WEBVTT - I Will Piledrive You If You Mention AI Again Ft. Nik Suresh & Robert Evans 0:00:01.840 --> 0:00:06.600 All zone media, mother father, come quick, it's better offline. 0:00:06.640 --> 0:00:21.959 The tech podcast I'm Your host ed Zetron What. A 0:00:22.000 --> 0:00:24.560 few weeks ago, the Internet was captivated by the work 0:00:24.600 --> 0:00:28.200 of one particularly pissed off software developer who claimed, and 0:00:28.280 --> 0:00:31.040 I quote the title of his blog, I will fucking 0:00:31.120 --> 0:00:34.720 pile drive you if you mention AI again. The writer, 0:00:35.159 --> 0:00:38.640 based in Australia, lays out in detail how infuriating a 0:00:38.680 --> 0:00:41.720 distraction AI has become from the wider problems riddling many 0:00:41.720 --> 0:00:45.240 modern tech companies, as well as the noxious, nasty presence 0:00:45.240 --> 0:00:48.320 of grift and management consultant types that are deliberately keeping 0:00:48.400 --> 0:00:51.879 it inflated, regardless of whether it actually means anything. One 0:00:51.920 --> 0:00:55.080 particular quote stood out to me. You see, while hype 0:00:55.200 --> 0:00:58.440 is nice, it's only nice in small bursts. For practitioners, 0:00:58.840 --> 0:01:01.400 we have a few things that a grifter does not have, 0:01:01.520 --> 0:01:05.679 such as job stability, genuine friendships and souls. What we 0:01:05.760 --> 0:01:09.319 do not have is the ability to trivially switch fields 0:01:09.319 --> 0:01:11.880 the moment the gold rush is over, due to the 0:01:11.920 --> 0:01:14.680 sad fact that we actually need to study things and 0:01:14.880 --> 0:01:19.119 build experience. Grifters, on the other hand, wield the omnitool 0:01:19.360 --> 0:01:23.040 that they self aggrandizingly call politics. That is to say, 0:01:23.440 --> 0:01:25.880 it turns out that the core competency of smiling and 0:01:26.120 --> 0:01:30.200 promising people things that you can't actually deliver is highly transferable. 0:01:30.959 --> 0:01:33.160 This piece caught both my eye and that of Cool 0:01:33.240 --> 0:01:36.360 Zone Media's Robert Evans, who assured me he promised that 0:01:36.480 --> 0:01:38.840 I'd be allowed to leave the Cool Zone Media pain 0:01:38.920 --> 0:01:40.760 cage if I was able to get the author of 0:01:40.760 --> 0:01:43.360 the piece on the show. So that's what I did. 0:01:45.480 --> 0:01:49.280 Adjoining us to talk about this incredibly aggressive blog is 0:01:49.480 --> 0:01:53.200 Nikol Suresh, who's the director at Hermit Technology and an 0:01:53.200 --> 0:01:55.520 excellent blogger. Nick, thank you so much for joining us. 0:01:55.960 --> 0:01:58.960 Oh, thank you for the very kind words. I'm super 0:01:58.960 --> 0:02:02.040 excited to be here. When that piece came out, at 0:02:02.120 --> 0:02:03.960 least three people said, hey, have you heard of ed 0:02:04.040 --> 0:02:07.560 Citron And I was like, well, I'm talking to Tomorrow'll 0:02:07.600 --> 0:02:09.560 be excited now I will. 0:02:10.040 --> 0:02:13.120 So you're angry, I think we've kind of got that, 0:02:13.200 --> 0:02:16.680 and I share your anger. But what is it about 0:02:16.680 --> 0:02:19.560 this that really boils your blood that makes you want 0:02:19.600 --> 0:02:20.680 to pile drive folks? 0:02:21.919 --> 0:02:22.359 Yeah. 0:02:22.480 --> 0:02:26.200 So it's interesting because a lot of people read that 0:02:26.280 --> 0:02:29.120 piece and they reached out and thought that, you know, 0:02:29.360 --> 0:02:34.280 I'm actually super angry about AI specifically, but a pretty 0:02:34.320 --> 0:02:35.720 consistent through line through my blog. 0:02:35.720 --> 0:02:37.359 It's on a lot of traffic over the past year 0:02:37.360 --> 0:02:37.920 and a half. 0:02:38.440 --> 0:02:42.160 It's not about AI specifically, it's about this massive class 0:02:42.200 --> 0:02:47.000 of non technical managers that have entered the IT space 0:02:47.440 --> 0:02:51.280 and you're