WEBVTT STYLE ::cue(v[voice="EDEB1289-37EE-489E-91FE-AF0D914CDCE5"]) { color: #00C3D0; } ::cue(v[voice="1AC7D02B-029D-459B-9583-3575AAA94272"]) { color: #FF383C; } ::cue(v[voice="9A469E6A-7DCF-4D85-B605-281EAB20BF64"]) { color: #CB30E0; } ::cue(v[voice="4C565A47-80DA-4EAF-A640-3A47230B8E62"]) { color: #6155F5; } ::cue(v[voice="91CAF432-16BC-48DE-8182-B7E7F0CF7766"]) { color: #34C759; } 1 00:00:00.000 --> 00:00:20.000 [Music] 2 00:00:10.500 --> 00:00:15.960 This is GamesItWork.biz, your weekly podcast about gaming, technology, and play. 3 00:00:15.960 --> 00:00:19.399 Your hosts are Michael Martine, Andy Piper, and Michael Rowe. 4 00:00:19.399 --> 00:00:23.000 The thoughts and opinions on this podcast are those of the hosts and guests alone, 5 00:00:23.000 --> 00:00:28.000 and are not the opinions of any organization which they have been, are, or may be, affiliated with. 6 00:00:33.500 --> 00:00:37.020 This is episode 540, saucer, separation button. 7 00:00:38.500 --> 00:00:41.079 [upbeat music] 8 00:00:47.000 --> 00:01:04.760 Hello everyone and welcome again to another episode of GamesItWork.biz, your weekly technology podcast. This is Michael Martine, Freshly Back from Terror Hote, and I'm glad to be back reunited with Mike Podcast co-host Michael Rowe and Andy Piper, and how you're doing these days. 9 00:01:06.000 --> 00:01:10.359 I'm well Michael, thanks for asking and say Tera Ho. 10 00:01:10.500 --> 00:01:12.500 Terahote rhymes with boat 11 00:01:11.799 --> 00:01:13.040 Tera Ho. 12 00:01:13.040 --> 00:01:14.480 Oh, okay. 13 00:01:13.780 --> 00:01:16.640 It's an Indiana of all places so 14 00:01:14.480 --> 00:01:15.319 Okay. 15 00:01:17.420 --> 00:01:18.260 Wow. 16 00:01:18.659 --> 00:01:21.280 The things you learn on a podcast like this, right? 17 00:01:19.239 --> 00:01:20.079 Nice trip. 18 00:01:21.000 --> 00:01:23.000 It's not terahoot and ryan's with boot? 19 00:01:23.319 --> 00:01:29.560 I just wanna give a shout out to Michael Rowe for keeping things rolling last week with our good friend E-Pread. 20 00:01:30.500 --> 00:01:48.340 Yeah. Ian was awesome. I was very glad to find Ian and not find him, but get him, convince him to come back and have fun with us. It was loads. Yes. Yes. Yes. 21 00:01:30.519 --> 00:01:32.640 Since neither of us could be part of the show. 22 00:01:40.000 --> 00:01:48.319 to drag him kicking and screaming as a retired person to have to express an opinion on a topic or three. 23 00:01:43.000 --> 00:01:45.000 Yes 24 00:01:47.200 --> 00:01:52.039 Yeah, you know, you know the thing that struck me as I was listening to the episode is like, you know 25 00:01:52.799 --> 00:01:57.519 Now that he's retired I wonder how busy he might be hmm 26 00:01:58.000 --> 00:02:07.840 Why do you think? No, that's not why I asked him, but hey, yeah, he could. Maybe like a third part of a novel trilogy. 27 00:01:58.500 --> 00:02:02.500 He could always spend some more time writing I guess. 28 00:02:09.520 --> 00:02:13.020 So, we have got some fun links to share this week. 29 00:02:13.000 --> 00:02:23.000 And Andy, we're going to start off with you because you have a new piece of shiny new equipment that you've gotten off of Kickstarter, which kind of kickstarted our run of show. 30 00:02:23.000 --> 00:02:30.000 So, let's talk about this great device you have for your phone. 31 00:02:28.500 --> 00:02:49.120 That noise was my M-con controller, which is a MagSafe connectable, or magnetically attachable controller for your mobile device, so it's pretty much the same footprint as my iPhone, 32 00:02:29.000 --> 00:02:32.979 The thing that you're showing, ooh, yes. 33 00:02:30.000 --> 00:02:32.000 Ooh, did you hear that? 34 00:02:49.120 --> 00:02:58.460 whatever version I've got, the one that's two years old, 15 something like that, connects on the back using mech safe. Really nicely even though I'm 35 00:02:58.500 --> 00:03:01.039 where it's got a case on my phone. 