:23:37,760 --> 00:23:40,800 Speaker 2: is as many as eight hundred million weekly active users. 464 00:23:41,680 --> 00:23:43,800 Speaker 2: That's proof of adoption, right, That's going to be an 465 00:23:43,840 --> 00:23:46,520 Speaker 2: argument that people have. There's some bloke on Blue Sky 466 00:23:47,000 --> 00:23:49,240 Speaker 2: who has just been responding to me every few days 467 00:23:49,440 --> 00:23:51,399 Speaker 2: with this kind of argument, saying, look, look at all 468 00:23:51,440 --> 00:23:54,920 Speaker 2: the users, and look, I get the look. You might 469 00:23:54,960 --> 00:23:57,200 Speaker 2: be a bit horny about this number, but something don't 470 00:23:57,200 --> 00:24:00,000 Speaker 2: make no sense about this number. Or March the first 471 00:24:00,000 --> 00:24:02,240 Speaker 2: twenty twenty five, open ai said that it had five 472 00:24:02,320 --> 00:24:05,320 Speaker 2: hundred million people who used chat GPT every week. Two 473 00:24:05,359 --> 00:24:07,960 Speaker 2: weeks later, Sam Mortman claimed that something like ten percent 474 00:24:08,000 --> 00:24:10,560 Speaker 2: of the world uses their systems a lot. They're referring 475 00:24:10,560 --> 00:24:13,160 Speaker 2: to chat GPTs, and the media took this to mean 476 00:24:13,160 --> 00:24:16,119 Speaker 2: that chat GPT is eight hundred million weekly active users. 477 00:24:16,440 --> 00:24:18,040 Speaker 2: I just want to be clear about something as well. 478 00:24:18,040 --> 00:24:21,480 Speaker 2: Sam Moltman didn't say that, he said the weird vague 479 00:24:21,520 --> 00:24:24,560 Speaker 2: thing about something like ten percent of the world, like 480 00:24:24,640 --> 00:24:26,760 Speaker 2: that's what he said, and everyone just went, oh shit, 481 00:24:26,760 --> 00:24:29,120 Speaker 2: we gotta help help Sam Moorman out got to push 482 00:24:29,119 --> 00:24:31,760 Speaker 2: this bad boy over the edge. And there are three 483 00:24:31,760 --> 00:24:34,040 Speaker 2: ways to interpret what he said, and you tell me 484 00:24:34,080 --> 00:24:36,920 Speaker 2: which one sounds real. Number One, open AI's user base 485 00:24:36,960 --> 00:24:39,840 Speaker 2: increased by three hundred million weekly active users in two weeks. 486 00:24:40,880 --> 00:24:44,560 Speaker 2: Number two. Open ai understated it's user base in the 487 00:24:44,560 --> 00:24:47,520 Speaker 2: announcement of their funding announcement on OpenAI dot com by 488 00:24:47,520 --> 00:24:52,320 Speaker 2: three hundred million users or three. Number three, How about this, 489 00:24:52,600 --> 00:24:56,240 Speaker 2: Sam Moltman fucking lied. I get that some members of 490 00:24:56,320 --> 00:24:58,760 Speaker 2: the media of a weird attachment to this damp little man, 491 00:24:58,800 --> 00:25:01,119 Speaker 2: But have any of you ever considered that he's just 492 00:25:01,160 --> 00:25:03,280 Speaker 2: fucking saying things knowing that you'll print them with the 493 00:25:03,320 --> 00:25:07,399 Speaker 2: kindest possible interpretation. Sam Oltman is a liar. He's lied before, 494 00:25:07,400 --> 00:25:10,040 Speaker 2: and he'll lie again. I wrote an entire newsletter called 495 00:25:10,040 --> 00:25:11,960 Speaker 2: sam Oltman is Full of Shit. You should read it. 496 00:25:11,960 --> 00:25:14,560 Speaker 2: I'm gonna link to it, but way ed Google says 497 00:25:14,560 --> 00:25:16,720 Speaker 2: it as three hundred and fifty million monthly active uses 498 00:25:16,760 --> 00:25:20,359 Speaker 2: on Gemini eat shit Zichron No, you eat shit? Yes, 499 00:25:20,440 --> 00:25:23,600 Speaker 2: Google Gemini has three hundred and fifty million monthly active users, 500 00:25:23,600 --> 00:25:26,919 Speaker 2: and that's because they started replacing Google Assistant with Google 501 00:25:26,960 --> 00:25:30,080 Speaker 2: Gemini in early March. You are being had, You are 502 00:25:30,119 --> 00:25:33,720 Speaker 2: being swindled. If Google replaced Google Search with Google Gemini, 503 00:25:33,720 --> 00:25:38,080 Speaker 2: it would have billions of monthly active users. Jesus Christ, 504 00:25:38,840 --> 00:25:42,520 Speaker 2: Jesus Christ on a goddamn cracker. Even reading this script out, 505 00:25:42,520 --> 00:25:45,240 Speaker 2: I get like some of you have suggested that this 506 00:25:45,280 --> 00:25:48,960 Speaker 2: is at all