essentially cause playing and doing software engineering. The 0:02:51.360 --> 0:02:55.200 thing that is infuriating about the LLLM thing in particular 0:02:55.560 --> 0:02:59.239 is that it's uniquely suited for drifting. So if you 0:02:59.320 --> 0:03:02.079 look at the average company that they can't ship incredibly 0:03:02.120 --> 0:03:04.880 basic applications. And even if you look at something like 0:03:04.919 --> 0:03:07.920 the crypto space, which I don't like, but you look 0:03:07.960 --> 0:03:12.480 around Melbourne, crypto companies actually have some level of engineering discipline. 0:03:12.880 --> 0:03:15.880 You know, they use newer technology, they've got there's things 0:03:15.880 --> 0:03:18.080 in the job descriptions to indicate seriousness. 0:03:18.440 --> 0:03:18.520 Right. 0:03:18.600 --> 0:03:21.919 AI companies don't have that, and I don't think at 0:03:21.919 --> 0:03:25.480 big corporations CEOs are aware that their stoff for just 0:03:25.520 --> 0:03:28.400 typing an import open AI. You know, you're not doing 0:03:28.480 --> 0:03:32.000 any complex work. You just call like an API. You 0:03:32.040 --> 0:03:34.600 just you just send data to a company in the US. 0:03:35.280 --> 0:03:38.320 They do everything, and people down here in Melbourne are 0:03:38.840 --> 0:03:42.360 making out like they're the greatest engineers ever, their thought leaders. 0:03:42.400 --> 0:03:45.360 You should put them on comper stages. It's not hard. 0:03:45.440 --> 0:03:47.320 It's so easy. It takes like a week to make 0:03:47.360 --> 0:03:48.280 something from scratch. 0:03:49.360 --> 0:03:49.760 Yeah. 0:03:49.840 --> 0:03:52.200 I really appreciated that part of your piece because it 0:03:52.240 --> 0:03:55.400 gets at something that's like frustrated me as well, which 0:03:55.440 --> 0:03:57.200 is and I'm not coming at this from the same 0:03:57.320 --> 0:03:59.560 kind of technical background you are, but knowing that there's 0:04:00.040 --> 0:04:03.840 there's legitimate technology here, but all of the people being 0:04:04.040 --> 0:04:06.760 trusted with marshaling it and marshaling these companies that are 0:04:06.800 --> 0:04:09.560 supposed to deliver it to us are on their face. Conman, 0:04:09.600 --> 0:04:12.200 I'm thinking about Sam Altman today, or at least the 0:04:12.200 --> 0:04:14.360 clip came out today where he's like, soon you'll be 0:04:14.360 --> 0:04:18.560 able to just ask asking chat GPT to solve physics for. 0:04:18.520 --> 0:04:20.280 You, solveys. 0:04:20.720 --> 0:04:24.479 That's just a nonsense statement. That's just that's absurd. The 0:04:24.520 --> 0:04:26.680 way we know that, right, we can all tell that 0:04:26.680 --> 0:04:27.520 that's nonsense. 0:04:28.240 --> 0:04:31.440 If if you're not defaming him, that's a crazy thing. 0:04:31.720 --> 0:04:33.520 Like if you guess those are the. 0:04:33.440 --> 0:04:35.400 Woods, I'll find the exact quote. 0:04:35.440 --> 0:04:38.640 But yeah, no, he he, he was very much. That's 0:04:38.680 --> 0:04:41.359 what he said. You've spoken of this kind of like 0:04:41.560 --> 0:04:45.120 organizational cancer inside tech films? Is it this kind of grift? 0:04:45.960 --> 0:04:49.719 Yeah, yeah, it's not specific to technology. It's just the 0:04:49.760 --> 0:04:52.640 technology makes it very very easy. I don't know of 0:04:52.760 --> 0:04:55.960 any industry that doesn't have this massive horde of people 0:04:56.000 --> 0:04:58.880 who don't really know what they're doing, right, but they 0:04:59.000 --> 0:05:02.680 just kind of bulldoze peaceeople socially put it this way. 0:05:03.160 --> 0:05:05.600 So I think my blog is done two million hits 0:05:05.640 --> 0:05:09.720 in the past year and a million realized. And when 0:05:09.800 --> 0:05:12.320 people talked about, you know, offering me work or anything, 0:05:13.440 --> 0:05:16.560 they said they would give me a reference at their workplace, 0:05:16.960 --> 0:05:19.839 but