36 00:03:01.039 --> 00:03:04.960 And you press this button, it pops up so that the underneath the phone slides out, 37 00:03:04.960 --> 00:03:12.240 a standard control pad with a D-pad and two joysticks left and right thumbsticks. 38 00:03:12.240 --> 00:03:15.439 And then you have four buttons on the face, 39 00:03:15.439 --> 00:03:18.240 and then there's also four shoulder buttons. 40 00:03:18.240 --> 00:03:21.639 You can fold out the little wings. 41 00:03:21.639 --> 00:03:25.319 So it's more like a regular console controller. 42 00:03:25.319 --> 00:03:27.280 A bit more comfortable to hold that way. 43 00:03:27.280 --> 00:03:28.479 And the other thing you can do is to test. 44 00:03:28.500 --> 00:03:40.500 And you can attach the top section, which is the magnetic part, and prop your iDevice or mobile device on the table with a little kickstand instead if you prefer. 45 00:03:40.500 --> 00:03:44.500 So yeah, it's quite nice. 46 00:03:41.000 --> 00:03:44.240 So, so I haven't asked, does it work with Android? 47 00:03:44.500 --> 00:03:46.500 It does, yes. 48 00:03:46.500 --> 00:03:54.500 They attach, in case you don't have a maxa for a key magnet on the back of your device. 49 00:03:54.500 --> 00:03:58.460 they've added one of those all easy to stick on rings. 50 00:03:58.500 --> 00:04:18.399 I have to say I think this was a kickstarter that was about 18 months ago, I backed it and it's taken a long time and they did some really nice updates through the campaign showing the iterations of the thing and yeah, it's really nice device. 51 00:04:18.399 --> 00:04:28.420 The only thing is that a lot of mobile games are deliberately made for portrait mode because they're more casual games for when you're waiting for the bus or whatever and just 52 00:04:28.500 --> 00:04:41.860 wiping away, rather than the landscape mode that this expects which has been my one annoyance finding, easily finding the games that it supports or that support it. 53 00:04:42.000 --> 00:04:56.439 So, sticking on this theme of controllers, there's another one that we were chatting about a minute or two before getting into the show itself, and this is the Abernich's wireless controller, Abernich, Abernich. 54 00:04:54.000 --> 00:05:20.000 Anvanic. Yes. Anvanic. They make a lot of retro devices for these fairly small consoles that you can play retro games on these days, although. So this is apparently, this is a story on the verge that actually just came out today as we record about the new Anvanic controller and it's 55 00:05:08.000 --> 00:05:11.000 And it has a very special feature, doesn't it? 56 00:05:20.560 --> 00:05:23.839 you know, a fairly standard looking console. 57 00:05:24.160 --> 00:05:27.199 controller, not dissimilar to what I've just been talking about with the phone. 58 00:05:28.560 --> 00:05:45.120 And it's not the first one that has a screen in the middle between the thumbsticks or the other controls. Of course, we can go all the way back to the Sega Dreamcast for that kind of aesthetic where you had a little display in the middle, but this one, as well as showing, you know, 59 00:05:39.000 --> 00:05:39.839 Check out! 60 00:05:45.120 --> 00:05:52.079 your battery charge and various other bits of information on there, it will show your heart rate. 61 00:05:53.500 --> 00:05:59.879 I guess so you can calm down in your first person shooter. 62 00:05:54.000 --> 00:05:58.560 Exactly. Yes. 63 00:05:59.879 --> 00:06:18.920 You know, the one thing I was thinking about with the availability of various retro arcade tools on mobile, I'm wondering if any of those would, by definition, support a gamepad. 64 00:06:18.920 --> 00:06:23.439 And therefore, if you ran one of those arcade simulators, if it would give you more 65 00:06:24.000 --> 00:06:53.839 one of the things that when I went into Apple Arcade on my phone it was recommending was the retrocade okay simulator game yes I don't know whether I also went to one that said it supported a controller and it was immediately trying to do things in portrait mode so in principle is you still need to tell the games framework on on iOS to to pump the the game gamepad through in the case of the ambient act. 66 00:06:55.000 --> 00:07:01.000 This is apparently incorporating the heart rate sensor going back to this story. 