1 00:00:02,800 --> 00:00:07,480 Speaker 1: Zone Media. Welcome to Better Offline. I'm your host ed 2 00:00:07,560 --> 00:00:22,079 Speaker 1: zitrun as Eva. Check out the episode notes for merchandise, 3 00:00:22,200 --> 00:00:25,680 Speaker 1: my newsletter, various links, things that I just write for myself. 4 00:00:25,880 --> 00:00:28,080 Speaker 1: I don't actually leave little notes, but maybe I should. 5 00:00:28,560 --> 00:00:31,000 Speaker 1: But as promised in my last monologue, we've got a 6 00:00:31,000 --> 00:00:34,040 Speaker 1: three part episode this week, and we're talking about AI boosters, 7 00:00:34,080 --> 00:00:36,800 Speaker 1: the people who insist despite very little proof and all 8 00:00:36,840 --> 00:00:39,680 Speaker 1: evidence of the contrary. The generative AI is the future, 9 00:00:39,960 --> 00:00:42,800 Speaker 1: open AI and anthropic of perfectly healthy companies. And what 10 00:00:42,840 --> 00:00:45,760 Speaker 1: we're witnessing isn't an inflation of a massive, dangerous bubble 11 00:00:45,800 --> 00:00:48,800 Speaker 1: that might eat our entire economy, but rather the emergence 12 00:00:48,800 --> 00:00:51,040 Speaker 1: of a brand new paradigm in tech that's just as 13 00:00:51,080 --> 00:00:55,000 Speaker 1: meaningful as the smartphone and cloud computing revolutions. I even 14 00:00:55,120 --> 00:00:57,600 Speaker 1: saying that out loud makes me feel a little crazy, 15 00:00:58,080 --> 00:01:00,560 Speaker 1: but yeah, that's what they believe. These people exist. They're 16 00:01:00,560 --> 00:01:03,680 Speaker 1: often wrong. In fact, they're mostly wrong, and they're also 17 00:01:03,760 --> 00:01:06,800 Speaker 1: really really annoying. And so as a public service, I'm 18 00:01:06,800 --> 00:01:09,080 Speaker 1: going to spend the next three episodes explaining how to 19 00:01:09,120 --> 00:01:11,640 Speaker 1: talk with them without losing your sanity, will to live, 20 00:01:11,720 --> 00:01:14,840 Speaker 1: or your ability to operate in society, and now what 21 00:01:15,000 --> 00:01:17,240 Speaker 1: makes me qualified to do this well? In the last 22 00:01:17,319 --> 00:01:19,880 Speaker 1: two years, I've written no less than half a million words, 23 00:01:19,920 --> 00:01:22,560 Speaker 1: with many of them dedicated to breaking both existent and 24 00:01:22,600 --> 00:01:25,160 Speaker 1: previous myths about the state of technology and the tech 25 00:01:25,200 --> 00:01:29,400 Speaker 1: industry itself. Now let's start with something every critic is experienced, 26 00:01:29,480 --> 00:01:32,680 Speaker 1: the massive double standard between those perceived as skeptics and 27 00:01:32,720 --> 00:01:36,480 Speaker 1: those who are optimistic or optimists. To be skeptical of 28 00:01:36,520 --> 00:01:39,120 Speaker 1: ais to commit yourself to a near constant amount of 29 00:01:39,160 --> 00:01:42,000 Speaker 1: demands to prove yourself and endless nags of But what 30 00:01:42,200 --> 00:01:44,920 Speaker 1: about with each one, no matter how small, presented as 31 00:01:44,959 --> 00:01:47,319 Speaker 1: a fact that defeats any points you may have had, 32 00:01:47,800 --> 00:01:50,920 Speaker 1: have or will have in the future. Conversely, being an 33 00:01:50,920 --> 00:01:53,760 Speaker 1: optimist allows you to take things like AI twenty twenty seven, 34 00:01:53,800 --> 00:01:56,360 Speaker 1: which I will fucking get to seriously, to the point 35 00:01:56,360 --> 00:01:59,080 Speaker 1: that you can write an entire feature length fan fiction 36 00:01:59,200 --> 00:02:01,600 Speaker 1: piece in the New York Times and nobody will bat 37 00:02:01,600 --> 00:02:04,760 Speaker 1: an eye lead. In any case, things are beginning to 38 00:02:04,760 --> 00:02:07,800 Speaker 1: fall UPALM two of the actual reporters, the real ones 39 00:02:07,840 --> 00:02:10,120 Speaker 1: at the New York Times rather than column nests, suit 40 00:02:10,520 --> 00:02:13,760 Speaker 1: kind of act like adult babies. Reported in late August 41 00:02:13,800 --> 00:02:16,560 Speaker 1: that Meta is yet again restructuring its AI department for 42 00:02:16,600 --> 00:02:19,880 Speaker 1: the fourth time, and that it's considering downsizing their overall 43 00:02:19,919 --> 00:02:22,760 Speaker 1: AI division, which sure doesn't seem like something you'd do 44 00:02:22,840 --> 00:02:25,640 Speaker 1: if you thought AI was the future. Meanwhile, the markets 45 00:02:25,639 --> 00:02:28,680 Speaker 1: are thoroughly spooked by an MIT study covered by Fortune 46 00:02:28,720 --> 00:02:31,440 Speaker 1: that found that ninety five percent of generative AI pilots 47 00:02:31,440 --> 00:02:34,480 Speaker 1: and companies are failing in providing no ROI. And though 48 00:02:34,600 --> 00:02:37,000 Speaker 1: MIT Nanda has now replaced the link to the study 49 00:02:37,280 --> 00:02:40,160 Speaker 1: with some sort of Google form to request access, you 50 00:02:40,200 --> 00:02:42,639 Speaker 1: can find the full pdf linked in the show notes. 51 00:02:42,840 --> 00:02:44,120 Speaker 1: By the way, this is the kind of thing that 52 00:02:44,280 --> 00:02:46,560 Speaker 1: is a PR firm wanting to try and set up interviews, 53 00:02:46,600 --> 00:02:48,520 Speaker 1: not for me, thank you, just like to read the 54 00:02:48,520 --> 00:02:51,280 Speaker 1: thing that you put out very rude. In any case, 55 00:02:51,320 --> 00:02:53,800 Speaker 1: the report is actually grimmer than Fortune made its sound, 56 00:02:53,880 --> 00:02:56,519 Speaker 1: saying that ninety five percent of organizations are getting zero 57 00:02:56,600 --> 00:02:59,880 Speaker 1: return on GENERATIVAI. The report says that adoption is high, 58 00:03:00,080 --> 00:03:03,160 Speaker 1: transformation is low, adding that few industries show the deep 59 00:03:03,200 --> 00:03:06,880 Speaker 1: structural shifts associated with past general purpose technologies, such as 60 00:03:06,960 --> 00:03:10,440 Speaker 1: new market leaders, disrupted business models, or measurable changes in 61 00:03:10,480 --> 00:03:13,200 Speaker 1: customer behavior, and saying that out loud. That is right, 62 00:03:13,280 --> 00:03:17,040 Speaker 1: what is being disrupted by AI? Search? Kind of that 63 00:03:17,120 --> 00:03:20,280 Speaker 1: just means that search sucks anyway. Yet the most damning 64 00:03:20,360 --> 00:03:22,959 Speaker 1: part was there was a part called the five myths 65 00:03:22,960 --> 00:03:26,160 Speaker 1: about AI in the Enterprise, which is probably the most 66 00:03:26,200 --> 00:03:28,960 Speaker 1: wilting takedown of this movement I've ever seen. And I'm 67 00:03:28,960 --> 00:03:31,840 Speaker 1: going to quote an adverbatum okay number one. And they 68 00:03:32,000 --> 00:03:35,240 Speaker 1: present these as the first statement is something that isn't 69 00:03:35,280 --> 00:03:38,160 Speaker 1: going to happen, or they're questioning AI will replace most 70 00:03:38,240 --> 00:03:40,760 Speaker 1: jobs in the next few years. What it actually says 71 00:03:40,840 --> 00:03:43,720 Speaker 1: is research found limited layoffs from JENAI, and only in 72 00:03:43,720 --> 00:03:47,280 Speaker 1: industries that are already affected significantly by AI. There is 73 00:03:47,320 --> 00:03:51,080 Speaker 1: no consensus among executives as the hiring levels over the 74 00:03:51,160 --> 00:03:54,560 Speaker 1: next three to five years. Another statement they challenge generative 75 00:03:54,600 --> 00:03:58,000 Speaker 1: AI is transforming business. No, no, No, adoption is high, but 76 00:03:58,040 --> 00:04:01,280 Speaker 1: transformation is rare. Only five percent of enterprises have AI 77 00:04:01,360 --> 00:04:04,880 Speaker 1: tools integrated into workflows at scale, and seven of nine 78 00:04:04,920 --> 00:04:07,560 Speaker 1: sectors show no real structural change. And I want to 79 00:04:07,600 --> 00:04:09,760 Speaker 1: thank them for putting this there. I made this exact 80 00:04:09,800 --> 00:04:12,480 Speaker 1: point in February in my newsletter. There's no AI revolution. 