1 00:00:01,840 --> 00:00:07,560 Speaker 1: Also media hell, and welcome to this week's better Offline Monologue. 2 00:00:07,560 --> 00:00:17,000 Speaker 1: I'm your host, led Zeitron, and this has been a 3 00:00:17,000 --> 00:00:19,799 Speaker 1: tough week. I'm not gonna lie. Monday, we recorded a 4 00:00:19,880 --> 00:00:23,040 Speaker 1: wonderful episode with Victoria's song and Ashmun Rodriguez, which got 5 00:00:23,040 --> 00:00:25,640 Speaker 1: lost due to a technical four. And then I recorded 6 00:00:25,720 --> 00:00:28,920 Speaker 1: this in time monolog, which then got lost to a 7 00:00:29,000 --> 00:00:33,600 Speaker 1: completely different different computer, different room, different place, technical error. 8 00:00:34,000 --> 00:00:36,320 Speaker 1: I love making this show for all of you. It's 9 00:00:36,320 --> 00:00:38,400 Speaker 1: been kind of a pisser, but it's important we get 10 00:00:38,440 --> 00:00:41,680 Speaker 1: on top of this goddamn subject. So check the episode notes, 11 00:00:41,720 --> 00:00:44,760 Speaker 1: buy a challenge coin, read the newsletter. A's a premium 12 00:00:44,840 --> 00:00:47,320 Speaker 1: version unrelated to this show. I'd love if you'd subscribe, 13 00:00:47,320 --> 00:00:50,280 Speaker 1: and if you don't, I won't feel anything. Don't worry 14 00:00:50,280 --> 00:00:52,760 Speaker 1: about it, but let's get to it. Last week open 15 00:00:52,800 --> 00:00:55,680 Speaker 1: Ai launched GPT five, a new flagship model of some 16 00:00:55,760 --> 00:00:58,240 Speaker 1: sort that's allegedly better at coding and writing, but in 17 00:00:58,240 --> 00:01:00,840 Speaker 1: reality is much more of the same. Another model is 18 00:01:00,840 --> 00:01:04,320 Speaker 1: interterminately better at benchmarks built specifically for large language moddles, 19 00:01:04,360 --> 00:01:07,720 Speaker 1: because they can't do actual work. The Wall Street Journal 20 00:01:07,760 --> 00:01:11,080 Speaker 1: reported late last year that it took multiple half billion 21 00:01:11,160 --> 00:01:13,880 Speaker 1: dollar training runs to get GPT five off the ground, 22 00:01:13,959 --> 00:01:16,360 Speaker 1: and Altman him said said in the podcast with theovon 23 00:01:16,440 --> 00:01:19,000 Speaker 1: of all people at GPT five scared him and made 24 00:01:19,080 --> 00:01:21,440 Speaker 1: him say, what have we done? And that's a good 25 00:01:21,480 --> 00:01:24,880 Speaker 1: bloody question, Sammy. According to open Ai, GPT five is 26 00:01:24,920 --> 00:01:27,640 Speaker 1: a unified system with a smart, efficient model that answers 27 00:01:27,680 --> 00:01:30,280 Speaker 1: most questions, a deeper reasoning model, and a real time 28 00:01:30,319 --> 00:01:32,840 Speaker 1: router that quickly decides which model to use based on 29 00:01:32,880 --> 00:01:36,200 Speaker 1: conversation type, complexity, tool needs. And you're explicit in tent. 30 00:01:36,680 --> 00:01:38,240 Speaker 1: I read all of that out because I wanted you 31 00:01:38,280 --> 00:01:40,680 Speaker 1: to hear how convoluted GPT five is and how much 32 00:01:40,720 --> 00:01:43,000 Speaker 1: effort open Ai has had to put in to create 33 00:01:43,040 --> 00:01:46,279 Speaker 1: something that, based on all reports is fine. To quote 34 00:01:46,319 --> 00:01:51,720 Speaker 1: Simon Willison, it just does stuff. Wowie zowie. In simpler terms, 35 00:01:51,800 --> 00:01:54,840 Speaker 1: Chad GPT's version of GPT five takes the user's prompt 36 00:01:54,880 --> 00:01:56,960 Speaker 1: and decides which modele to use as a result, using 37 00:01:56,960 --> 00:01:59,760 Speaker 1: one of a few submodels GPT five Regular, Mini or 38 00:01:59,840 --> 00:02:02,600 Speaker 1: now I know, and then spits out an output. And 39 00:02:02,640 --> 00:02:04,240 Speaker 1: there are rate limits, by the way, so if you 40 00:02:04,320 --> 00:02:06,920 Speaker 1: use it too much you get kicked down to many automatically. 