they wouldn't mention that piece because leadership would be 0:05:19.920 --> 0:05:22.720 so embarrassed, like they can't have someone like me around 0:05:23.000 --> 0:05:26.680 going to someone yeah somewhat cool things, it says, fuck 0:05:26.760 --> 0:05:30.279 sometimes can't have that. But no, it's just because like 0:05:30.320 --> 0:05:33.440 the Emperor very clearly has no clothes. I don't even 0:05:33.480 --> 0:05:36.120 really get hate mail because people know I'll just point 0:05:36.160 --> 0:05:39.800 out that they're drifting and they'll have no defense, so 0:05:39.839 --> 0:05:41.280 they just leave me alone. 0:05:41.440 --> 0:05:43.880 Yet, So, how do you think that this has happened? 0:05:43.880 --> 0:05:46.359 It clearly isn't something that happened overnight, but it feels like, 0:05:46.800 --> 0:05:50.400 especially the tech industry, has just been poisoned by product 0:05:50.440 --> 0:05:53.040 managers and management concerns. How did this happen? 0:05:53.839 --> 0:05:57.040 Off the top of my head, something that really influences 0:05:57.080 --> 0:06:00.800 this is just very strange ideas on how things scale. 0:06:01.839 --> 0:06:04.760 There is this I listened to those two episodes you suggested, 0:06:04.880 --> 0:06:07.720 and I think the main one is the Rot Combobble, 0:06:07.760 --> 0:06:11.400 which people should go listen to. And there is this 0:06:11.480 --> 0:06:15.200 idea that everything should scale up forever, and the only 0:06:15.240 --> 0:06:18.600 way to handle that communicatively is like this hierarchical structure. 0:06:19.160 --> 0:06:21.480 But it's really hard to find good people. 0:06:21.520 --> 0:06:23.560 And I think a huge hierarchy can work if you 0:06:23.640 --> 0:06:27.159 only have really great people in it, but it's really hard. 0:06:27.480 --> 0:06:28.440 There's one good. 0:06:28.320 --> 0:06:32.080 Software engineering company I know in Melbourne, Flip Down in 0:06:32.200 --> 0:06:36.480 the CBD, and they regularly interview forty people for one role, 0:06:37.120 --> 0:06:40.240 right because just the quality of candidates not there. If 0:06:40.279 --> 0:06:42.360 you insist you need this massive structure and you need 0:06:42.400 --> 0:06:44.039 a hierarchy for it, you're just going to have to 0:06:44.080 --> 0:06:46.360 insert people who don't know what they're talking about. You 0:06:46.440 --> 0:06:50.040 can't find good people fast enough, and then those people 0:06:50.040 --> 0:06:52.039 who aren't very good to hire other people who aren't 0:06:52.080 --> 0:06:53.839 very good, and soon that's the whole industry. 0:06:53.960 --> 0:06:57.760 Right, and that feels something I'd be writing that will 0:06:57.880 --> 0:06:59.960 end up being a podcast that may run before or off. 0:07:00.080 --> 0:07:02.919 This is just this idea of shareholder supremacy, that the 0:07:02.920 --> 0:07:07.320 companies have been engineered to make a market need work 0:07:07.400 --> 0:07:12.680 to hit analyst expectations. Jack Welch, classic bastard. And it 0:07:12.720 --> 0:07:16.600 feels like this that llms in particular have become the 0:07:16.720 --> 0:07:20.920 ultimate form like the super Saiyan of rock capitalism, where 0:07:21.040 --> 0:07:24.000 people just kind of, like you said, just bullshit about them. 0:07:24.000 --> 0:07:26.480 It's like, yeah, you can do all this shit, right, Yeah, 0:07:27.200 --> 0:07:30.240 it's the future. What is your actual background? 0:07:30.920 --> 0:07:33.960 So I've got a four year qualification in psychology, so 0:07:33.960 --> 0:07:37.480 I mostly did cognitive psychology and circadian science, and then 0:07:37.480 --> 0:07:39.880 I moved over to a postgraduate qualification at one of 0:07:39.920 --> 0:07:42.920 Australia's bigger universities, so that was in data science. I 0:07:43.000 --> 0:07:45.680 did Masters with a little bit of a thesis work 0:07:45.680 --> 0:07:48.880 in it, so I'm reasonably good. I'm not a PhD, 0:07:49.120 --> 0:07:50.840 but I do know what I'm talking about. 