81 00:04:12,880 --> 00:04:16,000 Speaker 1: Links in the show notes as ever, enterprises are slow 82 00:04:16,040 --> 00:04:18,720 Speaker 1: in adopting new tech is another commonly held thing, and 83 00:04:18,800 --> 00:04:22,120 Speaker 1: they say enterprises are extremely eager to adopt AI, and 84 00:04:22,200 --> 00:04:26,159 Speaker 1: ninety percent of seriously explored buying an AI solution. Another 85 00:04:26,200 --> 00:04:28,839 Speaker 1: statement they challenge, the biggest thing holding back AI is 86 00:04:28,880 --> 00:04:33,520 Speaker 1: model quality, legal data, and risk. And then they say, 87 00:04:33,560 --> 00:04:37,280 Speaker 1: which is my favorite quot, what's really holding it back 88 00:04:37,360 --> 00:04:39,600 Speaker 1: is that most AI tools don't learn and don't integrate 89 00:04:39,600 --> 00:04:43,800 Speaker 1: well into workflows. I really do love that. I love 90 00:04:43,839 --> 00:04:46,400 Speaker 1: that so much because it's like the thing holding AI 91 00:04:46,480 --> 00:04:49,320 Speaker 1: back is that it sucks. And the final one, the 92 00:04:49,320 --> 00:04:52,120 Speaker 1: final challenge statement they make is the best enterprises are 93 00:04:52,120 --> 00:04:55,479 Speaker 1: building their own tools. Internal builds fail twice as often, 94 00:04:56,320 --> 00:04:59,239 Speaker 1: and that's twice as often in a thing where ninety 95 00:04:59,240 --> 00:05:03,760 Speaker 1: five percent anyway, These are brutal, dispassionate points that directly 96 00:05:03,760 --> 00:05:07,719 Speaker 1: deal with the most common boosterisms. Generative AI isn't transforming anything. 97 00:05:07,800 --> 00:05:11,520 Speaker 1: AI isn't replacing anyone. Enterprises are trying to adopt generative AI, 98 00:05:11,760 --> 00:05:13,920 Speaker 1: but it doesn't fucking work. And the thing holding back 99 00:05:13,960 --> 00:05:16,560 Speaker 1: AI is the fact it doesn't fucking work. This isn't 100 00:05:16,560 --> 00:05:18,880 Speaker 1: a case where the enterprise is suddenly going to save 101 00:05:18,960 --> 00:05:21,840 Speaker 1: these companies because the enterprise has already tried and it 102 00:05:21,920 --> 00:05:24,520 Speaker 1: is not working. And what's hilarious as well is so 103 00:05:24,560 --> 00:05:27,640 Speaker 1: many people say, well, there's just this enterprise opportunity they 104 00:05:27,640 --> 00:05:31,120 Speaker 1: haven't got into yet. Actually they have, They've got into 105 00:05:31,160 --> 00:05:34,240 Speaker 1: the enterprise. They're sticking their fingers and all the bits 106 00:05:34,279 --> 00:05:36,640 Speaker 1: of the sas and it isn't working. And the enterprise 107 00:05:36,720 --> 00:05:39,000 Speaker 1: loves money. The enterprise will do even if it's a 108 00:05:39,279 --> 00:05:42,799 Speaker 1: genuinely evil product. They will push it, Like look at Salesforce, 109 00:05:42,839 --> 00:05:46,640 Speaker 1: an entire company based on them. Anyway, an incorrect read 110 00:05:46,680 --> 00:05:49,400 Speaker 1: of the study that's been going around is that the 111 00:05:49,480 --> 00:05:53,000 Speaker 1: learning gap that makes these things less useful is because 112 00:05:53,000 --> 00:05:56,239 Speaker 1: of human issues like human error, when the study actually 113 00:05:56,279 --> 00:05:59,279 Speaker 1: says that the fundamental gap that defines the GENAI divide 114 00:05:59,320 --> 00:06:02,680 Speaker 1: is that users tools that don't adapt model quality fills 115 00:06:02,680 --> 00:06:05,800 Speaker 1: without context, and UX suffers when systems can't remember this 116 00:06:05,920 --> 00:06:08,440 Speaker 1: isn't something you as a person learn your way out of. 117 00:06:08,480 --> 00:06:10,440 Speaker 1: These products don't do what they're meant to do, and 118 00:06:10,480 --> 00:06:13,880 Speaker 1: people are realizing it when they try and use them. Nevertheless, 119 00:06:13,880 --> 00:06:15,880 Speaker 1: boosters will still find a way to twist this study 120 00:06:15,880 --> 00:06:18,360 Speaker 1: to mean something else. They'll claim that AI is still early, 121 00:06:18,480 --> 00:06:21,000 Speaker 1: that the opportunity is still there, that we didn't confirm 122 00:06:21,040 --> 00:06:25,400 Speaker 1: that the Internet and smartphones with productivity boosting, or that 123 00:06:25,440 --> 00:06:28,119 Speaker 1: we're in the early days of AI somehow three years 124 00:06:28,160 --> 00:06:31,279 Speaker 1: and hundreds of billions of dollars and thousands of articles 125 00:06:31,279 --> 00:06:33,640 Speaker 1: in Yeah, we're in the early days. We're in the 126 00:06:33,640 --> 00:06:35,600 Speaker 1: early days when all the money in all the land 127 00:06:35,600 --> 00:06:38,560 Speaker 1: has gone into this. And I'm tired. I'm tired of 128 00:06:38,600 --> 00:06:41,160 Speaker 1: having the same arguments with these people, and frankly, I'm 129 00:06:41,200 --> 00:06:44,120 Speaker 1: sure you are too. No matter how blindly obvious evidence 130 00:06:44,920 --> 00:06:47,279 Speaker 1: is to the contrary, they will find ways to ignore it. 131 00:06:47,400 --> 00:06:50,440 Speaker 1: They continually make these smug comments about people wishing things 132 00:06:50,480 --> 00:06:53,400 Speaker 1: would be bad or suggesting you are stupid, and yes, 133 00:06:53,440 --> 00:06:55,560 Speaker 1: that is their belief, by the way, for not believing 134 00:06:55,600 --> 00:06:58,920 Speaker 1: that generative AI is disruptive today. And in this three 135 00:06:59,000 --> 00:07:00,960 Speaker 1: part tern I'm going to give you the tools to 136 00:07:01,000 --> 00:07:03,599 Speaker 1: fight back against the AI boosters in your life. I'm 137 00:07:03,640 --> 00:07:06,280 Speaker 1: going to go into the generalities of the booster movement, 138 00:07:06,600 --> 00:07:08,600 Speaker 1: the way they argue, the tropes they cling to, and 139 00:07:08,640 --> 00:07:10,560 Speaker 1: the ways in which they use your own self doubt 140 00:07:10,560 --> 00:07:13,160 Speaker 1: against you. They're your buddy, your boss, a man in 141 00:07:13,200 --> 00:07:16,200 Speaker 1: a Ginghamshire Epic steakhouse who won't leave you the fuck alone. 142 00:07:16,320 --> 00:07:19,520 Speaker 1: A editor, a writer, a founderer, or just a common 143 00:07:19,640 --> 00:07:22,680 Speaker 1: or garden conn artist. Whoever the booster in your life is, 144 00:07:22,880 --> 00:07:24,960 Speaker 1: I want you to have the words to fight them with. 145 00:07:35,200 --> 00:07:37,720 Speaker 1: So an AI booster is not, in many cases, an 146 00:07:37,760 --> 00:07:42,120 Speaker 1: actual fan of artificial intelligence. People like Simon Willison or 147 00:07:42,160 --> 00:07:44,480 Speaker 1: Max Wolf who actually work with LMS on a daily 148 00:07:44,520 --> 00:07:47,920 Speaker 1: basis don't see the need to repeatedly harass everybody or 149 00:07:47,960 --> 00:07:50,360 Speaker 1: talk down to them about their unwillingness to pledge allegiance 150 00:07:50,400 --> 00:07:53,440 Speaker 1: to the graveyard smash of generative AI. In fact, the 151 00:07:53,480 --> 00:07:56,400 Speaker 1: closer I've found someone to actually building things with LLMS, 152 00:07:56,560 --> 00:07:59,000 Speaker 1: the less likely they are to emphatically argue that I'm 153 00:07:59,040 --> 00:08:03,320 Speaker 1: missing out on something by not doing so myself. No, no, no. 154 00:08:03,400 --> 00:08:07,080 Speaker 1: The AI booster is symbolically aligned with generative AI. They're 155 00:08:07,080 --> 00:08:08,960 Speaker 1: fans in the same way that somebody is a fan 156 00:08:09,000 --> 00:08:11,400 Speaker 1: of a sports team. Their houses in blazons with every 157 00:08:11,440 --> 00:08:14,520 Speaker 1: possible piece of tat and crap they can find. They're Sundays, 158 00:08:14,520 --> 00:08:17,040 Speaker 1: living and dying by the successes of the team. Except 159 00:08:17,080 --> 00:08:19,520 Speaker 1: even fans of the Dallas Cowboys of a tighter grasp 160 00:08:19,560 --> 00:08:24,040 Speaker 1: on reality, but not Micah Parsons anyway. Kevin Rus and 161 00:08:24,120 --> 00:08:26,480 Speaker 1: Casey Newton are two of the most notable boosters, and 162 00:08:26,520 --> 00:08:29,680 Speaker 1: as I'll get to later in this series, neither of 163 00:08:29,720 --> 00:08:32,560 Speaker 1: them have a consistent or comprehensive knowledge of AI, despite 164 00:08:32,679 --> 00:08:35,520 Speaker 1: being at The New York Times. Though Casey Newton is 165 00:08:35,520 --> 00:08:38,559 Speaker 1: a contractor, he is a contractor just for a podcast, 166 00:08:38,600 --> 00:08:43,880 Speaker 1: which I can't insult due to my own contractual relationships. Nevertheless, 167 00:08:43,920 --> 00:08:46,560 Speaker 1: they will insist that everybody is using AI for everything, 168 00:08:46,600 --> 00:08:48,600 Speaker 1: which is the title of an article they put out, 169 00:08:48,800 --> 00:08:51,280 Speaker 1: a statement that even a booster should realize is incorrect 170 00:08:51,280 --> 00:08:54,440 Speaker 1: based on the actual abilities of the models. But that's 171 00:08:54,480 --> 00:08:57,480 Speaker 1: because it isn't about what's happening. It's not about what's 172 00:08:57,520 --> 00:09:02,240 Speaker 1: actually happening. It's about allegiance. AI symbolizes something to the 173 00:09:02,280 --> 00:09:05,080 Speaker 1: AI booster, a way that they're better than other people. 