41 00:02:07,160 --> 00:02:09,440 Speaker 1: If you ask it to think about something, it will 42 00:02:09,520 --> 00:02:11,560 Speaker 1: choose to engage the reasoning part of the model. These 43 00:02:11,600 --> 00:02:14,440 Speaker 1: things do not think, by the way. They are probabilistic models, 44 00:02:14,639 --> 00:02:17,520 Speaker 1: so reasoning is kind of like you get a prompt 45 00:02:17,520 --> 00:02:20,600 Speaker 1: and then it reads the prompt with another model and says, Okay, 46 00:02:20,639 --> 00:02:23,320 Speaker 1: what would the steps be to execute this? It has 47 00:02:23,600 --> 00:02:26,640 Speaker 1: some returns, but recent papers suggest that it doesn't work 48 00:02:26,639 --> 00:02:29,480 Speaker 1: that well anyway. Using this, you mostly have to trust 49 00:02:29,480 --> 00:02:31,600 Speaker 1: the open ai will choose the best model for the job, 50 00:02:31,600 --> 00:02:34,200 Speaker 1: as opposed to the cheapest one for open ai to serve, 51 00:02:34,240 --> 00:02:36,680 Speaker 1: which is what I think they're actually doing. As part 52 00:02:36,720 --> 00:02:38,920 Speaker 1: of the launch, open ai has also killed access to 53 00:02:38,919 --> 00:02:41,040 Speaker 1: all other models that at least is planning to, and 54 00:02:41,080 --> 00:02:45,079 Speaker 1: truncated user access to two choices GPT five or GPT 55 00:02:45,120 --> 00:02:48,239 Speaker 1: five thinking with legacy models like four four many and 56 00:02:48,280 --> 00:02:51,640 Speaker 1: their associated rate limits gone immediately for most, though some 57 00:02:51,680 --> 00:02:53,880 Speaker 1: of them are back, and in sixty days of people 58 00:02:53,960 --> 00:02:56,919 Speaker 1: paying two hundred dollars a month, you'll work out why 59 00:02:56,960 --> 00:03:00,800 Speaker 1: I'm dithering in a second. This enraged the chat gp subreddit, 60 00:03:00,840 --> 00:03:03,240 Speaker 1: with users claiming that GPT five was and I quote, 61 00:03:03,280 --> 00:03:05,720 Speaker 1: wearing the skin of their dead friend, referring to GPT 62 00:03:05,760 --> 00:03:08,280 Speaker 1: four oh, and another saying that GPT four point five 63 00:03:08,360 --> 00:03:10,639 Speaker 1: genuinely talk to them and as pathetic as it sounds, 64 00:03:10,720 --> 00:03:13,800 Speaker 1: was their only friend. And it must be clear, we 65 00:03:13,840 --> 00:03:15,840 Speaker 1: can make fun of these people if we want, but 66 00:03:15,960 --> 00:03:19,120 Speaker 1: this is actually genuinely sad. There is something going on 67 00:03:19,240 --> 00:03:21,680 Speaker 1: here where people are so lonely that they want to 68 00:03:21,720 --> 00:03:24,160 Speaker 1: talk to a chat bot. Mock them if you want, 69 00:03:24,840 --> 00:03:26,720 Speaker 1: and some of you will, and I don't know if 70 00:03:26,760 --> 00:03:30,200 Speaker 1: I even want to. But something is happening here and 71 00:03:30,240 --> 00:03:33,120 Speaker 1: it isn't brilliant. But after a few days, Clammy Samuel 72 00:03:33,280 --> 00:03:36,960 Speaker 1: and restored access to GPT four overpaid users, and then 73 00:03:37,120 --> 00:03:39,160 Speaker 1: this only matched the stem the type briefly, with one 74 00:03:39,240 --> 00:03:41,480 Speaker 1: user saying that their baby was back, that they cried 75 00:03:41,520 --> 00:03:43,400 Speaker 1: a lot, and that they were crying as they wrote 76 00:03:43,400 --> 00:03:46,080 Speaker 1: the post, ending by saying love you. I assumed to 77 00:03:46,120 --> 00:03:50,400 Speaker 1: GPT four zero. Here's the problem. Though, users are now 78 00:03:50,440 --> 00:03:52,840 Speaker 