67 00:07:01.000 --> 00:07:10.000 Into the actual grips of the controller, which I think is quite interesting. 68 00:07:10.000 --> 00:07:15.000 There's a few other interesting elements to the device. 69 00:07:15.000 --> 00:07:22.000 I'm wondering whether people will be interested in integrating the heart rate sensor into games, 70 00:07:22.000 --> 00:07:24.000 or whether it will just be used for something. 71 00:07:24.000 --> 00:07:37.000 I haven't done any more than just say it's coming soon, so it's a bit of a gimmick at the moment. 72 00:07:37.000 --> 00:07:40.000 But it also doesn't. 73 00:07:37.500 --> 00:07:42.339 Well, they're not a development shop though. Are they? Does dampernik write games as well? 74 00:07:40.000 --> 00:07:46.000 Ambonic? No, no, no, they make a lot of hardware, yeah, they make a lot of hardware. 75 00:07:43.139 --> 00:07:45.139 They're just a hardware manager 76 00:07:46.459 --> 00:07:49.899 So I mean if you if you think about all the other you know 77 00:07:50.779 --> 00:07:56.860 Not controllers, but devices that I'll read other information health information whether it's a watch or ring or what have you? 78 00:07:57.420 --> 00:08:08.259 In theory those data streams exist and an enterprising developer might be able to then incorporate that stream into doing something you know like for example 79 00:08:07.500 --> 00:08:28.500 To suggest rest breaks or to leverage kind of like with Pokemon Go that you want to get your heart rate up a bit you know to maybe unlock a certain feature at a certain location because you're now exercising and I can tell yeah 80 00:08:19.500 --> 00:08:43.940 Or Pokemon sleep, which I've been playing now for two years consistently and using it as, you know, because I already, I haven't collected more, I don't think it's actually possible to collect them all, I'm convinced that the reason they keep bringing out new Pokemon games is so that you'll always have to try collecting more, is that not the reason? 81 00:08:31.000 --> 00:08:34.379 Have you clicked to them all? 82 00:08:34.379 --> 00:08:36.879 OK. 83 00:08:43.500 --> 00:08:45.500 novel. 84 00:08:46.500 --> 00:08:57.220 Well, the next set of articles that we've got here are all talking about how do you solve the puzzles and how do you do them with a little bit of help from your, in this case, AI friends? 85 00:08:56.500 --> 00:09:03.220 Well, is it? I'm curious to tell your interpretations of both of these stories. Go on. 86 00:08:57.220 --> 00:08:59.379 So this one, is it? 87 00:09:04.100 --> 00:09:10.620 Well, the first one is around letting Claude play text adventures and we've had over the last, 88 00:09:11.740 --> 00:09:15.059 we'll just say, is it two years, it's been a while. 89 00:09:12.000 --> 00:09:16.799 two years. I think it's two years. 90 00:09:15.059 --> 00:09:16.500 you know, various examples. 91 00:09:16.500 --> 00:09:36.500 examples of playing info comp games with large language models and having them kind of sort their way through things and text adventures are certainly really well suited to this kind of thing because it is essentially a chat interface, the game being one end of the chat, the LLM could be the other end of the chat. 92 00:09:36.500 --> 00:09:46.500 And for those that playing a game is rough and you want to twitch type of experience, you could say, alright, you know, hey Claude, use my credits and go play this game. 93 00:09:46.500 --> 00:10:14.500 And let's see how you do so I we talked about this for two years Michael like you said that this is a good test to see how the experience goes from a large language model and how sophisticated can it be and how far can it get versus you know, is it just sort of flailing around and not knowing to go down the stairs or pick up the candlestick or the sword or whatever it might be because many of these. 