174 00:09:05,120 --> 00:09:08,719 Speaker 1: That makes them superior because they're, unlike cynics and skeptics, 175 00:09:08,760 --> 00:09:12,079 Speaker 1: able to see the incredible potential in the future of AI, 176 00:09:12,200 --> 00:09:14,559 Speaker 1: but also how great it is today, though it never 177 00:09:14,600 --> 00:09:16,760 Speaker 1: seemed to be able to explain why it is other 178 00:09:16,800 --> 00:09:19,040 Speaker 1: than it replaced search for me and I use it 179 00:09:19,040 --> 00:09:21,720 Speaker 1: to draw connections between articles I write, which is something 180 00:09:21,800 --> 00:09:24,400 Speaker 1: I do for free without AI with my fucking brain. 181 00:09:25,240 --> 00:09:28,560 Speaker 1: Boosterism is a kind of religion interested in finding symbolic 182 00:09:28,640 --> 00:09:31,680 Speaker 1: proof that things are getting better in some indeterminate way, 183 00:09:31,880 --> 00:09:34,280 Speaker 1: and that anyone that chooses to believe otherwise is ignorant 184 00:09:34,320 --> 00:09:38,080 Speaker 1: or stupid or I actually don't know what it is 185 00:09:38,120 --> 00:09:40,240 Speaker 1: that they're meant to be missing. Let me give you 186 00:09:40,240 --> 00:09:44,080 Speaker 1: an example. Thomas Potastic. He wrote a piece called my 187 00:09:44,160 --> 00:09:46,840 Speaker 1: ais skeptic friends are all nuts, and it was catnip 188 00:09:46,920 --> 00:09:50,120 Speaker 1: for boosters. A software engineer using technical terms like interact 189 00:09:50,200 --> 00:09:52,880 Speaker 1: with GIT and MCP, vague chants, and of course an 190 00:09:52,880 --> 00:09:56,000 Speaker 1: extremely vague statement that says hallucinations aren't a problem, and 191 00:09:56,040 --> 00:09:58,760 Speaker 1: I quote now, I'm sure there are still environments where 192 00:09:58,800 --> 00:10:02,280 Speaker 1: hallucination matters. But hallucination is the first thing developers bring 193 00:10:02,360 --> 00:10:05,240 Speaker 1: up when somebody suggests using llms, despite it being more 194 00:10:05,360 --> 00:10:11,000 Speaker 1: or less a solved problem. Is it anyway? My favorite 195 00:10:11,000 --> 00:10:13,520 Speaker 1: favorite part though, let me quote this. A lot of 196 00:10:13,640 --> 00:10:17,720 Speaker 1: LLM skepticism probably isn't really about llms. It's projection. People 197 00:10:17,720 --> 00:10:20,080 Speaker 1: say llms can't code, when what they really mean is 198 00:10:20,280 --> 00:10:22,400 Speaker 1: llms can't write Rust which by the way, is a 199 00:10:22,440 --> 00:10:26,360 Speaker 1: coding language. Fair enough, but people select languages in part 200 00:10:26,400 --> 00:10:29,000 Speaker 1: based on how well lms work with them, So RUSS 201 00:10:29,040 --> 00:10:33,360 Speaker 1: people should get on that. What nobody projects more than 202 00:10:33,360 --> 00:10:36,000 Speaker 1: an AI booster. They thrive on the sense that they're 203 00:10:36,000 --> 00:10:39,600 Speaker 1: oppressed and villainized after years of seemingly every other, every 204 00:10:39,640 --> 00:10:42,800 Speaker 1: goddamn outlet on Earth claiming they're right, regardless of whether 205 00:10:42,840 --> 00:10:46,199 Speaker 1: there's any proof. They sneer and jeer and cry constantly 206 00:10:46,200 --> 00:10:48,720 Speaker 1: the people not showing adequate amounts of awe when an 207 00:10:48,760 --> 00:10:51,320 Speaker 1: AI lab says we did something in private, we can't 208 00:10:51,320 --> 00:10:54,079 Speaker 1: share it with you, but it's so cool, and constantly 209 00:10:54,120 --> 00:10:57,079 Speaker 1: act as if they're victims as they spread outright misinformation, 210 00:10:57,440 --> 00:10:59,920 Speaker 1: either through getting things wrong or never really caring enough 211 00:10:59,920 --> 00:11:02,760 Speaker 1: to check if they're right. Also, none of the booster 212 00:11:02,880 --> 00:11:06,600 Speaker 1: arguments actually survive a thorough response, as Nick Saresh proved 213 00:11:06,600 --> 00:11:10,439 Speaker 1: with his hilarious and brutal takedown of Potastics. Peachs Siesh 214 00:11:10,520 --> 00:11:12,320 Speaker 1: is a great guy. He's been on the show before, 215 00:11:12,480 --> 00:11:13,960 Speaker 1: and I've linked to his piece in the show notes, 216 00:11:14,000 --> 00:11:15,640 Speaker 1: and I'm going to bring him back on. He's written 217 00:11:15,640 --> 00:11:18,600 Speaker 1: for my newsletter as well. Absolute legend now there are 218 00:11:18,720 --> 00:11:21,600 Speaker 1: I believe some people who truly do love using llms, 219 00:11:21,679 --> 00:11:24,240 Speaker 1: yet they are not the ones defending them. But Tastic's 220 00:11:24,320 --> 00:11:26,920 Speaker 1: peace strips with condescension to the point that I feel 221 00:11:26,920 --> 00:11:29,520 Speaker 1: like he's trying to convince himself how good lms are. 222 00:11:29,559 --> 00:11:33,319 Speaker 1: Because boosters are eternal victims. He wrote them a piece 223 00:11:33,360 --> 00:11:35,920 Speaker 1: that they could send around to skeptics saying hh see 224 00:11:36,040 --> 00:11:38,160 Speaker 1: without being able to explain why it was such a 225 00:11:38,160 --> 00:11:41,520 Speaker 1: brutal takedown, mostly because they can't express why other than well, 226 00:11:42,120 --> 00:11:45,400 Speaker 1: this guy gets it. One cannot be a big, smart 227 00:11:45,400 --> 00:11:47,880 Speaker 1: genius that understands the glory and power of AI while 228 00:11:47,960 --> 00:11:51,080 Speaker 1: also acting like a scared little puppy every time somebody 229 00:11:51,120 --> 00:11:54,920 Speaker 1: tells them it sucks. You know what, This is a 230 00:11:54,960 --> 00:11:57,120 Speaker 1: great place to start. This is a great place to 231 00:11:57,200 --> 00:12:00,959 Speaker 1: get into how to deal with AI boosters because AI 232 00:12:01,080 --> 00:12:04,160 Speaker 1: boosters love being victims, and you should not play into it. 233 00:12:05,559 --> 00:12:07,800 Speaker 1: When you speak to an AI booster, you may get 234 00:12:07,840 --> 00:12:10,120 Speaker 1: the instinct to shake them vigorously or respond to their 235 00:12:10,120 --> 00:12:14,000 Speaker 1: posts by saying to do something with your something, or 236 00:12:14,000 --> 00:12:17,199 Speaker 1: that they're stupid. I understand the temptation, but you want 237 00:12:17,200 --> 00:12:19,400 Speaker 1: to keep a level head here. Keep your head on 238 00:12:19,440 --> 00:12:23,080 Speaker 1: the swivel. They thrive on this victimization. I'm sorry. If 239 00:12:23,080 --> 00:12:25,040 Speaker 1: you're an AI booster and this makes you feel bad, 240 00:12:25,080 --> 00:12:26,920 Speaker 1: Please reflect on your work and how many times you've 241 00:12:26,960 --> 00:12:29,280 Speaker 1: referred to somebody who didn't understand AI in a manner 242 00:12:29,360 --> 00:12:31,760 Speaker 1: that suggested that they were ignorant or tried to gaslight 243 00:12:31,800 --> 00:12:34,800 Speaker 1: them by saying AI was powerful while providing no actionable 244 00:12:34,800 --> 00:12:38,160 Speaker 1: ways of proving this or prove or just being able 245 00:12:38,200 --> 00:12:41,240 Speaker 1: to point in it being powerful. You cannot and should 246 00:12:41,280 --> 00:12:43,160 Speaker 1: not allow these people to act as if they're being 247 00:12:43,240 --> 00:12:46,720 Speaker 1: victimized or other Throughout this series, I'm going to address 248 00:12:46,760 --> 00:12:49,000 Speaker 1: a very specific thing. I'm going to use a term 249 00:12:49,240 --> 00:12:51,680 Speaker 1: boost the quip. This refers to the things that they 250 00:12:51,760 --> 00:12:54,800 Speaker 1: say and how often you hear them. These are lines 251 00:12:54,840 --> 00:12:56,920 Speaker 1: that you'll hear them say again and again and again 252 00:12:57,000 --> 00:13:00,600 Speaker 1: and again. They're common arguments, common cliches that demand response. 253 00:13:00,920 --> 00:13:03,640 Speaker 1: And let's start with our first booster quip. You are 254 00:13:03,760 --> 00:13:06,760 Speaker 1: just being a hater for attention. Contrarians just do it 255 00:13:06,800 --> 00:13:10,360 Speaker 1: for clicks and headlines. First and foremost, there are boosters 256 00:13:10,480 --> 00:13:13,199 Speaker 1: are pretty much every major think tank, government agency, and 257 00:13:13,240 --> 00:13:17,040 Speaker 1: media out there. It's extremely lucrative being an AI booster. 