manufactured. No. Reading this stuff makes me 507 00:25:49,080 --> 00:25:53,200 Speaker 2: very angry because I didn't grow up popular or intelligent 508 00:25:53,200 --> 00:25:55,040 Speaker 2: in any way. I've had to pick this shit up 509 00:25:55,200 --> 00:25:57,520 Speaker 2: as I go, and I don't think what I'm saying 510 00:25:57,640 --> 00:26:00,119 Speaker 2: is crazy, but I am sometimes treated that way, and 511 00:26:00,160 --> 00:26:03,800 Speaker 2: this episode I realize I'm doing myself no favors. But anyway, 512 00:26:03,800 --> 00:26:06,679 Speaker 2: back to the critics, really quickly open AI having hundreds 513 00:26:06,720 --> 00:26:08,960 Speaker 2: of millions of free years as each losing it money 514 00:26:09,000 --> 00:26:11,720 Speaker 2: is proof that the free version of chat GPT is popular, 515 00:26:12,000 --> 00:26:14,600 Speaker 2: largely because the entirety of the media has written about 516 00:26:14,640 --> 00:26:17,520 Speaker 2: AI NonStop for two straight years and mentioned chat GPT 517 00:26:17,680 --> 00:26:20,760 Speaker 2: every single fucking time. Yes, yes, there is a degree 518 00:26:20,800 --> 00:26:22,720 Speaker 2: here of marketing at partnerships of word of mouth of 519 00:26:22,840 --> 00:26:25,720 Speaker 2: some degree of utility. But when you remove the non 520 00:26:25,760 --> 00:26:29,440 Speaker 2: stop free media campaign chat GPT would have petered out 521 00:26:29,720 --> 00:26:32,480 Speaker 2: by now along with this stupid fucking bubble, but edits 522 00:26:32,480 --> 00:26:34,960 Speaker 2: pouf sambody's doing something. Yeah, it's proved that something is 523 00:26:34,960 --> 00:26:39,240 Speaker 2: broken in society. Generative AI has never ever had the 524 00:26:39,359 --> 00:26:43,119 Speaker 2: kind of meaningful business returns or utility that actually underpins 525 00:26:43,119 --> 00:26:46,080 Speaker 2: something meaningful, but it has had enough to make people 526 00:26:46,119 --> 00:26:50,080 Speaker 2: give it a try. Do you not? Actually? No, I 527 00:26:50,119 --> 00:26:52,840 Speaker 2: know you listening. You're gonna get this. In some ways 528 00:26:52,880 --> 00:26:54,880 Speaker 2: this episode has been I mean, in all ways it's 529 00:26:54,880 --> 00:26:57,399 Speaker 2: been pretty rant. In the second one going to be 530 00:26:57,440 --> 00:26:59,720 Speaker 2: even more so. What I'm trying to do here is 531 00:26:59,760 --> 00:27:03,000 Speaker 2: show you how fascicle all this crap is, how ridiculous 532 00:27:03,000 --> 00:27:07,240 Speaker 2: it is, how silly these posits are. These projections are 533 00:27:07,280 --> 00:27:10,080 Speaker 2: the suggestion that what we have today will become something else, 534 00:27:10,359 --> 00:27:13,439 Speaker 2: when all we've had is proof that it won't. Do 535 00:27:13,480 --> 00:27:16,600 Speaker 2: you see the obvious cracks in the wall here, No 536 00:27:16,640 --> 00:27:20,840 Speaker 2: matter how strenuously people like professional credulous dipshits that the 537 00:27:21,240 --> 00:27:24,320 Speaker 2: other big publications tried to pave over them, does any 538 00:27:24,320 --> 00:27:26,720 Speaker 2: of this make sense to you? Because I, even when 539 00:27:26,760 --> 00:27:29,040 Speaker 2: I try and steal man. My own arguments. Can't wrap 540 00:27:29,080 --> 00:27:31,520 Speaker 2: my head around how any of this survives, let alone 541 00:27:31,520 --> 00:27:33,840 Speaker 2: becomes an industry where the biggest player has annual revenue 542 00:27:33,880 --> 00:27:37,479 Speaker 2: is greater than some major industrialized countries. And I know 543 00:27:37,680 --> 00:27:40,439 Speaker 2: some of you the emotions a lot, and I know 544 00:27:40,480 --> 00:27:44,320 Speaker 2: the aggressions a lot. I'm frustrated because I truly believe 545 00:27:44,440 --> 00:27:47,480 Speaker 2: this stuff's falling apart. I truly believe that this was 546 00:27:47,760 --> 00:27:50,960 Speaker 2: never really anything well. I'm saying this. Kevin Rose is 547 00:27:51,000 --> 00:27:53,000 Speaker 2: in the New York Times going, I believe that Agi 548 00:27:53,080 --> 00:27:55,639 Speaker 2: is my friend. I believe AGI will rise out of 549 00:27:55,680 --> 00:27:57,560 Speaker 2: the ground and hug me in the way no one 550 00:27:57,640 --> 00:28:02,199 Speaker 2: ever has. I think that's disgusting