94 00:10:14.500 --> 00:10:16.500 older and newer. 95 00:10:16.500 --> 00:10:46.139 So that was going through my mind to begin with, but I really liked the way that this particular article from Borretti.me explains how to go about doing it. 96 00:10:46.500 --> 00:11:03.700 It gives now some instruction sets about how might you use the harness, how might you connect the LLM, and then how might you be able to take it even further to maintain and memory, what with the experience, and maybe draw some things out to create maps that weren't 97 00:11:06.000 --> 00:11:10.200 So when they brought out the open source, 98 00:11:10.200 --> 00:11:14.039 what was the games that Microsoft recently open source from Infocomp? 99 00:11:15.000 --> 00:11:17.000 Happy with the Zork, yeah. 100 00:11:16.759 --> 00:11:18.279 Zork, thank you, of course. 101 00:11:18.279 --> 00:11:20.519 Classic Zork, when they, when they did that, 102 00:11:20.519 --> 00:11:33.639 I downloaded a Python interpreter and started messing around with that and I made a little desktop UI in Python so I could play it with a sort of graphical wrapper and things like that. 103 00:11:33.639 --> 00:11:34.639 That was quite fun. 104 00:11:34.639 --> 00:11:36.000 There's a programming. 105 00:11:37.000 --> 00:11:39.000 And of course, 106 00:11:39.000 --> 00:11:47.000 giving the LLM the context for how to play these kinds of games is quite straightforward, 107 00:11:47.000 --> 00:11:49.000 especially with the older games. 108 00:11:49.000 --> 00:11:50.000 Even with Zork, 109 00:11:50.000 --> 00:11:54.000 you had a very sophisticated range of things that were possible. 110 00:11:54.000 --> 00:12:00.000 That was one of the joys of Zork and discovering how the game could be. 111 00:12:00.000 --> 00:12:03.000 The tools could be used. 112 00:12:03.000 --> 00:12:05.840 So I do think, I agree, I think, fanander bretties. 113 00:12:06.000 --> 00:12:23.000 But blog post here is really interesting to talk about how he implemented this using a Python wrapper, and being a two-dimensional, essentially a two-dimensional game, with a text interface. 114 00:12:23.000 --> 00:12:32.000 It's interesting. The cost of running it and letting it get creative is quite challenging, it seems, which is not completely surprising. 115 00:12:32.000 --> 00:12:36.039 I am a little bit tempted to have a go mess around with this myself. 116 00:12:36.000 --> 00:12:40.639 I don't particularly want to spend a load of AI tokens on doing that. 117 00:12:41.500 --> 00:12:51.419 I'm wondering if you could take this and restructure the code to run against a local Olama instance. 118 00:12:50.500 --> 00:13:12.899 Oh sure, you definitely could. Yeah, yeah. You definitely could. It will all be about the efficiency and effectiveness of the models, right? And the commercial ones typically have been the more successful ones. You could also train your local model to more specifically understand how to do the game without having to use more of a general purpose one. 119 00:12:51.419 --> 00:12:53.480 So you're not spending any money on tokens, right? 120 00:13:00.000 --> 00:13:05.000 Yeah. I... 121 00:13:13.500 --> 00:13:15.700 There was an article this week, 122 00:13:15.700 --> 00:13:18.860 and I forgot to put it in the show, Lynx, 123 00:13:18.860 --> 00:13:28.820 about it was a Git repo and some of the things on how to have Claude code running locally on your infrastructure. 124 00:13:29.000 --> 00:13:33.240 But it went me. I can't be- can't be correct. Right. 125 00:13:30.980 --> 00:13:32.740 It can't be the whole thing, 126 00:13:32.740 --> 00:13:36.179 but I'm gonna find it just for us, 127 00:13:36.179 --> 00:13:38.940 maybe we can read it and talk about it next week. 128 00:13:38.440 --> 00:13:42.279 Hmm. I- I- 129 00:13:38.940 --> 00:13:41.620 But that way we're all knowledgeable on it, 130 00:13:41.620 --> 00:13:43.500 But because it was really interesting. 