258 00:13:17,320 --> 00:13:20,360 Speaker 1: You're showered with panel invites access to executives, and you're 259 00:13:20,360 --> 00:13:22,360 Speaker 1: able to get headlines by saying how scared you are 260 00:13:22,360 --> 00:13:24,560 Speaker 1: of the computer. And it's really easy to do. So 261 00:13:24,920 --> 00:13:28,080 Speaker 1: being a booster is easy. And I must be clear 262 00:13:28,120 --> 00:13:31,760 Speaker 1: when I say booster, it doesn't always have to mean 263 00:13:31,920 --> 00:13:33,920 Speaker 1: over It could just mean the things you choose not 264 00:13:34,000 --> 00:13:36,360 Speaker 1: to do. It could mean the things you choose not 265 00:13:36,360 --> 00:13:38,360 Speaker 1: to criticize them for, or the things you just write 266 00:13:38,400 --> 00:13:40,880 Speaker 1: down that they say. But really we're talking about the 267 00:13:40,880 --> 00:13:43,560 Speaker 1: worst of them. But if you hear that sentence and 268 00:13:43,640 --> 00:13:46,520 Speaker 1: you don't think you're a booster, you can be a booster, 269 00:13:46,840 --> 00:13:49,400 Speaker 1: by the way, if you just choose not to criticize them. 270 00:13:49,400 --> 00:13:52,440 Speaker 1: But we're talking about the real assholes. Being a critic 271 00:13:52,559 --> 00:13:54,920 Speaker 1: requires you to constantly have to explain yourself in a 272 00:13:54,920 --> 00:13:57,480 Speaker 1: way that boosters never have to. Now, if a booster 273 00:13:57,600 --> 00:13:59,559 Speaker 1: says this to you, if they say you're just being 274 00:13:59,600 --> 00:14:02,200 Speaker 1: a hater for attention, you're just doing this for clicks, 275 00:14:02,559 --> 00:14:04,840 Speaker 1: ask them to explain first of all what they mean 276 00:14:04,880 --> 00:14:07,120 Speaker 1: by clicks or attention, and how they think you are 277 00:14:07,120 --> 00:14:10,600 Speaker 1: monetizing it, how this differs in its success from say, 278 00:14:10,760 --> 00:14:14,000 Speaker 1: anybody who interviews and quotes Sam Altman or Dario Wario, 279 00:14:14,080 --> 00:14:17,960 Speaker 1: Ama Day or whomever from Anthropic on Hard Fork. Ask 280 00:14:18,040 --> 00:14:20,600 Speaker 1: them what the difference is, and ask them why do 281 00:14:20,680 --> 00:14:23,360 Speaker 1: they believe your intentions as a critic are somehow malevolent 282 00:14:23,400 --> 00:14:25,640 Speaker 1: as opposed to those literally reporting what the rich and 283 00:14:25,680 --> 00:14:28,680 Speaker 1: powerful want them to. There's no answer here, because this 284 00:14:28,840 --> 00:14:32,480 Speaker 1: is not a coherent point of view. Boosters are more successful, 285 00:14:32,680 --> 00:14:35,360 Speaker 1: get more perks, and are in general treated better than 286 00:14:35,360 --> 00:14:39,120 Speaker 1: any critic and pretty much every major out there. Fundamentally, 287 00:14:39,160 --> 00:14:41,400 Speaker 1: these people exist in the land of the vague, and 288 00:14:41,440 --> 00:14:43,600 Speaker 1: they don't like it when you force them to get specific. 289 00:14:44,000 --> 00:14:46,440 Speaker 1: They will drag you toward what's just on the horizon, 290 00:14:46,480 --> 00:14:48,600 Speaker 1: but never quite define what the thing that dazzles you 291 00:14:48,720 --> 00:14:51,840 Speaker 1: so much will be or when it will arrive. Really, 292 00:14:51,880 --> 00:14:54,400 Speaker 1: their argument comes down to one thought. You must get 293 00:14:54,480 --> 00:14:56,400 Speaker 1: on board now because at some point it will be 294 00:14:56,480 --> 00:14:59,600 Speaker 1: so good you'll feel so stupid for not believing something 295 00:14:59,640 --> 00:15:02,520 Speaker 1: that it sucks wouldn't be really good. And if this 296 00:15:02,640 --> 00:15:04,640 Speaker 1: line sounds familiar, it's because you've heard it a million 297 00:15:04,640 --> 00:15:10,560 Speaker 1: times before, most notably with cryptocurrency, NFTs, Mataverse, Clubhouse, tons 298 00:15:10,600 --> 00:15:13,680 Speaker 1: of movements. Actually, they will make you define what would 299 00:15:13,720 --> 00:15:16,160 Speaker 1: impress you, which is not your job. In the same way, 300 00:15:16,200 --> 00:15:18,960 Speaker 1: finding the use case for them isn't your job. In fact, 301 00:15:19,160 --> 00:15:22,400 Speaker 1: you're the customer, you're the consumer. You are the person 302 00:15:22,520 --> 00:15:25,160 Speaker 1: AI needs to prove itself too, not the other way around. 303 00:15:25,880 --> 00:15:28,560 Speaker 1: But let's go to another booster quip. When they go, 304 00:15:29,760 --> 00:15:33,920 Speaker 1: you just don't get it, here's a great place to start. Say, 305 00:15:34,080 --> 00:15:36,920 Speaker 1: that's a really weird thing to say. It is peculiar 306 00:15:36,960 --> 00:15:40,240 Speaker 1: to suggest that somebody that doesn't get how to use 307 00:15:40,280 --> 00:15:44,360 Speaker 1: a product is weird, and that we as the consumer, 308 00:15:44,480 --> 00:15:48,920 Speaker 1: as the customer, must justify ourselves to our own purchases. No, no, 309 00:15:48,920 --> 00:15:51,560 Speaker 1: no no. If I don't get it, it's the booster's 310 00:15:51,600 --> 00:15:56,280 Speaker 1: job to tell me. Why make them justify their attitude. 311 00:15:56,560 --> 00:15:59,680 Speaker 1: Just like any product, we buy software to serve a need. 312 00:16:00,080 --> 00:16:02,800 Speaker 1: This is meant to be artificial intelligence. Why is it 313 00:16:02,880 --> 00:16:05,040 Speaker 1: so fucking stupid that I have to work out why 314 00:16:05,080 --> 00:16:07,600 Speaker 1: it's useful? The answer is, of course, that it has 315 00:16:07,640 --> 00:16:10,600 Speaker 1: no intellect. It is not intelligent, and large language models 316 00:16:10,600 --> 00:16:13,000 Speaker 1: are being pushed up mounted by a cadre of people 317 00:16:13,000 --> 00:16:16,080 Speaker 1: who are either easily impressed or invested, either emotionally or 318 00:16:16,160 --> 00:16:19,200 Speaker 1: financially in its success. Due to the company. They keep 319 00:16:19,200 --> 00:16:21,760 Speaker 1: all their intentions for the world. And if a booster 320 00:16:21,840 --> 00:16:24,120 Speaker 1: suggests you just don't get it, ask them to explain 321 00:16:24,160 --> 00:16:27,880 Speaker 1: the following What am I missing, What specifically is it 322 00:16:27,960 --> 00:16:30,240 Speaker 1: that is so life changing about this product based on 323 00:16:30,280 --> 00:16:33,040 Speaker 1: your own experience, not on any anecdotes from other people, 324 00:16:33,240 --> 00:16:35,040 Speaker 1: because they will say, well, I heard of a guy 325 00:16:35,080 --> 00:16:37,040 Speaker 1: who are ten billion lines of coat, and then the 326 00:16:37,080 --> 00:16:40,880 Speaker 1: baby looked at me and I cried. They don't have 327 00:16:40,960 --> 00:16:44,520 Speaker 1: real things themselves, So cut off the exits brought up 328 00:16:44,560 --> 00:16:47,560 Speaker 1: the doors. And then also ask them what use cases 329 00:16:47,560 --> 00:16:51,680 Speaker 1: are truly transformative about AI. Don't let them say well, 330 00:16:51,680 --> 00:16:54,680 Speaker 1: I heard in an industry. Actually make them prove themselves. 331 00:16:55,120 --> 00:16:57,440 Speaker 1: The use cases will likely be the AI is replaced. 332 00:16:57,480 --> 00:17:00,480 Speaker 1: Search for them that they use it for brainstorming or journaling, 333 00:17:00,520 --> 00:17:02,680 Speaker 1: proof reading an article, or looking through a big pile 334 00:17:02,720 --> 00:17:05,159 Speaker 1: of their notes or some other corpus of information and 335 00:17:05,200 --> 00:17:09,119 Speaker 1: summarizing it or pulling out insights. Who gives a shit? 336 00:17:10,160 --> 00:17:12,920 Speaker 1: Sorry not to be too a cerbic, but really, who 337 00:17:12,960 --> 00:17:16,240 Speaker 1: fucking cares? That shit's so boring? Hundreds of billions of 338 00:17:16,280 --> 00:17:18,959 Speaker 1: dollars of wasted investment on this and this is what 339 00:17:19,000 --> 00:17:22,880 Speaker 1: We've got three years in fucking Humpty dumpty. Could never 340 00:17:22,920 --> 00:17:26,199 Speaker 1: have it this good anyway. Our next booster quip is 341 00:17:26,280 --> 00:17:29,800 Speaker 1: one of my faves. Ahem, AI is powerful and getting 342 00:17:29,840 --> 00:17:33,280 Speaker 1: exponentially more powerful. Now, if a booster ever refers to 343 00:17:33,320 --> 00:17:36,680 Speaker 1: AI being powerful and getting more powerful, ask them the following, 344 00:17:37,040 --> 00:17:40,400 Speaker 1: what does powerful mean? In the event that they mentioned benchmarks, 345 00:17:40,440 --> 00:17:43,360 Speaker 1: ask them how those benchmarks apply to real world scenarios. 