on levels. But I 551 00:28:02,240 --> 00:28:06,120 Speaker 2: also think it's genuinely irresponsible. I think all of this 552 00:28:06,200 --> 00:28:10,119 Speaker 2: is I think when this collapses, we're going to have 553 00:28:10,200 --> 00:28:13,160 Speaker 2: to look back and take inventory of how we got here. 554 00:28:15,119 --> 00:28:17,480 Speaker 2: And I need you to in the next episode listen 555 00:28:17,560 --> 00:28:21,119 Speaker 2: to it through the kind of len listen to it 556 00:28:21,160 --> 00:28:23,919 Speaker 2: through the lens. That's how lenses work. I need you 557 00:28:23,960 --> 00:28:25,720 Speaker 2: to just stick with it and realize that all of 558 00:28:25,720 --> 00:28:28,040 Speaker 2: the what this is is trying to show you and 559 00:28:28,119 --> 00:28:31,000 Speaker 2: hopefully other people that you talk to how silly this is, 560 00:28:31,160 --> 00:28:33,159 Speaker 2: how ridiculous this is, and that we have a major 561 00:28:33,200 --> 00:28:36,119 Speaker 2: problem in tech and business media. We have a problem 562 00:28:36,160 --> 00:28:39,000 Speaker 2: where people can come out and just say whatever the 563 00:28:39,160 --> 00:28:43,680 Speaker 2: Charlie Brown had hose of the tech media. And it's 564 00:28:43,720 --> 00:28:46,520 Speaker 2: disgusting to me because there are startups that could use 565 00:28:46,520 --> 00:28:48,719 Speaker 2: this money. There are better things to be done with 566 00:28:48,760 --> 00:28:51,400 Speaker 2: this money. Perhaps they're not hypergrowth markets, but there are 567 00:28:51,400 --> 00:28:54,040 Speaker 2: things that actually exist that could be piled into instead. 568 00:28:54,080 --> 00:28:57,640 Speaker 2: We've done this to make companies look like they can grow, 569 00:28:57,880 --> 00:29:01,520 Speaker 2: to make Sam Altman able to buy another five million 570 00:29:01,600 --> 00:29:04,960 Speaker 2: dollar koncig car. Is that the one he has? Either way, 571 00:29:05,760 --> 00:29:07,959 Speaker 2: I'm not gonna lower the temperature on the next episode. 572 00:29:07,960 --> 00:29:10,000 Speaker 2: I'm gonna be honest. It's gonna be just the spicy. 573 00:29:10,640 --> 00:29:12,480 Speaker 2: But I want you to know all of this frustration 574 00:29:12,600 --> 00:29:14,400 Speaker 2: comes from a place of knowing that we can do better, 575 00:29:14,480 --> 00:29:17,400 Speaker 2: and knowing that the tech industry could do better. Perhaps 576 00:29:17,480 --> 00:29:19,200 Speaker 2: it won't be as big as it is today in 577 00:29:19,240 --> 00:29:22,400 Speaker 2: the future, I don't know, but for it to get better, 578 00:29:22,840 --> 00:29:26,080 Speaker 2: this shit needs to end. Stick around for the next part, 579 00:29:26,120 --> 00:29:28,400 Speaker 2: where I'll talk about how we actually got here, how 580 00:29:28,400 --> 00:29:31,240 Speaker 2: this bubble got inflated, and how nasty the result could 581 00:29:31,280 --> 00:29:42,080 Speaker 2: be At the end, thank you for listening to Better Offline. 582 00:29:42,200 --> 00:29:44,640 Speaker 2: The editor and composer of the Better Offline theme song 583 00:29:44,720 --> 00:29:47,360 Speaker 2: is Matasowski. You can check out more of his music 584 00:29:47,360 --> 00:29:50,880 Speaker 2: and audio projects at Matasowski dot com, M A T 585 00:29:51,040 --> 00:29:55,640 Speaker 2: T O. S O w Ski dot com. You can 586 00:29:55,680 --> 00:29:58,160 Speaker 2: email me at easy at Better offline dot com or 587 00:29:58,240 --> 00:30:00,720 Speaker 2: visit Better Offline dot com to find more podcast links 588 00:30:00,760 --> 00:30:03,920 Speaker 2: and of course, my newsletter. I also really recommend you 589 00:30:03,920 --> 00:30:06,080 Speaker 2: go to chat dot Where's youoead dot at to visit 590 00:30:06,120 --> 00:30:08,760 Speaker 2: the discord, and go to our slash Better Offline to 591 00:30:08,800 --> 00:30:11,920 Speaker 2: check out our reddit. Thank you so much for listening. 592 00:30:12,760 --> 00:30:15,440 Speaker 2: Better Offline is a production of cool Zone Media. For 593 00:30:15,560 --> 00:30:18,720 Speaker 2: more from cool Zone Media, visit our website cool Zonemedia 594 00:30:18,760 --> 00:30:21,640 Speaker 2: dot com, or check us out on the iHeartRadio app, 595 00:30:21,680 --> 00:30:24,360 Speaker 2: Apple Podcasts, or wherever you get your podcasts.