0:07:50.640 --> 0:07:51.160 In the space. 0:07:51.840 --> 0:07:53.840 I did want to drill into something because you're coming 0:07:53.880 --> 0:07:56.440 at this like from the perspective of somebody who was 0:07:56.480 --> 0:07:59.880 working in a technical capacity, seeing the way this group 0:07:59.880 --> 0:08:02.360 of people who are talking nonsense are coming in and, 0:08:02.400 --> 0:08:04.760 as you said, kind of like bulldozing their way into 0:08:04.760 --> 0:08:08.120 conversations with people. Can you talk socially about the way 0:08:08.160 --> 0:08:10.080 that kind of happens, because we all sort of see 0:08:10.080 --> 0:08:13.160 the way these people try to monopolize conversations about technology 0:08:13.200 --> 0:08:15.480 at events, you know, if you're coming at it from 0:08:15.520 --> 0:08:17.920 like press at something like CEES, but also just kind 0:08:17.920 --> 0:08:20.480 of if you're looking at things on Twitter or looking 0:08:20.560 --> 0:08:23.240 at just sort of the way popular conversations go. But 0:08:23.320 --> 0:08:27.120 I'm I'm curious about like the social dynamics of these 0:08:27.160 --> 0:08:30.440 people who are coming into organizations full of real people 0:08:30.800 --> 0:08:34.120 trying to do real things and hijacking that for the 0:08:34.160 --> 0:08:35.800 purpose of cashing out. 0:08:35.960 --> 0:08:39.120 Yeah, I think a lot of them are latching onto 0:08:39.160 --> 0:08:41.680 some pre existing level of grift that's just been in 0:08:41.720 --> 0:08:46.079 the industry. So when I entered around twenty eighteen twenty nineteen, 0:08:46.120 --> 0:08:49.160 that's when I got into the Australian tech market, MLMs 0:08:49.200 --> 0:08:51.240 were not this massive thing, right. There were a couple 0:08:51.280 --> 0:08:53.760 of GPT two demos, but there is no way for 0:08:53.840 --> 0:08:58.360 you at a normal company to roll out something like Jenai, right, 0:08:58.400 --> 0:08:59.760 you really had to know what you were doing. You 0:08:59.760 --> 0:09:02.679 didn't have the resources, like you couldn't just hit an 0:09:02.760 --> 0:09:05.840 endpoint somewhere and get a result. And I cannot emphasize 0:09:05.840 --> 0:09:09.960 that enough for like non technical listeners, it's like two 0:09:10.000 --> 0:09:13.040 lines of Python to use open AI. People making these 0:09:13.120 --> 0:09:16.560 chatbots are not serious practitioners. I could teach like my 0:09:16.640 --> 0:09:19.920 thirteen year old cousin how to do this, but there's 0:09:19.920 --> 0:09:23.400 this kind of this pre existing class of person that talkslot, 0:09:23.440 --> 0:09:26.040 gets on stages, doesn't really know what's been going on. 0:09:26.160 --> 0:09:28.520 But they were much smaller back then. They did talk 0:09:28.520 --> 0:09:32.000 about AI relentlessly. But the difference between twenty nineteen and 0:09:32.080 --> 0:09:35.760 twenty twenty four is those people in twenty nineteen didn't 0:09:35.760 --> 0:09:39.120 have any working products. They would talk and then they 0:09:39.160 --> 0:09:41.960 would leave organizations every two years before they were on 0:09:42.040 --> 0:09:45.960 the hook for their failed projects. The difference now is 0:09:46.280 --> 0:09:49.600 people are going to those same people, so you just 0:09:49.640 --> 0:09:51.040 turn up and you go, hey, I'm going to be 0:09:51.200 --> 0:09:53.079 your lead data scientist, your chief. 0:09:52.960 --> 0:09:54.320 Data officer, whatever you want. 0:09:54.840 --> 0:09:57.360 I'm going to do this AI revolution for you, and 0:09:57.400 --> 0:09:59.960 then they can very quickly ship stuff that doesn't quite work. 0:10:00.840 --> 0:10:04.280 But organizations are really bad at tracking what actually drives revenue, 0:10:04.520 --> 0:10:07.320 So the moment you get the flashy product out, people 0:10:07.320 --> 0:10:08.920 think you're some sort of superhero. 