346 00:17:43,680 --> 00:17:47,000 Speaker 1: If they bring up swe bench the standard benchmark for coding, 347 00:17:47,160 --> 00:17:48,960 Speaker 1: ask them if they can code, and if they cannot, 348 00:17:49,600 --> 00:17:52,960 Speaker 1: ask them for another example. I mean, they will tell 349 00:17:53,000 --> 00:17:55,119 Speaker 1: you that they've spoken with coders. I've talked with a 350 00:17:55,119 --> 00:17:57,120 Speaker 1: lot of coders. I have a great one with Colt 351 00:17:57,200 --> 00:18:01,880 Speaker 1: Vogie coming up, another software engineer talking about MS. It's 352 00:18:01,920 --> 00:18:04,959 Speaker 1: so funny. It's so funny when you actually lay this 353 00:18:05,000 --> 00:18:08,040 Speaker 1: stuff out, how weak their arguments are. But in the 354 00:18:08,160 --> 00:18:11,120 Speaker 1: event they mentioned reasoning, ask them to define them. Once 355 00:18:11,119 --> 00:18:14,000 Speaker 1: they've defined reasoning, ask them to explain in plain English, 356 00:18:14,000 --> 00:18:16,879 Speaker 1: what reasoning allows you to do on a use case level, 357 00:18:17,160 --> 00:18:19,679 Speaker 1: not just how it works. They will likely bring up 358 00:18:19,680 --> 00:18:22,000 Speaker 1: the gold medal performance the open AI's model got on 359 00:18:22,000 --> 00:18:25,399 Speaker 1: the meth OLYMPIAD. Ask them why open ai hasn't released 360 00:18:25,400 --> 00:18:28,160 Speaker 1: that model. Then ask them what the actual practical use 361 00:18:28,200 --> 00:18:30,840 Speaker 1: case is that this success has opened up. They will 362 00:18:30,840 --> 00:18:33,320 Speaker 1: say it's an innovation. You've got to be patient, and 363 00:18:33,359 --> 00:18:37,240 Speaker 1: then pairper spray them. No, don't, don't pepper spray anyone anyway. 364 00:18:38,119 --> 00:18:40,040 Speaker 1: You should also then ask them what use cases have 365 00:18:40,160 --> 00:18:42,920 Speaker 1: arrived as a result of models becoming more powerful. If 366 00:18:42,960 --> 00:18:45,360 Speaker 1: they say vague things like oh in coding and oh 367 00:18:45,480 --> 00:18:48,880 Speaker 1: in medicine, ask them to get specific, and then ask 368 00:18:48,880 --> 00:18:51,240 Speaker 1: them what new products have arrived as a result. If 369 00:18:51,240 --> 00:18:53,399 Speaker 1: they say coding LMS, they will likely add that this 370 00:18:53,480 --> 00:18:57,320 Speaker 1: is replacing coders. Ask them where that has happened and 371 00:18:57,400 --> 00:19:01,040 Speaker 1: ask them to show you proof. Blinks and it will 372 00:19:01,040 --> 00:19:03,879 Speaker 1: not be sufficient to say that a CEO mentioned that 373 00:19:03,960 --> 00:19:09,160 Speaker 1: they did something with AI and efficiency. Get numbers. They 374 00:19:09,240 --> 00:19:11,800 Speaker 1: just won't. They won't do this. They will turn into 375 00:19:11,840 --> 00:19:13,679 Speaker 1: a pillar of salt. And we got two more fucking 376 00:19:13,720 --> 00:19:16,000 Speaker 1: parts of this. I mean you're gonna have a ball 377 00:19:16,040 --> 00:19:18,800 Speaker 1: with this. Look. The core of the AI booster's argument 378 00:19:18,920 --> 00:19:20,840 Speaker 1: is that they need to make you feel bad. They 379 00:19:20,920 --> 00:19:23,440 Speaker 1: like the gaslight, and you should. You need to refuse 380 00:19:23,520 --> 00:19:25,439 Speaker 1: to let them. You need to push back heavily. If 381 00:19:25,520 --> 00:19:27,879 Speaker 1: there's ever a point where you feel like they are 382 00:19:27,880 --> 00:19:29,840 Speaker 1: trying to make you feel stupid, ask them why they're 383 00:19:29,840 --> 00:19:33,159 Speaker 1: doing so. And to be clear, anyone with a compelling 384 00:19:33,240 --> 00:19:35,399 Speaker 1: argument doesn't have to make you feel bad to convince you. 385 00:19:35,720 --> 00:19:39,880 Speaker 1: The iPhone didn't need a spurious hype cycle of proof 386 00:19:40,080 --> 00:19:42,679 Speaker 1: of people saying you must look at this, it is 387 00:19:42,760 --> 00:19:46,679 Speaker 1: so important. When it didn't really work. It worked immediately. 388 00:19:47,160 --> 00:19:49,000 Speaker 1: Now I said in my news that it didn't need 389 00:19:49,040 --> 00:19:52,040 Speaker 1: a fucking marketing campaign. Yes there was marketing dollars behind 390 00:19:52,040 --> 00:19:55,680 Speaker 1: the iPhone. Eric newcomer, come at me with better WORKMATEE 391 00:19:55,720 --> 00:19:58,119 Speaker 1: respond to the rest of this, But we all kind 392 00:19:58,119 --> 00:19:59,479 Speaker 1: of got it with the iPhone. In fact, the moment 393 00:19:59,520 --> 00:20:02,600 Speaker 1: Steve job announced their piece of shit that he was, 394 00:20:02,400 --> 00:20:05,280 Speaker 1: he was. He said, here's the phone, here's an IPO, 395 00:20:05,680 --> 00:20:08,240 Speaker 1: here's here's ema. You could do all these on one device, 396 00:20:08,320 --> 00:20:12,120 Speaker 1: and everyone, oh, yeah, that is good. That is really 397 00:20:12,160 --> 00:20:15,359 Speaker 1: obvious that that was good. It was impressive because it 398 00:20:15,440 --> 00:20:19,880 Speaker 1: was impressive. And boosters will suggest you are intentional or 399 00:20:19,960 --> 00:20:21,879 Speaker 1: in not liking AI because you're a hater or a 400 00:20:21,880 --> 00:20:24,439 Speaker 1: cynic or a luddie. They'll suggest that you're ignorant for 401 00:20:24,480 --> 00:20:28,199 Speaker 1: not being amazed by chat GPT. Let me tell you something, 402 00:20:28,560 --> 00:20:30,679 Speaker 1: You don't have to be impressed by anything. By defam 403 00:20:30,960 --> 00:20:34,760 Speaker 1: and any product, especially software designed to make you feel 404 00:20:34,800 --> 00:20:38,320 Speaker 1: stupid for not getting it is poorly designed. Chat GPT 405 00:20:38,440 --> 00:20:42,280 Speaker 1: is the ultimate form of silicon value sociopathy. You must 406 00:20:42,280 --> 00:20:44,080 Speaker 1: do the work to find the use cases and thank 407 00:20:44,160 --> 00:20:47,160 Speaker 1: them for giving you the chance to do so. AI 408 00:20:47,280 --> 00:20:50,400 Speaker 1: is not even good reliable software. It resembles the death 409 00:20:50,400 --> 00:20:54,240 Speaker 1: of the art of technology. Inconsistent and unreliable by definition, 410 00:20:54,440 --> 00:20:57,800 Speaker 1: inefficient by design, financially ruin us, and adds to the 411 00:20:57,840 --> 00:20:59,960 Speaker 1: cognitive load of the user by requiring them to be 412 00:21:00,160 --> 00:21:02,520 Speaker 1: ever vigilant of the shit asked outputs that can come 413 00:21:02,560 --> 00:21:05,399 Speaker 1: out of them. So here's a really easy way to 414 00:21:05,480 --> 00:21:07,960 Speaker 1: deal with this. If a booster ever suggests you are 415 00:21:08,000 --> 00:21:10,879 Speaker 1: stupid or ignorant, ask them why it's necessary to demean 416 00:21:10,920 --> 00:21:13,320 Speaker 1: you to get their point across, even if you are 417 00:21:13,400 --> 00:21:15,919 Speaker 1: unable to argue on a technical level, make them explain 418 00:21:15,960 --> 00:21:19,040 Speaker 1: why the software itself can't convince you, and be vigilant. 419 00:21:19,080 --> 00:21:21,400 Speaker 1: Boosters seldom live in reality and will do everything they 420 00:21:21,400 --> 00:21:24,040 Speaker 1: can to pull you off course. And I should add 421 00:21:24,040 --> 00:21:27,080 Speaker 1: that there is a fair criticism here. I do insult people, 422 00:21:27,080 --> 00:21:29,600 Speaker 1: I do demean them. I call them babies, and I 423 00:21:29,760 --> 00:21:32,480 Speaker 1: do funny voices, and I do that because I don't 424 00:21:32,520 --> 00:21:35,880 Speaker 1: respect them. I'm not. I'm I am here to tell 425 00:21:35,880 --> 00:21:38,679 Speaker 1: you how I feel. I will convince you through the 426 00:21:39,040 --> 00:21:41,119 Speaker 1: large amounts of work I do in the research I have. 427 00:21:41,440 --> 00:21:44,359 Speaker 1: I don't if you disagree with me, you disagree with me, Eh, 428 00:21:44,359 --> 00:21:47,359 Speaker 1: it's fine. And yeah, these people do sound kind of silly. 