1: doubting that the four O that open Ai is restored 79 00:03:52,880 --> 00:03:55,280 Speaker 1: is actually the same model. One post claims that four 80 00:03:55,360 --> 00:03:57,440 Speaker 1: roh has lost its soul, and another says that fourro 81 00:03:57,560 --> 00:04:00,280 Speaker 1: is lobotomized all the way down in one three. One 82 00:04:00,360 --> 00:04:02,920 Speaker 1: user said that four row had gotten markedly worse suddenly, 83 00:04:03,120 --> 00:04:05,360 Speaker 1: and another said it's definitely not the same, though others 84 00:04:05,360 --> 00:04:07,880 Speaker 1: in the thread claimed that it was. Another post said 85 00:04:07,960 --> 00:04:11,240 Speaker 1: legacy GPT four row is GPT five in cosplay, even 86 00:04:11,240 --> 00:04:13,760 Speaker 1: as others pleaded with them and said it was exactly 87 00:04:13,760 --> 00:04:16,640 Speaker 1: the same and that people were experiencing some strange phantom 88 00:04:16,680 --> 00:04:19,720 Speaker 1: love for placebo effect. And I think that's actually what's 89 00:04:19,800 --> 00:04:22,920 Speaker 1: going on, writ large. Chat GPT was never a success 90 00:04:22,960 --> 00:04:26,159 Speaker 1: based on its actual abilities or outputs or things it 91 00:04:26,200 --> 00:04:28,839 Speaker 1: could do, but a global marketing campaign perpetuated by it 92 00:04:28,880 --> 00:04:31,640 Speaker 1: taking business media asleep at the wheel, or worse still, 93 00:04:31,960 --> 00:04:35,000 Speaker 1: that wanted these companies to win and help them by lying. 94 00:04:35,160 --> 00:04:37,960 Speaker 1: I have done a comprehensive evaluation of the last three 95 00:04:38,040 --> 00:04:41,440 Speaker 1: years of press around chat, GPT and GPT itself, and 96 00:04:41,520 --> 00:04:43,400 Speaker 1: you will look in the things in twenty twenty three 97 00:04:43,400 --> 00:04:45,680 Speaker 1: and there's shit that's just fucking made up. There's a 98 00:04:45,720 --> 00:04:49,200 Speaker 1: whole go and look up task Grabbit GPT four. There 99 00:04:49,200 --> 00:04:51,280 Speaker 1: are so many I'm a link one in the fucking notes. 100 00:04:52,080 --> 00:04:55,039 Speaker 1: There are people that claim that GPT four ordered a 101 00:04:55,160 --> 00:04:58,039 Speaker 1: task grab it to complete a capture. Now, on top 102 00:04:58,080 --> 00:04:59,960 Speaker 1: of this not being a thing that task grabbit generally 103 00:05:00,000 --> 00:05:02,799 Speaker 1: it does. This is from the system card of GPT 104 00:05:03,000 --> 00:05:07,240 Speaker 1: four and it claims that it hired a taskrab, except 105 00:05:07,240 --> 00:05:08,920 Speaker 1: when you look at just said it messaged them. It's 106 00:05:09,040 --> 00:05:11,880 Speaker 1: very clearly made up. But everyone reported it as agents 107 00:05:11,960 --> 00:05:15,520 Speaker 1: existing in twenty twenty three. Aha. Every time I read 108 00:05:15,560 --> 00:05:19,039 Speaker 1: this stuff, I feel little goddamn insane. But anyway, because 109 00:05:19,080 --> 00:05:22,120 Speaker 1: these models do not have obvious replicable ways outside of 110 00:05:22,200 --> 00:05:24,800 Speaker 1: benchmarks of testing what they can do, each user is 111 00:05:24,800 --> 00:05:27,159 Speaker 1: effectively in a constant vibe check with the models, and 112 00:05:27,160 --> 00:05:29,920 Speaker 1: the sycophantic qualities of GPT four oh were clearly enough 113 00:05:29,920 --> 00:05:35,920 Speaker 1: to endear them to the platform. People using GPT four Oh, 114 00:05:36,080 --> 00:05:38,479 Speaker 1: they couldn't tell you why it's different to GPT five 115 00:05:38,520 --> 00:05:41,000 Speaker 1: other than it feels less human or doesn't do the 116 00:05:41,040 --> 00:05:43,320 Speaker 1: same things, even if those things are kind of