0:10:09.840 --> 0:10:12.360 Yeah. And if it's broken, yeah, absolutely so. 0:10:12.400 --> 0:10:14.880 I've been I probably shouldn't say this, but I will 0:10:14.880 --> 0:10:17.400 and just leave the details out. I've been leaked a 0:10:17.480 --> 0:10:20.920 couple of confidential documents from i would say small to 0:10:21.000 --> 0:10:24.439 medium businesses on how JENNYI has worked in their chetbots. 0:10:24.960 --> 0:10:27.000 I can't name the companies because I'm not supposed to. 0:10:27.000 --> 0:10:29.760 The documents like, it's not impressive. Even the ones that 0:10:29.960 --> 0:10:33.959 sort of work. You get really weird statistics like interacting 0:10:33.960 --> 0:10:39.240 with the chatbot predicts tickets being escalated to human support teams, 0:10:39.320 --> 0:10:40.480 like way worse than usually. 0:10:41.160 --> 0:10:43.040 But they're still rolled about, right, that's good. 0:10:43.120 --> 0:10:44.960 So the one thing that they say, this is going 0:10:45.040 --> 0:10:47.880 to be able to do customer service, very cool. 0:10:48.360 --> 0:10:49.920 I definitely can't do it now. 0:10:50.360 --> 0:10:51.559 Great Jesus. 0:10:51.920 --> 0:10:55.160 Do you get the sense they're rolling it out even 0:10:55.200 --> 0:10:58.080 though they're not saying an immediate like positive effect, because 0:10:58.160 --> 0:11:02.480 the kind of effect on their potential valuation by throwing 0:11:02.520 --> 0:11:05.600 AI in is higher than whatever they lose by having 0:11:05.640 --> 0:11:07.160 a less efficient actual solution. 0:11:07.720 --> 0:11:08.719 I think it's threefold. 0:11:09.160 --> 0:11:12.440 So the first one is it's speculative, which means even 0:11:12.440 --> 0:11:15.040 if I if I was dishonest and I thought it 0:11:15.080 --> 0:11:17.520 was bad for the company, I might want AI on 0:11:17.559 --> 0:11:20.600 my CV specifically, it doesn't actually have anything to do 0:11:20.640 --> 0:11:23.680 with the company, right, it's just for me. The second 0:11:23.679 --> 0:11:26.240 one is, yeah, there's obviously it's very easy to get funding. 0:11:26.800 --> 0:11:30.000 Someone reached out to me a CEO of a company 0:11:30.040 --> 0:11:33.280 that does LM work, and this person admitted to me 0:11:33.320 --> 0:11:35.480 their product doesn't work, so they don't accept funding because 0:11:35.480 --> 0:11:37.280 they're trying to get it to work and people are 0:11:37.320 --> 0:11:39.840 still flinging money at them. This thing that, like self 0:11:39.880 --> 0:11:41.199 admittedly doesn't function. 0:11:41.840 --> 0:11:45.400 That's very cool. Yeah, it's great. It's great that the 0:11:45.440 --> 0:11:46.480 industry is in that position. 0:11:47.160 --> 0:11:49.600 And the last one is good for them. 0:11:50.840 --> 0:11:52.679 The last one is, you know, the chatbot piece. 0:11:53.000 --> 0:11:55.840 I don't work in that domain, but I kind of 0:11:55.880 --> 0:11:59.880 wonder is it cheaper to just provide bad customer service, 0:12:00.120 --> 0:12:02.960 lay off all your supports offf Like, when I'm trying 0:12:02.960 --> 0:12:05.760 to cancel my mobile plan, do they actually care that 0:12:05.800 --> 0:12:07.440 I got frustrated and put the call down? 0:12:07.440 --> 0:12:08.199 That sounds good for. 0:12:08.160 --> 0:12:11.840 Them, right, Well, yeah, I mean that's one of the 0:12:11.840 --> 0:12:14.600 things that's so insane about this is that this could 0:12:14.640 --> 0:12:17.120 work out for them. At least in I still don't 0:12:17.120 --> 0:12:19.600 think that when you're looking kind the kind of like 0:12:19.640 --> 0:12:22.760 the dollar amount of valuations that like open Ai is seeing, 0:12:22.840 --> 0:12:25.560 I don't think that letting companies that don't give a 0:12:25.559 --> 0:12:30.200 shit about customer service economize more is going to be 0:12:30.240 --> 0:12:32.720 the size of industry that they're hoping for. You're certainly 0:12:32.760 --> 0:12:35.720 not going to It's not trillions of dollars in value, right, 0:12:35.840 --> 0:12:39.120 that could be a profitable business and just making everything 0:12:39.160 --> 0:12:41.960 a little bit worse for everybody like that, that could 0:12:42.040 --> 0:12:46.439 be a very good business to be it. Yeah, just Google, No, 0:12:47.400 --> 0:12:49.640 I mean yeah, I mean yeah, that's Google's done well 0:12:49.640 --> 0:12:50.480 off of that. 0:12:50.720 --> 0:12:54.160 It's just just making things gradually, iteratively worse. 