429 00:21:47,440 --> 00:21:49,560 Speaker 1: I'll get you a Casey Newton thing in a couple 430 00:21:49,640 --> 00:21:52,359 Speaker 1: of episodes. I think that really it's impossible to call 431 00:21:52,400 --> 00:22:09,080 Speaker 1: them otherwise. Look, if you say that none of these 432 00:22:09,160 --> 00:22:11,399 Speaker 1: companies make money, they'll say it's in the early days. 433 00:22:11,760 --> 00:22:14,879 Speaker 1: If you say AI companies burn billions, they'll say the 434 00:22:14,880 --> 00:22:16,919 Speaker 1: cost of inference is coming down. If you say the 435 00:22:16,920 --> 00:22:20,080 Speaker 1: industry is massively overbuilding, they'll say that this is actually 436 00:22:20,119 --> 00:22:22,000 Speaker 1: just let the dot com boom and that the infrastructre 437 00:22:22,000 --> 00:22:23,920 Speaker 1: will be picked up at used in the future. If 438 00:22:23,920 --> 00:22:25,800 Speaker 1: you say there are no real use cases, they'll say 439 00:22:25,880 --> 00:22:30,040 Speaker 1: chat GBT as seven hundred billion weekly users. Every single time. 440 00:22:30,160 --> 00:22:33,640 Speaker 1: There's the same goddamn arguments, which is why from here 441 00:22:33,680 --> 00:22:35,719 Speaker 1: on out, I'm going to give you the responses to 442 00:22:35,760 --> 00:22:40,080 Speaker 1: all of them. And there's an article versus you could 443 00:22:40,080 --> 00:22:43,520 Speaker 1: print up and stuff it in them. No no, no, no, 444 00:22:43,520 --> 00:22:47,119 Speaker 1: no no violence. Everyone be nice, but you're going to 445 00:22:47,160 --> 00:22:49,040 Speaker 1: be able to use these arguments going forward. And let's 446 00:22:49,040 --> 00:22:54,760 Speaker 1: start with my favorite booster, quip AI will. Any time 447 00:22:55,000 --> 00:22:57,840 Speaker 1: an AI booster says AI will tell them to stop, 448 00:22:58,040 --> 00:23:00,679 Speaker 1: tell them to stop and explain what AI can do now. 449 00:23:01,040 --> 00:23:04,120 Speaker 1: And if they insist, ask them both when they expect 450 00:23:04,119 --> 00:23:06,560 Speaker 1: things to happen in the way they're talking about, and 451 00:23:06,600 --> 00:23:09,399 Speaker 1: if they say very soon, ask them to be more specific. 452 00:23:09,880 --> 00:23:11,400 Speaker 1: Get them to agree to a date and then call 453 00:23:11,440 --> 00:23:13,760 Speaker 1: them on that day or show up at their house. Hell, 454 00:23:13,760 --> 00:23:16,120 Speaker 1: you could be waiting inside when they get there, assuming 455 00:23:16,160 --> 00:23:19,119 Speaker 1: you have a key legally. Now here's another one, another 456 00:23:19,160 --> 00:23:21,680 Speaker 1: boost the whip. They will say agents will automate large 457 00:23:21,720 --> 00:23:24,240 Speaker 1: parts of our economy, and I will say, fucking stop. 458 00:23:25,920 --> 00:23:29,760 Speaker 1: There's that will bullshit again. There's that will will will will. 459 00:23:30,600 --> 00:23:32,960 Speaker 1: Agents don't work. They don't work at all. The term 460 00:23:33,040 --> 00:23:36,679 Speaker 1: agent means, to quote Maxwolf, a workflow where the LLM 461 00:23:36,680 --> 00:23:38,680 Speaker 1: can make its own decisions, such as in the case 462 00:23:38,720 --> 00:23:41,239 Speaker 1: of web search, where the LLM is told you can 463 00:23:41,280 --> 00:23:43,520 Speaker 1: search the web if you need to, then can output. 464 00:23:43,560 --> 00:23:46,280 Speaker 1: I should search the web and search the web. Yet 465 00:23:46,359 --> 00:23:49,200 Speaker 1: agent has now become a mythical creature that means totally 466 00:23:49,200 --> 00:23:52,159 Speaker 1: autonomous AI that can do an entire job. If anyone 467 00:23:52,200 --> 00:23:55,000 Speaker 1: tells you agents are dot dot dot, you should ask 468 00:23:55,040 --> 00:23:56,960 Speaker 1: them to point to one. If they say coding, please 469 00:23:57,000 --> 00:23:59,680 Speaker 1: demand that they explain how autonomous these things are. If 470 00:23:59,680 --> 00:24:02,200 Speaker 1: they say they can reflect her entire code bases, ask 471 00:24:02,240 --> 00:24:04,200 Speaker 1: them what that means, and also laugh at them because 472 00:24:04,200 --> 00:24:07,520 Speaker 1: they will not know. Ask them to explain how self force. 473 00:24:07,600 --> 00:24:09,760 Speaker 1: His own research shows that agents only have a fifty 474 00:24:09,800 --> 00:24:13,119 Speaker 1: eight percent success rate on single step tasks that you 475 00:24:13,200 --> 00:24:15,639 Speaker 1: ask it to do one thing, and thirty five percent 476 00:24:15,680 --> 00:24:19,000 Speaker 1: on multi step tasks. And what's crazy as well? And 477 00:24:19,040 --> 00:24:20,919 Speaker 1: I won't have this in the episode now, it's just 478 00:24:20,960 --> 00:24:24,080 Speaker 1: gonna have to look it up. There is a story 479 00:24:24,080 --> 00:24:27,119 Speaker 1: that just came out about open Aiyes projections, they reduce 480 00:24:27,160 --> 00:24:28,960 Speaker 1: the amount of money they're making over the next few 481 00:24:29,040 --> 00:24:32,639 Speaker 1: years in agents down twenty six billion dollars less they 482 00:24:32,800 --> 00:24:35,880 Speaker 1: just remove that part. Do you think agents exist now? 483 00:24:36,520 --> 00:24:39,239 Speaker 1: Agents are not autonomous. They do not replace jobs. They 484 00:24:39,240 --> 00:24:41,880 Speaker 1: cannot replace code as they are not going to do so, 485 00:24:42,760 --> 00:24:46,440 Speaker 1: because probabilistic models are a horrible means of taking precise actions. 486 00:24:46,600 --> 00:24:49,159 Speaker 1: And almost anyone who brings up agents as a booster 487 00:24:49,440 --> 00:24:53,679 Speaker 1: is either misinformed or in the business of misinformation. Now 488 00:24:53,720 --> 00:24:56,440 Speaker 1: here's another booster grip for you. What I mean is 489 00:24:56,480 --> 00:24:58,960 Speaker 1: that AI is like the early days of the Internet 490 00:25:00,080 --> 00:25:03,800 Speaker 1: any cases. I think they're referring to AI as being 491 00:25:03,960 --> 00:25:07,120 Speaker 1: early as a reference to those early days, and they 492 00:25:07,160 --> 00:25:09,520 Speaker 1: never really refer to what that means, because the early 493 00:25:09,600 --> 00:25:11,719 Speaker 1: days of the Internet can refer to just about anything. 494 00:25:12,160 --> 00:25:14,600 Speaker 1: Are we talking about dial up DSL? Are we talking 495 00:25:14,600 --> 00:25:17,080 Speaker 1: about the pre platform days when people accessed it the 496 00:25:17,119 --> 00:25:19,800 Speaker 1: Internet via compy server? Al well, yes, yes, I remember 497 00:25:19,840 --> 00:25:21,720 Speaker 1: the article from Newsweek. I already explained it in my 498 00:25:21,760 --> 00:25:24,639 Speaker 1: newsletter reality Check. I'm going to quote myself about this 499 00:25:24,720 --> 00:25:27,200 Speaker 1: fucking article where the guy said the Internet wouldn't take off. 500 00:25:27,760 --> 00:25:30,000 Speaker 1: In any case, one guy said, saying that the Internet 501 00:25:30,040 --> 00:25:32,200 Speaker 1: won't be big doesn't mean a fucking thing about generat 502 00:25:32,200 --> 00:25:34,520 Speaker 1: iv AI, And you're a simpleton if you think it does. 503 00:25:35,040 --> 00:25:36,800 Speaker 1: One guy being wrong in some way is not a 504 00:25:36,800 --> 00:25:38,600 Speaker 1: response to by work, and I will crush you like 505 00:25:38,640 --> 00:25:41,840 Speaker 1: a bug. Again, I'm using ad hominin attacks. Who cares 506 00:25:42,119 --> 00:25:44,760 Speaker 1: these fucking people are rude as hell. Now, if your 507 00:25:44,840 --> 00:25:48,200 Speaker 1: argument is that the early Internet required expensive Sun microsystem 508 00:25:48,280 --> 00:25:51,159 Speaker 1: service to run, Jim Cavello of Goldman Sachs addressed that 509 00:25:51,440 --> 00:25:54,280 Speaker 1: in June twenty twenty four by saying that the costs 510 00:25:54,280 --> 00:25:57,320 Speaker 1: were that the cost piled in comparison, adding that we 511 00:25:57,359 --> 00:25:59,480 Speaker 1: also didn't need to expand our fucking power grid to 512 00:25:59,520 --> 00:26:04,119 Speaker 1: build the early web. Well, sir, there's another booster whip, sir. Actually, 513 00:26:04,200 --> 00:26:07,399 Speaker 1: people said smartphones wouldn't be big. This is a straight 514 00:26:07,480 --> 00:26:09,399 Speaker 1: up lie. By the way, I've heard this a good amount. 