hard 117 00:05:43,360 --> 00:05:46,159 Speaker 1: to define. This is what happens when you build a 118 00:05:46,200 --> 00:05:48,920 Speaker 1: following for a product based on specious hype and vague 119 00:05:48,960 --> 00:05:51,839 Speaker 1: promises and lines of inference, of course, and then allow 120 00:05:52,080 --> 00:05:54,279 Speaker 1: users to make up the reasons that they care. You 121 00:05:54,360 --> 00:05:57,159 Speaker 1: begin engaging with the gamer mindset of vibes based fandom 122 00:05:57,200 --> 00:05:59,920 Speaker 1: that's completely unbreakable unless you make one subtle change that 123 00:06:00,040 --> 00:06:02,480 Speaker 1: you could never see coming that breaks the illusion, leading 124 00:06:02,480 --> 00:06:05,839 Speaker 1: to gamer like distrust and anger. You see, I theorized 125 00:06:05,839 --> 00:06:08,200 Speaker 1: that the vast majority of chat GPT users do not 126 00:06:08,360 --> 00:06:10,440 Speaker 1: know where they used it in the first place. Three 127 00:06:10,520 --> 00:06:12,800 Speaker 1: years of immediate pressure to use AI that AI was 128 00:06:12,839 --> 00:06:15,600 Speaker 1: the future, their boss is saying AI is important and 129 00:06:15,640 --> 00:06:17,720 Speaker 1: that you would be left behind if you didn't use AI. 130 00:06:17,839 --> 00:06:19,880 Speaker 1: Mean that people come to chat GPT to work out 131 00:06:19,920 --> 00:06:21,680 Speaker 1: why they're using it in the first place, which has 132 00:06:21,760 --> 00:06:24,760 Speaker 1: led to all sorts of bizarre emotional attachments, kind of 133 00:06:24,800 --> 00:06:27,080 Speaker 1: like one's attachments to a live service game and shout 134 00:06:27,080 --> 00:06:29,440 Speaker 1: out to Catharsis twenty three on Blue Sky, who made 135 00:06:29,440 --> 00:06:33,240 Speaker 1: this observation. As a result, users were incredibly sensitive to 136 00:06:33,320 --> 00:06:36,120 Speaker 1: changes like removing or changing a model because their association 137 00:06:36,200 --> 00:06:38,760 Speaker 1: with chat GPT was based on However, GPT four zero 138 00:06:38,800 --> 00:06:41,680 Speaker 1: works and sounds by ripping it out and replacing it 139 00:06:41,680 --> 00:06:45,080 Speaker 1: with GPT five, users immediately felt jilted and swindled by 140 00:06:45,120 --> 00:06:47,800 Speaker 1: open AI, and much like a dying live service game, 141 00:06:47,839 --> 00:06:50,240 Speaker 1: any changes that have been made as a result were 142 00:06:50,240 --> 00:06:54,000 Speaker 1: met with paranoia and confusion. Clamiel Altman's attempts to paper 143 00:06:54,000 --> 00:06:56,560 Speaker 1: over the problem by boosting rate limits on GPT five's 144 00:06:56,600 --> 00:06:59,159 Speaker 1: thinking and restoring access to GPT four to ZH and 145 00:06:59,200 --> 00:07:01,799 Speaker 1: other models for pay maid users were not enough because, 146 00:07:01,800 --> 00:07:03,960 Speaker 1: in a very real sense, many of those users could 147 00:07:04,000 --> 00:07:05,440 Speaker 1: not tell you what they liked four to oh to 148 00:07:05,480 --> 00:07:08,480 Speaker 1: begin with four oh wasn't good so much as it 149 00:07:08,520 --> 00:07:11,360 Speaker 1: was an investment of time. By showing that open ai 150 00:07:11,520 --> 00:07:14,160 Speaker 1: is willing to cut things arbitrarily, users can no longer 151 00:07:14,160 --> 00:07:17,000 Speaker 1: trust that this investment of time is worthy, especially as 152 00:07:17,040 --> 00:07:19,680 Speaker 1: many complain that it launch GPD five deleted a bunch 153 00:07:19,680 --> 00:07:22,880 Speaker 1: of conversations. Now open ai sits in an odd spot 154 00:07:22,920 --> 00:07:27,120 Speaker 1: where they're supposedly huge Manhattan Project level launch has been 155 00:07:27,160 --> 00:07:30,559 Speaker 1: met with either apathy or agony. While they've plicated users 156 00:07:30,560 --> 00:07:32,720 Speaker 1: in the short term, it's very clear that the vast 157 00:07:32,720 --> 00:07:35,600 Speaker 1: majority of users this, like GPT five and power users, 158 00:07:35,640 --> 00:07:38,520 Speaker 1: don't seem particularly impressed with it either. This was meant 159 00:07:38,560 --> 00:07:40,400 Speaker 1: to be the big launch that changed things for open 160 00:07:40,400 --> 00:07:42,560 Speaker 1: ai forever, but it's turned into something of a mass 161 00:07:42,600 --> 00:07:45,400 Speaker 1: betrayal or just kind of a mass letdown. And because 162 00:07:45,400 --> 00:07:47,720 Speaker 1: it's based on vibes rather than its actual ability to 163 00:07:47,760 --> 00:07:50,080 Speaker 1: do something, there's very little one could do to fix 164 00:07:50,160 --> 00:07:53,560 Speaker 1: this problem. It's unclear how all of this affects the 165 00:07:53,560 --> 00:07:56,440 Speaker 1: company long term, but things do not seem good. Sam 166 00:07:56,480 --> 00:07:58,640 Speaker 1: Wortman has already said that open ai is having to 167 00:07:58,680 --> 00:08:02,080 Speaker 1: reallocate capacity for the couple of months prioritizing paying Chat 168 00:08:02,120 --> 00:08:05,119 Speaker 1: GPT users of an API demand up from the current 169 00:08:05,160 --> 00:08:08,360 Speaker 1: allocated capacity and commitments that they've made to their customers 170 00:08:08,400 --> 00:08:11,080 Speaker 1: code for those who do not want to pay priority processing, 171 00:08:11,320 --> 00:08:14,800 Speaker 1: which is now available for any developer. It's also unclear 172 00:08:14,840 --> 00:08:18,240 Speaker 1: what happens next. GPT five is not the future. Open 173 00:08:18,280 --> 00:08:20,520 Speaker 1: ai is running out of capacity, and their product, despite 174 00:08:20,560 --> 00:08:23,280 Speaker 1: the fanfare, has no capabilities or reasons to adopt it 175 00:08:23,560 --> 00:08:27,160 Speaker 1: that are really new or interesting. Years of allowing the 176 00:08:27,160 --> 00:08:30,200 Speaker 1: media to spin out ridiculous narratives about what AI can 177 00:08:30,320 --> 00:08:33,400 Speaker 1: or could do using vague pablem that kind of suggests 178 00:08:33,400 --> 00:08:35,679 Speaker 1: they're more powerful than they are has created a pr 179 00:08:35,760 --> 00:08:38,200 Speaker 1: campaign for a product that does not exist, and the 180 00:08:38,240 --> 00:08:41,320 Speaker 1: seven hundred million weekly active users of chat GPT have 181 00:08:41,400 --> 00:08:44,880 Speaker 1: clearly arrived there without much guidance. The attachment born of 182 00:08:44,920 --> 00:08:48,600 Speaker 1: compulsion and societal pressure rather than any real use cases. 183 00:08:49,360 --> 00:08:52,200 Speaker 1: When you allow people to define an indeterminately powerful tool 184 00:08:52,200 --> 00:08:54,679 Speaker 1: by any standard they like, with no interest in correcting them, 185 00:08:54,920 --> 00:08:57,520 Speaker 1: with no interest in guiding them, with no interest in 186 00:08:57,559 --> 00:08:59,600 Speaker 1: actually showing them what it was that they were paying 187 00:08:59,640 --> 00:09:03,040 Speaker 1: for others. Then it can generate staff, You'll create an 188 00:09:03,040 --> 00:09:05,600 Speaker 1: attachment to it that devis any real ability you have 189 00:09:05,640 --> 00:09:09,000 Speaker 1: to control things. Open ai was never forced to productize 190 00:09:09,040 --> 00:09:11,880 Speaker 1: and at scale. It's a very real possibility that people 191 00:09:11,880 --> 00:09:14,840 Speaker 1: have pressure by the media and society itself force themselves 192 00:09:14,840 --> 00:09:18,480 Speaker 