0:12:54.760 --> 0:12:57.640 Well, I will say that I get a lot of 0:12:57.679 --> 0:12:59.720 email from people who left Google, and you know, some 0:12:59.800 --> 0:13:01.680 of them. It's a huge company, so they do some 0:13:01.800 --> 0:13:03.640 useful stuff, but a lot of people were like, I 0:13:03.679 --> 0:13:04.960 left because it sucks there now. 0:13:05.440 --> 0:13:07.400 So that's that's definitely a thing. 0:13:23.440 --> 0:13:26.600 So you said the data science field was large but 0:13:26.720 --> 0:13:30.480 largely fraudulent. What exactly does that? Does it mean that 0:13:30.880 --> 0:13:33.200 most of data science is just grifting or is it 0:13:33.320 --> 0:13:34.840 just something a little more deep. 0:13:35.360 --> 0:13:38.880 It's a little deeper than that. So as a as 0:13:38.920 --> 0:13:42.160 a technical discipline, machine learning is very very serious, right. 0:13:42.880 --> 0:13:45.839 You do see it used and genuinely helpful things. It's 0:13:45.840 --> 0:13:48.480 probably being used right now as we're recording this. It's 0:13:48.480 --> 0:13:52.320 somewhere in the supply chain, and you'll hear more old 0:13:52.360 --> 0:13:55.160 cool practitioners. They get very annoyed at the words AI 0:13:55.200 --> 0:13:57.480 and machine learning. They say things like, it's just statistics, 0:13:58.880 --> 0:14:01.600 you know it, Ellen are obviously doing something. It would 0:14:01.679 --> 0:14:04.040 be wrong to just say they're classical statistics and not 0:14:04.120 --> 0:14:08.320 impressive in any way. But when I say it's grifting, 0:14:08.440 --> 0:14:12.119 I mean this whole artificial intelligence thing. The people deciding 0:14:12.200 --> 0:14:15.000 on the size of the data science market are not technicians, 0:14:15.320 --> 0:14:18.640 so when you say AI, they're thinking Skynet, generative AI, 0:14:19.680 --> 0:14:22.560 general intelligence. Like they really have no idea what they're 0:14:22.600 --> 0:14:25.720 hiring for. But they'll do things like say we need 0:14:25.720 --> 0:14:28.080 to hire twenty data scientists tomorrow, right, like that's. 0:14:28.040 --> 0:14:31.080 Right, without really knowing what they do with them, no. 0:14:31.040 --> 0:14:33.080 Idea what they do with them. And it's really hard 0:14:33.120 --> 0:14:36.560 to solve problems with data science, right. It is entirely 0:14:36.600 --> 0:14:42.000 possible to have a problem that is fundamentally intractable, no 0:14:42.040 --> 0:14:45.200 matter how discipline your team is because the data doesn't exist, 0:14:45.640 --> 0:14:48.200 or the problem's just too hard mathematically. 0:14:48.280 --> 0:14:49.240 That happens all the time. 0:14:50.360 --> 0:14:52.320 It almost feels like the problem with the tech industry 0:14:52.400 --> 0:14:56.800 is that nobody knows what they're doing. You have people 0:14:56.840 --> 0:15:01.920 that know stuff, how to do things, and then they 0:15:02.360 --> 0:15:05.040 get a job at Google in a pile of people 0:15:05.120 --> 0:15:09.600 that can do stuff run by someone who goes all right, dickheads, 0:15:09.920 --> 0:15:13.560 put the AI in it now, and then stuff comes out. 0:15:13.640 --> 0:15:15.800 It almost like the problem you mentioned earlier of just 0:15:16.040 --> 0:15:19.160 they're not being enough good people also appears to be 0:15:19.320 --> 0:15:20.640 there are no good managers. 