515 00:26:09,600 --> 00:26:12,960 Speaker 1: I've heard at least five people say, yeah, well, people 516 00:26:13,000 --> 00:26:16,199 Speaker 1: didn't didn't think the iPhone would be big. The iPhone 517 00:26:16,320 --> 00:26:20,440 Speaker 1: wasn't going to be big. Actually, huhh. This is a lie. 518 00:26:20,600 --> 00:26:23,320 Speaker 1: It's a lie. It's a lie. You're lying. Also, as 519 00:26:23,400 --> 00:26:25,800 Speaker 1: Jim Cavello from gobmin Sex noted, there were hundreds of 520 00:26:25,840 --> 00:26:28,720 Speaker 1: presentations in the early two thousands that included road maps 521 00:26:28,720 --> 00:26:31,840 Speaker 1: that accurately fit how smartphones rolled out, and that no 522 00:26:31,920 --> 00:26:35,720 Speaker 1: such roadmap exists for generative AI. The iPhone was also 523 00:26:35,760 --> 00:26:38,240 Speaker 1: an immediate success as a thing that people paid for 524 00:26:38,400 --> 00:26:40,639 Speaker 1: with Apple selling four million units in the space of 525 00:26:40,680 --> 00:26:43,480 Speaker 1: six months, and this was on an exclusive contract for 526 00:26:43,840 --> 00:26:46,920 Speaker 1: several years, I think with Singular Wireless now called AT 527 00:26:47,000 --> 00:26:49,280 Speaker 1: and T hell in two thousand and six. Since the 528 00:26:49,359 --> 00:26:52,360 Speaker 1: year before the iPhone launched, there were smartphones, and there 529 00:26:52,400 --> 00:26:56,080 Speaker 1: was an estimated seven point seven million worldwide smartphone shipments, 530 00:26:56,200 --> 00:26:59,560 Speaker 1: mostly from BlackBerry, Windows Mobile and Palm. Though, to be 531 00:26:59,600 --> 00:27:02,920 Speaker 1: generous to the GENERATIVEAI boosters, I'm going to disregard those 532 00:27:03,000 --> 00:27:06,159 Speaker 1: because they actually he'll prove my point more and I 533 00:27:06,240 --> 00:27:07,760 Speaker 1: just want to be clear that the early days of 534 00:27:07,760 --> 00:27:11,199 Speaker 1: the Internet are not a sensible comparison to GENERATIVEAI. The 535 00:27:11,280 --> 00:27:13,840 Speaker 1: original Attentional is all you Need paper, the one that 536 00:27:13,920 --> 00:27:16,520 Speaker 1: kicked off the transformer based large language model ERA, was 537 00:27:16,560 --> 00:27:20,200 Speaker 1: published in June twenty seventeen. Chat GPT launched in November 538 00:27:20,240 --> 00:27:24,439 Speaker 1: twenty twenty two. It's not early. We're not early anymore. 539 00:27:24,560 --> 00:27:26,480 Speaker 1: We haven't been for a while. But nevertheless, if we're 540 00:27:26,520 --> 00:27:29,399 Speaker 1: saying early days here, we should actually define what that means. 541 00:27:29,600 --> 00:27:32,560 Speaker 1: As I mentioned previously, people paid for the iPhone immediately 542 00:27:32,560 --> 00:27:34,639 Speaker 1: despite it being a device that was completely and utterly 543 00:27:34,680 --> 00:27:38,000 Speaker 1: new on one specific carrier. While there was a small 544 00:27:38,040 --> 00:27:40,840 Speaker 1: group of consumers that might have used similar devices like 545 00:27:40,880 --> 00:27:44,800 Speaker 1: the compact ipack. It's kind of a cool device. The 546 00:27:44,840 --> 00:27:47,320 Speaker 1: iPhone was a completely new kind of computing sold it 547 00:27:47,359 --> 00:27:49,840 Speaker 1: at premium, requiring you to have a contract with that 548 00:27:49,920 --> 00:27:54,560 Speaker 1: specific carrier. Conversely, chat GPT's annualized revenue in December twenty 549 00:27:54,560 --> 00:27:56,920 Speaker 1: twenty three was one point six billion dollars, so about 550 00:27:56,920 --> 00:27:58,919 Speaker 1: one hundred and thirty three million in that month for 551 00:27:58,960 --> 00:28:01,159 Speaker 1: a product that had by that time raised over ten 552 00:28:01,200 --> 00:28:03,840 Speaker 1: billion dollars. And while we don't know what OpenAI lost 553 00:28:03,880 --> 00:28:06,440 Speaker 1: in twenty twenty three, reports suggested it burned over five 554 00:28:06,480 --> 00:28:09,560 Speaker 1: billion dollars in twenty twenty four. Big Tech is spent 555 00:28:09,600 --> 00:28:12,159 Speaker 1: over five hundred billion dollars in capital expenditures in the 556 00:28:12,240 --> 00:28:15,800 Speaker 1: last eighteen months, and all told, between investments of cloud 557 00:28:15,840 --> 00:28:19,320 Speaker 1: credits and infrastructure, will likely sink over six hundred billion 558 00:28:19,320 --> 00:28:21,840 Speaker 1: dollars by the year's end. The early days of the 559 00:28:21,840 --> 00:28:24,439 Speaker 1: Internet would define not by its lack of investment or attention, 560 00:28:24,520 --> 00:28:27,520 Speaker 1: but by its obscurity. Even in two thousand, around the 561 00:28:27,520 --> 00:28:29,639 Speaker 1: time of the dot com bubble, only fifty two percent 562 00:28:29,680 --> 00:28:31,760 Speaker 1: of US adults used the Internet, and it would take 563 00:28:31,800 --> 00:28:34,879 Speaker 1: another nineteen years for ninety percent of US adults to 564 00:28:34,920 --> 00:28:37,520 Speaker 1: do so. These early days were also defined by its 565 00:28:37,560 --> 00:28:41,000 Speaker 1: early functionality. The Internet would become so much more because 566 00:28:41,000 --> 00:28:43,440 Speaker 1: of the things that hyperconnectivity allowed us to do, and 567 00:28:43,480 --> 00:28:46,240 Speaker 1: both faster inte net connections and the ability to host 568 00:28:46,240 --> 00:28:49,920 Speaker 1: software and the cloud would change everything. We could define 569 00:28:49,960 --> 00:28:52,440 Speaker 1: what better would mean and make reasonable predictions about what 570 00:28:52,480 --> 00:28:55,960 Speaker 1: people could do on a better Internet. Yet, even in 571 00:28:56,000 --> 00:28:58,400 Speaker 1: those early days, it was obvious why you were using 572 00:28:58,400 --> 00:29:01,000 Speaker 1: the Internet and how it might grow from them. One 573 00:29:01,000 --> 00:29:03,080 Speaker 1: did not have to struggle to explain why buying a 574 00:29:03,080 --> 00:29:05,640 Speaker 1: book online might be useful or quicker than a shop, 575 00:29:05,920 --> 00:29:08,040 Speaker 1: or why a website might make a quicker reference than 576 00:29:08,120 --> 00:29:10,280 Speaker 1: having to go to a library, or why downloading a 577 00:29:10,320 --> 00:29:12,720 Speaker 1: game or a song might be a good idea. While 578 00:29:12,800 --> 00:29:15,880 Speaker 1: habits might have needed adjusting, it was blatantly obvious what 579 00:29:15,920 --> 00:29:19,560 Speaker 1: the value of the early Internet was. It's also unclear 580 00:29:19,600 --> 00:29:22,400 Speaker 1: when the early days of the Internet ended. Only forty 581 00:29:22,440 --> 00:29:24,880 Speaker 1: four percent of US adults had access to broadband internet 582 00:29:24,880 --> 00:29:27,960 Speaker 1: in two thousand and six. Were those the early days 583 00:29:27,960 --> 00:29:31,160 Speaker 1: of the Internet. The answer is no, and that this 584 00:29:31,200 --> 00:29:33,080 Speaker 1: point is brought up by people with a poor grasp 585 00:29:33,120 --> 00:29:35,960 Speaker 1: of history and a flimsy attachment to reality. The early 586 00:29:36,040 --> 00:29:38,360 Speaker 1: days of the Internet were very, very, very different to 587 00:29:38,440 --> 00:29:41,080 Speaker 1: any associated tech boom since, and we need to stop 588 00:29:41,120 --> 00:29:44,120 Speaker 1: making the comparison. The Internet also grew in a vastly 589 00:29:44,120 --> 00:29:48,080 Speaker 1: different information ecosystem. Generative AI has had the benefit of 590 00:29:48,120 --> 00:29:51,479 Speaker 1: mass media driven by the Internet, along with social media 591 00:29:51,600 --> 00:29:55,280 Speaker 1: and social pressure to adopt AI for multiple years. And 592 00:29:55,320 --> 00:29:59,320 Speaker 1: now our last boost to quip for the episode, actually 593 00:29:59,440 --> 00:30:01,200 Speaker 1: that I meant something else. What I mean is that 594 00:30:01,240 --> 00:30:03,680 Speaker 1: we're in the early days of AI. All of the 595 00:30:03,720 --> 00:30:07,640 Speaker 1: other things you said were very misleading. You misread my statements. Somehow, 596 00:30:08,080 --> 00:30:10,400 Speaker 1: we are not in the early days of generat if AI, 597 00:30:10,440 --> 00:30:13,920 Speaker 1: and anyone using this argument is either ignorant or intentionally deceptive. 