1: to find meaning in LMS somehow, even if it feels 193 00:09:18,559 --> 00:09:20,600 Speaker 1: kind of stupid. And what I mean by this is 194 00:09:20,840 --> 00:09:22,640 Speaker 1: you get to the product and there really isn't that 195 00:09:22,679 --> 00:09:25,360 Speaker 1: much guidance. Go on an open AI's website and have 196 00:09:25,400 --> 00:09:27,360 Speaker 1: a look at the chat GPT page and look at 197 00:09:27,360 --> 00:09:29,960 Speaker 1: what it tells you to do. It's quite vague. You 198 00:09:30,000 --> 00:09:33,200 Speaker 1: can look at it can analyze data, right, it can 199 00:09:33,600 --> 00:09:37,120 Speaker 1: generate stuff. It helps you do ideas. Is that good? 200 00:09:37,360 --> 00:09:40,719 Speaker 1: Am I smart for using this? Everything else you look 201 00:09:40,720 --> 00:09:42,640 Speaker 1: at in the software world will tell you what it 202 00:09:42,679 --> 00:09:45,720 Speaker 1: is you're using it for any consumer driven software at 203 00:09:45,720 --> 00:09:49,839 Speaker 1: the very least. Yeah, chat GPT's never had to because 204 00:09:49,840 --> 00:09:52,080 Speaker 1: the media for three years or two years, I guess 205 00:09:52,080 --> 00:09:55,600 Speaker 1: with chat GPT has kind of just sat there doing 206 00:09:55,600 --> 00:09:57,960 Speaker 1: the work for them, telling them, oh, yeah, you can 207 00:09:58,600 --> 00:10:01,440 Speaker 1: use it as a powerful person a system assistant. How 208 00:10:01,679 --> 00:10:05,240 Speaker 1: to assist me with what? Nasty? Kevin Russe creepy Kevin 209 00:10:05,840 --> 00:10:07,920 Speaker 1: in the New York Times a few weeks ago, when 210 00:10:07,960 --> 00:10:12,080 Speaker 1: he did Everyone's Using AI piece with Casey Newton, he 211 00:10:12,200 --> 00:10:15,200 Speaker 1: said that it's a powerful assistant. It's like, for what 212 00:10:15,880 --> 00:10:18,760 Speaker 1: you can't do that? This shit can't control my calendar. 213 00:10:18,960 --> 00:10:21,240 Speaker 1: I don't want to touch in my emails. I don't 214 00:10:21,240 --> 00:10:23,800 Speaker 1: think most people do either. So you've just got people 215 00:10:23,800 --> 00:10:26,000 Speaker 1: to use it as a shit, our search engine, an 216 00:10:26,040 --> 00:10:30,040 Speaker 1: online companion, and a brainstorming thing, which is a natural 217 00:10:30,080 --> 00:10:32,760 Speaker 1: way to get people kind of addicted, but addicted to 218 00:10:32,760 --> 00:10:35,360 Speaker 1: a product you don't truly control and a product you 219 00:10:35,400 --> 00:10:38,200 Speaker 1: don't truly understand, one that can be swiped by just 220 00:10:38,240 --> 00:10:42,800 Speaker 1: about anyone. I actually think we're on the the kind 221 00:10:42,840 --> 00:10:46,120 Speaker 1: of downward spiral for this shit. I am winking like 222 00:10:46,200 --> 00:10:49,800 Speaker 1: a pig and squawking like a bird watching this happen. 223 00:10:49,880 --> 00:10:52,520 Speaker 1: Even core Weavers crashing as I record this. They're down 224 00:10:52,600 --> 00:10:55,520 Speaker 1: seventeen point nine one percent. Still up too much, though, 225 00:10:56,520 --> 00:10:58,599 Speaker 1: And I really do think something has shifted thanks to 226 00:10:58,679 --> 00:11:02,280 Speaker 1: GPT five, I'm a excited and like Doctor Stones once said, 227 00:11:02,400 --> 00:11:05,160 Speaker 1: get excited because I think the next few months, like 228 00:11:05,280 --> 00:11:09,560 Speaker 1: next year, is going to be chuckle heavy's. We're going 229 00:11:09,640 --> 00:11:12,520 Speaker 1: to be whimsy pilled as we go through the remainder 230 00:11:12,640 --> 00:11:14,640 Speaker 1: of the AI Boom, and I'll be here to guide 231 00:11:14,679 --> 00:11:16,280 Speaker 1: you through it. Thanks for listening.