0:15:21.280 --> 0:15:23.880 I wonder kind of pivoting or not pivoting, But building 0:15:23.920 --> 0:15:25.880 on that, I wonder how much of it is that 0:15:25.920 --> 0:15:27.840 the guys who are really good at keeping the hype 0:15:27.840 --> 0:15:30.840 going are also just better socially than the kind of 0:15:30.840 --> 0:15:33.040 people who are really good at coding and are really 0:15:33.080 --> 0:15:36.360 good at the actual like science, Like if some of 0:15:36.400 --> 0:15:39.800 the being bulldozed effect is that the people who really 0:15:39.800 --> 0:15:42.400 know their shit just are not used to the kind 0:15:42.480 --> 0:15:45.840 of energy that a hype man is capable of deploying 0:15:45.840 --> 0:15:46.680 against them. 0:15:47.040 --> 0:15:49.760 Yeah, so this has come up. I've thought about this 0:15:49.840 --> 0:15:53.280 a lot. So that's a kind of component of engineering. 0:15:53.320 --> 0:15:56.440 I work on a lot not hyping myself up, but 0:15:56.640 --> 0:15:59.960 actually being able to hold conversations. But that's not quite 0:16:00.120 --> 0:16:03.840 the attribute I see in the managers who do this 0:16:03.960 --> 0:16:06.960 kind of thing, because what you're describing is like a 0:16:07.040 --> 0:16:10.400 pure conman. And at least and not to be mean 0:16:10.400 --> 0:16:12.400 to people who aren't comment but I'm leaking, like at 0:16:12.440 --> 0:16:14.840 least there's some skill in that. But what we're talking 0:16:14.880 --> 0:16:15.960 about is they. 0:16:15.880 --> 0:16:19.360 Just like this is where Steve Jobs comes in, right, 0:16:19.400 --> 0:16:21.600 there's a version of this that requires talent. 0:16:21.800 --> 0:16:25.720 Right right, Like it may not be talented, I respect 0:16:25.840 --> 0:16:28.240 like morally, but I'm like that it's an actual skill set. 0:16:28.480 --> 0:16:31.440 But we're talking about people who have really just learned 0:16:31.840 --> 0:16:35.680 how to like mimic the trappings of success and responsibility. 0:16:36.400 --> 0:16:39.240 You know, they're real big on suits, they're real big 0:16:39.360 --> 0:16:46.360 on talking very slowly with long pauses, but like they 0:16:46.400 --> 0:16:47.480 don't have novel thoughts. 0:16:48.960 --> 0:16:50.240 Their brain has just had. 0:16:50.080 --> 0:16:53.280 Like Forbes magazine flashed onto it like there's nothing else there. 0:16:55.400 --> 0:16:58.080 So maybe talking about social skill and bulldozing, I'm. 0:16:57.920 --> 0:17:03.760 Not talking about the war, right, and you know, we're 0:17:03.760 --> 0:17:06.520 not talking about people who are doing this very savvy 0:17:06.640 --> 0:17:09.120 political piece where they finesse things, and you know, it's 0:17:09.240 --> 0:17:13.800 very machiavellian. We're talking about people who just turn up 0:17:14.040 --> 0:17:15.520 and if you point out they don't know what they're 0:17:15.560 --> 0:17:16.480 talking about. 0:17:16.840 --> 0:17:19.520 They're a fucking dick to you. And they do this 0:17:19.640 --> 0:17:22.280 until you leave them alone, and that's their whole career. 0:17:22.400 --> 0:17:26.000 The industry is just full of that. They're not even smart. 0:17:26.040 --> 0:17:28.960 I have seen an executive at a restaurant order the 0:17:29.000 --> 0:17:32.400 wrong dish, have their family eat the whole thing, then 0:17:32.480 --> 0:17:34.760 tell the weight stuff that like, hey, you brought us 0:17:34.800 --> 0:17:37.080 the wrong thing, and when a kid piped out, they 0:17:37.119 --> 0:17:37.880 shushed the kid. 0:17:38.680 --> 0:17:38.919 You know. 0:17:39.080 --> 0:17:43.320 It's it's not like a talent that they have cultivated 0:17:43.320 --> 0:17:46.840 in the menagerial space. It's like personality disorder that just 0:17:46.960 --> 0:17:48.040 leads into everything. 0:17:47.720 --> 0:17:50.040 In ruins in a reality. 