598 00:30:14,560 --> 00:30:17,800 Speaker 1: According to Pew, as of mid twenty twenty five, thirty 599 00:30:17,800 --> 00:30:20,400 Speaker 1: four percent of US adults have used chat GPT, with 600 00:30:20,440 --> 00:30:22,640 Speaker 1: seventy nine percent saying they'd at least heard of it 601 00:30:23,280 --> 00:30:26,600 Speaker 1: and a little about it. Even Furthermore, chat GPT has 602 00:30:26,640 --> 00:30:28,840 Speaker 1: always had a free version. On top of that, A 603 00:30:28,880 --> 00:30:31,240 Speaker 1: study from May twenty twenty three found that over ten 604 00:30:31,760 --> 00:30:35,240 Speaker 1: nine hundred news headlines mentioned chat GPT between November twenty 605 00:30:35,240 --> 00:30:37,600 Speaker 1: twenty two and March twenty twenty three, and a brand 606 00:30:37,680 --> 00:30:39,760 Speaker 1: Watch report found that in the first five months of 607 00:30:39,800 --> 00:30:43,320 Speaker 1: its release, chat GPT received over nine point two million 608 00:30:43,720 --> 00:30:47,080 Speaker 1: mentions on social media. Nearly eighty percent of people have 609 00:30:47,120 --> 00:30:49,440 Speaker 1: heard of chat GPT, in over a quarter of Americans 610 00:30:49,480 --> 00:30:52,240 Speaker 1: have used them. If we're defining the early days based 611 00:30:52,280 --> 00:30:55,600 Speaker 1: on consumer exposure, that ship has sailed. If we're defining 612 00:30:55,640 --> 00:30:58,080 Speaker 1: the early days by the passage of time, it's been 613 00:30:58,120 --> 00:31:00,320 Speaker 1: eight years since Attention Is All You Need and three 614 00:31:00,360 --> 00:31:03,840 Speaker 1: since CHADGBT came out. While three years might not seem 615 00:31:03,920 --> 00:31:06,400 Speaker 1: like a lot of time, the whole foundation of an 616 00:31:06,400 --> 00:31:09,800 Speaker 1: early day's argument is that in the early days, things 617 00:31:09,880 --> 00:31:14,200 Speaker 1: do not receive the venture funding, research, attention, infrastructural support, 618 00:31:14,280 --> 00:31:17,600 Speaker 1: or business interest necessary to make them big. In twenty 619 00:31:17,640 --> 00:31:20,600 Speaker 1: twenty four, nearly thirty three percent of all global venture 620 00:31:20,640 --> 00:31:23,800 Speaker 1: funding went to artificial intelligence, and according to the Information 621 00:31:24,120 --> 00:31:26,760 Speaker 1: AI startups have raised over forty billion dollars in twenty 622 00:31:26,760 --> 00:31:30,920 Speaker 1: twenty five alone, with Statista adding that AI absorbed seventy 623 00:31:30,960 --> 00:31:33,720 Speaker 1: one percent of VC funding in the first quarter of 624 00:31:33,760 --> 00:31:37,120 Speaker 1: twenty twenty five. These numbers also failed to account the 625 00:31:37,160 --> 00:31:41,000 Speaker 1: massive infrastructure costs that companies like open Ai and Anthropic 626 00:31:41,040 --> 00:31:43,920 Speaker 1: don't have to pay for the limitations of the early 627 00:31:44,160 --> 00:31:47,480 Speaker 1: Internet were twofold, the fiber optic cable boom that led 628 00:31:47,480 --> 00:31:51,120 Speaker 1: to the fiber optic bubble bursting when telecommunications companies massively 629 00:31:51,160 --> 00:31:54,400 Speaker 1: overinvested in infrastructure, which I will get too shortly. There was 630 00:31:54,440 --> 00:31:57,560 Speaker 1: also the lack of scalable cloud infrastructure to allow distinct 631 00:31:57,560 --> 00:32:00,000 Speaker 1: apps to be run online, a problem solved by Amazon 632 00:32:00,080 --> 00:32:04,440 Speaker 1: Web Services, among others. In Generative AI's case, Microsoft, Google 633 00:32:04,480 --> 00:32:07,080 Speaker 1: and Amazon have built the fiber optic cables for large 634 00:32:07,120 --> 00:32:10,960 Speaker 1: language models. Open AI and Anthropic have everything they need. 635 00:32:11,360 --> 00:32:14,240 Speaker 1: They have even if they say otherwise, plenty of compute, 636 00:32:14,480 --> 00:32:17,000 Speaker 1: access to the literal greatest minds in the field, the 637 00:32:17,080 --> 00:32:19,920 Speaker 1: constant attention of the media and global governments, and effectively 638 00:32:19,960 --> 00:32:23,200 Speaker 1: no regulations or restrictions stopping them from training their models 639 00:32:23,200 --> 00:32:26,080 Speaker 1: and the works of millions of people or destroying our environment. 640 00:32:26,480 --> 00:32:30,120 Speaker 1: They've already had this support too. Open ai was allowed 641 00:32:30,160 --> 00:32:33,040 Speaker 1: to burn half a billion dollars on a single training 642 00:32:33,120 --> 00:32:36,080 Speaker 1: run for GPT four point five and five, and they 643 00:32:36,080 --> 00:32:39,840 Speaker 1: did multiple runs. If anything, the massive amounts of capital 644 00:32:39,880 --> 00:32:42,320 Speaker 1: have allowed us to massively condense the time in which 645 00:32:42,320 --> 00:32:44,960 Speaker 1: a bubble goes from possible to bursting and washing out 646 00:32:44,960 --> 00:32:47,240 Speaker 1: a bunch of people. Because the tech industry is such 647 00:32:47,240 --> 00:32:49,800 Speaker 1: a powerful follow up work culture that only one or 648 00:32:49,840 --> 00:32:52,160 Speaker 1: two unique ideas can exist at one time, and I 649 00:32:52,200 --> 00:32:55,520 Speaker 1: think those ideas are currently open AI and anthropic. The 650 00:32:55,600 --> 00:32:59,520 Speaker 1: early Day's argument hinges on obscurity and limited resources, something 651 00:32:59,520 --> 00:33:02,800 Speaker 1: that general does not get to whine about. Companies that 652 00:33:02,840 --> 00:33:05,560 Speaker 1: make effectively no revenue can raise five hundred million dollars 653 00:33:05,640 --> 00:33:08,320 Speaker 1: to do the same AI coding bullshit that everybody else does. 654 00:33:09,040 --> 00:33:12,000 Speaker 1: In simpler terms, these companies are flush with cash, have 655 00:33:12,160 --> 00:33:15,200 Speaker 1: all the attention and investment they could possibly need. After all, 656 00:33:15,240 --> 00:33:18,120 Speaker 1: attention is all you need, and are still unable to 657 00:33:18,160 --> 00:33:21,680 Speaker 1: create a product with a defined, meaningful mass market use case, 658 00:33:21,880 --> 00:33:25,320 Speaker 1: let alone one that doesn't burn money. In fact, I 659 00:33:25,360 --> 00:33:28,680 Speaker 1: believe that, thanks to effectively infinite resources, we've speed run 660 00:33:28,720 --> 00:33:31,840 Speaker 1: the entire large language model era and we're nearing the end. 661 00:33:32,360 --> 00:33:35,440 Speaker 1: These companies got what they wanted, and I think I 662 00:33:35,520 --> 00:33:38,440 Speaker 1: want to die in Minecraft, obviously, but I must press on. 663 00:33:38,640 --> 00:33:41,320 Speaker 1: I must. Just saying all these things out loud, you 664 00:33:41,360 --> 00:33:44,440 Speaker 1: really get a sense for how illogical these people are 665 00:33:44,480 --> 00:33:46,680 Speaker 1: and how much bullshit is going around to try and 666 00:33:46,680 --> 00:33:50,120 Speaker 1: push back against skeptics. But it's not gonna work. It's 667 00:33:50,160 --> 00:33:52,440 Speaker 1: not gonna work. I hope you're not tired of me 668 00:33:52,480 --> 00:33:55,440 Speaker 1: talking about boost equips and doing silly voices. Is You've 669 00:33:55,480 --> 00:33:57,720 Speaker 1: got two more of these fucking episodes coming, and I 670 00:33:57,760 --> 00:34:01,840 Speaker 1: love recording them. I really do say recording. Jesus Christ, 671 00:34:02,200 --> 00:34:04,960 Speaker 1: my friends were in hell, but we're together in hell 672 00:34:05,000 --> 00:34:07,280 Speaker 1: and that's fun. And there's a lot more bullshit to 673 00:34:07,280 --> 00:34:18,640 Speaker 1: break down. Speak to you tomorrow. Thank you for listening 674 00:34:18,680 --> 00:34:21,319 Speaker 1: to Better Offline. The editor and composer of the Better 675 00:34:21,360 --> 00:34:24,359 Speaker 1: Offline theme song is Matasowski. You can check out more 676 00:34:24,360 --> 00:34:27,879 Speaker 1: of his music and audio projects at Matasowski dot com, 677 00:34:28,000 --> 00:34:32,880 Speaker 1: M A T t O. S O w Ski dot com. 678 00:34:32,960 --> 00:34:35,560 Speaker 1: You can email me at easy at Better offline dot com, 679 00:34:35,680 --> 00:34:38,000 Speaker 1: or visit Better offline dot com to find more podcast 680 00:34:38,040 --> 00:34:41,400 Speaker 1: links and of course, my newsletter. I also really recommend 681 00:34:41,400 --> 00:34:43,320 Speaker 1: you go to chat dot Where's youreed dot at to 682 00:34:43,400 --> 00:34:46,200 Speaker 1: visit the discord, and go to our slash Better Offline 683 00:34:46,239 --> 00:34:49,480 Speaker 1: to check out our reddit. Thank you so much for listening. 684 00:34:50,280 --> 00:34:53,000 Speaker 1: Better Offline is a production of cool Zone Media. For 685 00:34:53,120 --> 00:34:56,280 Speaker 1: more from cool Zone Media, visit our website Cool zonemedia 686 00:34:56,320 --> 00:34:59,160 Speaker 1: dot com, or check us out on the iHeartRadio app, 687 00:34:59,200 --> 00:35:15,160 Speaker 1: Apple podcast, or wherever you get your podcasts.