131 00:13:43.500 --> 00:13:49.899 And I want to give it a shot given the capacity of my server here and figure that I'd give it a shot. 132 00:13:50.500 --> 00:13:51.500 1-1 shot. 133 00:13:50.500 --> 00:14:10.980 I threw in both, I threw in both of the, this link that we've just been talking about the text adventure AI one and also the, the one I'm about to mention, which is also using Claude, Claude code in this case as the tool to drive roller coaster taking. 134 00:14:10.980 --> 00:14:20.259 I threw in both of these stories because I'd come across them both during the week at separate times and I thought they fitted our general themes. 135 00:14:11.000 --> 00:14:13.419 [humming] 136 00:14:20.500 --> 00:14:26.139 Of the show relatively well in the, they talk about games, they're both in this case, 137 00:14:26.139 --> 00:14:37.899 kind of retro classic games and whether we like it or not, we've been forced over the last couple of years to confront the reality that is LLM's and these AI tools. 138 00:14:37.899 --> 00:14:50.220 And equally I found this story really interesting, it's presented beautifully in a kind of classic Mac ish windows, ish window style webpage. 139 00:14:50.500 --> 00:15:07.580 With the isometric roller coaster tycoon art animated in the background and it's from a company called ramp and they talk about how they enabled Claude to run roller coaster tycoon. 140 00:15:07.580 --> 00:15:14.139 It turns out that this is of an open source game at this point and there is a CLI interface. 141 00:15:14.139 --> 00:15:44.139 One of the things that it shows, Claude is not good at, is the Vertica 142 00:15:20.500 --> 00:15:37.860 company, the three dimensionality of the game because it can't, it's used to the sort of flat surface and the management tasks of operating it but it can't do what we can do as humans and creatively come up with building taller and awesome roller coasters. 143 00:15:37.860 --> 00:15:40.700 So I thought that was particularly interesting. 144 00:15:40.700 --> 00:15:50.460 They also had to do a lot more in terms of modding open roller coaster tycoon to enable this interaction with Claude and again. 145 00:15:50.500 --> 00:15:55.919 I found that really quite interesting to see how they did it. 146 00:15:55.919 --> 00:16:01.899 Although they also pointed out that they vibe coded the interface itself, which was a little bit disappointing. 147 00:16:04.500 --> 00:16:16.860 I don't know, I love this because the idea behind a roller coaster tycoon or civilization or any of these games is that there are so many variables to try to optimize. 148 00:16:16.860 --> 00:16:20.740 And so the task in hand here was go optimize them all, right? 149 00:16:20.740 --> 00:16:26.419 In roller coaster tycoon, you want your visitors to your park to be happy. 150 00:16:26.419 --> 00:16:34.460 You want them to not wait in long lines, you don't want them to be throwing up from going on to the roller coasters that make them go crazy. 151 00:16:34.500 --> 00:16:42.379 And then whatever happens gets cleaned up quickly enough because you don't want your park to be messy, so I mean, and it goes on and on and on. 152 00:16:42.379 --> 00:16:43.779 There's so many of this. 153 00:16:43.779 --> 00:16:59.500 So this is such a great example of where a game environment where all the variables are known, which is a simulation of reality, which you cannot know all the variables in reality, but you can in a game to say, go optimize them all, go write bug reports and fix them. 154 00:16:58.000 --> 00:17:29.759 and going back to our original thought processes when we started recording these shows way back in the day it's one of those commands interfaces where you know you're learning through play how to optimize and manage large environments you've got that rich data you've got those dashboards you can see and as the game goes on as you say it gets more complicated to manage all of those tasks of course now the brilliant mine 155 00:17:28.000 --> 00:17:56.480 in the west coast of America at the moment are insisting that this should all be automated away and we shouldn't have fun but there you go that's that's right right right right right this this week's show brought to you by my I reading habits. 