0:17:51.800 --> 0:17:54.080 How I want. 0:17:52.920 --> 0:17:56.840 It's some of it is because like I talk about 0:17:56.920 --> 0:17:58.879 jobs a lot, and I think you have to because 0:17:59.119 --> 0:18:01.120 a lot of these guys, if not all of them, 0:18:01.440 --> 0:18:04.119 at some level wish they were him, right with the 0:18:04.160 --> 0:18:06.879 exception of the not you know, the medical issue, but 0:18:06.960 --> 0:18:10.520 like they's he's like mythologized to a significant degree. Especially 0:18:10.680 --> 0:18:13.800 I'm recording from San Francisco right now. I mean He's 0:18:13.800 --> 0:18:17.840 almost like a saint down here, and I feel like 0:18:17.880 --> 0:18:20.440 the act of copying him, you get these people who 0:18:20.440 --> 0:18:24.280 are like carbon copies of just the personality disorder without 0:18:24.320 --> 0:18:27.000 any of the actual insight. And I think that's like 0:18:27.240 --> 0:18:30.080 further exacerbated because they're all like, yeah, read the Art 0:18:30.119 --> 0:18:33.359 of War, read like one of three Malcolm Gladwell books, 0:18:33.800 --> 0:18:36.920 Read The Four Hour Body, and do not read anything else. 0:18:37.040 --> 0:18:39.920 There's no humanities, there's nothing like because all you need 0:18:39.920 --> 0:18:42.280 to learn how to do is be the most optimized 0:18:42.320 --> 0:18:43.240 asshole you can be. 0:18:44.040 --> 0:18:46.159 I like, how you just pick the If I had 0:18:46.240 --> 0:18:51.600 hit Man budget, like that would be my list as 0:18:51.640 --> 0:18:55.080 a data sideist. Malcolm Glenwell, Like I might need him 0:18:55.080 --> 0:18:56.680 more than anyone on the planet. 0:18:57.080 --> 0:19:00.000 Oh, I'm right there with you, buddy. He's he's been advertise, 0:19:00.240 --> 0:19:04.199 I got our shows and oh man, I don't love it. 0:19:04.640 --> 0:19:05.520 I don't love it. 0:19:07.040 --> 0:19:09.720 So with that in mind, Nick, what do you think 0:19:09.720 --> 0:19:10.520 of Sam Oltman. 0:19:11.200 --> 0:19:14.040 I don't really know a whole bunch about him because 0:19:14.560 --> 0:19:16.720 out here in Melbourne, I don't really play around in 0:19:16.760 --> 0:19:18.240 the fang super huge space. 0:19:18.440 --> 0:19:18.640 Right. 0:19:18.640 --> 0:19:21.680 Our clients are relatively small. I work at like medium 0:19:21.680 --> 0:19:25.720 sized businesses, tiny compared to the size of Google and 0:19:25.800 --> 0:19:30.400 open AI. All I will say is, if he said 0:19:30.440 --> 0:19:34.679 what you just said about physics, like, he needs to 0:19:34.680 --> 0:19:37.560 come out with the greatest innovation I've ever seen in 0:19:37.640 --> 0:19:39.760 the next two years, or he will be one of 0:19:39.800 --> 0:19:42.119 the most insane people I've ever heard of, right, Like, 0:19:42.160 --> 0:19:45.240 that's he really has to deliver on that statement, or 0:19:45.280 --> 0:19:46.600 he's obviously full of shit. 0:19:47.119 --> 0:19:49.439 It just it blows me away that he's even able 0:19:49.520 --> 0:19:51.679 to say it without someone just being like, what the 0:19:51.720 --> 0:19:54.879 fuck you talking about? Sam Oltman? Sam Oltman, did you 0:19:54.960 --> 0:19:57.800 hit your head somewhere on the way here? Because it's 0:19:57.840 --> 0:20:00.439 not even the first insane thing he said. He has 0:20:00.480 --> 0:20:02.960 said so many times. I was super smart. It's going 0:20:03.040 --> 0:20:05.479 to be a super smart friend that knows everything about you? 0:20:06.080 --> 0:20:08.400 The fuck are you talking? What does any of this mean? 0:20:08.520 --> 0:20:10.879 And my own frustration I've shared about the media is 0:20:10.920 --> 0:20:13.800 that no one ever seems to say the appropriate thing, 0:20:13.840 --> 0:20:17.960 which is wait, what was that? Excuse me, mate? What? 0:20:18.520 --> 0:20:21.359 But it almost feels with everything you're talking about, and 0:20:21.440 --> 0:20:24.240 in your excellent blog that also feels like the problem 0:20:24.280 --> 0:20:27.240 You've just got these guys who just say things raise 0:20:27.400 --> 0:20:32.439