156 00:17:38.500 --> 00:17:42.720 Well, that's why we have agents to have fun for us so we can enjoy the agent having fun. 157 00:17:42.720 --> 00:17:56.440 Now, Andy, you found this isometric New York City version and very similar to, you know, the 8-bit and structure and SIVS. 158 00:17:56.440 --> 00:18:06.279 Yeah, no, I'm glad for it and I love it and I was zooming in, you know, really small in certain places where I could see the various forts. 159 00:18:03.000 --> 00:18:07.519 But yeah, it doesn't zoom enough though. 160 00:18:04.000 --> 00:18:08.000 Now, this is another, this... 161 00:18:07.519 --> 00:18:17.400 I've zoomed all the way in and I recognize some parts in New York quite well because there's like a restaurant in the park that I went to and I found it but yeah, it doesn't zoom 162 00:18:17.500 --> 00:18:23.700 They've ended up quite high on hack and use and lobsters this week. 163 00:18:23.700 --> 00:18:31.180 It's an isometric New York City where the person who created it had the notion of, "Well, 164 00:18:31.180 --> 00:18:47.299 again, there's AI now," and we can use that to recreate classic kind of graphics similar to what we're actually looking at in open rollercoaster tycoon, or as Michael says, in You know, those 1990s pixel art. 165 00:18:47.500 --> 00:18:51.299 You know, the version of representations of cities. 166 00:18:51.299 --> 00:18:53.900 What if I could do that for the whole of a real place? 167 00:18:53.900 --> 00:18:57.619 So again, I found the right up of the process, 168 00:18:57.619 --> 00:18:59.460 particularly fascinating. 169 00:18:59.460 --> 00:19:01.900 This is described. 170 00:19:01.900 --> 00:19:05.220 Different ways of finding map data and tile information. 171 00:19:05.220 --> 00:19:17.579 In fact, he ends up using Google Maps 3D tiles and then having to train an LLM so that it's optimized for the task of creating this pixel art and a consistent 172 00:19:17.500 --> 00:19:47.420 manner so that it didn't all look completely disparate as you put peel piece together the whole of New York City. I found it also really interesting that he struggled most with green areas presumably central park primarily trees and and the water and he ended up just kind of throwing his hands up and giving up trying to to do that. He also talks about how complicated it is to manage that zoom in. 173 00:19:47.500 --> 00:19:54.500 You've just described and the over-tiling and depending on what you're looking at. 174 00:19:54.500 --> 00:20:03.500 I think it's cool to look at, for me, intellectually, it was again more interesting, similar to the last two stories. 175 00:20:03.500 --> 00:20:08.500 Thinking about what, thinking about the process of doing it. 176 00:20:08.500 --> 00:20:17.500 And I think it all three cases, which are using this AI, LLM technology, which we will have complicated feelings about. 177 00:20:18.500 --> 00:20:27.500 It's enabled interesting thought processes for the human behind it, as much as it has burned a bunch of tokens. 178 00:20:27.500 --> 00:20:35.500 And I really enjoyed those thought processes, as well as enjoying the aesthetic outputs. 179 00:20:35.500 --> 00:20:42.500 I'm frustrated that it all took loads of language models to achieve. 180 00:20:42.500 --> 00:20:48.099 Well, I mean, we've seen examples of this where humans have built in Minecraft and elsewhere, 181 00:20:44.500 --> 00:20:53.500 Yes, Sam. 182 00:20:48.099 --> 00:20:58.660 you know, similar kinds of things, right? And the digital twin, they're not sure. And but where I was going to go is that there's a digital twin kind of construct, 183 00:20:58.660 --> 00:21:05.059 whether we're talking about a city or a manufacturing line or what have you, that can be represented, 184 00:21:04.000 --> 00:21:12.759 I would like to ask Michael Rowe what his yes and is because I'd like to hear all of the angles here. 185 00:21:05.059 --> 00:21:06.259 Right? Yes, Sandy. 186 00:21:13.500 --> 00:21:27.259 Yeah, my yes, and I think the aspect of this that I found interesting was the constraints that you put in place, right, of how it was built. 187 00:21:24.500 --> 00:21:27.000 - Mm-hmm, yes. 188 00:21:27.259 --> 00:21:35.059 And then also the constraints in the experience, right, because my first reaction having played around with things like this all the time was, I want to turn it. 189 00:21:35.059 --> 00:21:37.819 I want to see the front of that building can't do any of that. 190 00:21:37.819 --> 00:21:43.500 It is truly just an image that you can zoom in that's giving you the isometric perspective. 191 00:21:42.000 --> 00:21:44.000 Yeah, that's a good point 192 00:21:43.500 --> 00:22:13.460 And then looking at the perspective of creating this and the point that you're making on the amount of tokens, there's so many ways that this type of thing has been done in the past to your Minecraft example is having AI do this specific task, proving something else that makes it valuable, or is it just, I want to see if an AI could redo something that somebody else has done, and we've got bazillions of examples of that because it has consumed that knowledge already and it is regurgitating. 193 00:22:13.500 --> 00:22:15.259 - Right, exactly. 194 00:22:13.500 --> 00:22:19.900 So the uniqueness of the experiment that was done here is what I'd like to understand more of. 195 00:22:17.460 --> 00:22:18.299 - Yes. 196 00:22:20.220 --> 00:22:23.380 I think that's a really good point. 197 00:22:23.380 --> 00:22:25.339 Yes, of course, I mean, since 32,000, 198 00:22:25.339 --> 00:22:27.059 which I think is what this is kind of modeled on, 199 00:22:27.059 --> 00:22:28.259 that's that graphical style, 200 00:22:28.259 --> 00:22:29.740 you also couldn't rotate the camera, 201 00:22:29.740 --> 00:22:32.539 you just had that one isometric view. 202 00:22:32.539 --> 00:22:41.339 But it's not like any of these three individuals have said, here's my entire idea to an AI model 203 00:22:41.420 --> 00:22:45.140 had it. Give it back. 204 00:22:43.500 --> 00:22:51.500 The fully completed version and such that there was no human in the loop and there was no iteration and process. 205 00:22:51.500 --> 00:23:07.500 There's some sophisticated, well, I hope and believe that we will never be there because I think that the creativity and the iteration and the what-ifs comes from the human mind and not from the machines here. 206 00:23:06.000 --> 00:23:12.819 Yeah, it's it's funny I've I've been in discussions with people talking about how AI 207 00:23:07.500 --> 00:23:09.019 Yeah, I... I... 208 00:23:14.279 --> 00:23:23.700 Gets them sparked creatively and like well It may be able to provide you examples of a bunch of stuff, but it's not creative. You are being creative 209 00:23:25.000 --> 00:23:33.500 Well, you're guiding it much in the same way that you're guiding a trowel as you're laying the brick on the wall, right? 210 00:23:33.000 --> 00:23:34.400 Another brick in the wall. 211 00:23:33.500 --> 00:23:45.299 You know, I realize, and I understand that a landscape is a changing and evolving thing, and it may not be evolving and changing minute by minute, but it certainly changes. 212 00:23:34.400 --> 00:23:36.440 [laughing] 213 00:23:45.299 --> 00:23:50.099 So when I zoomed in on the airport, for example, there's all the construction going on at the airport. 214 00:23:50.099 --> 00:23:55.019 So, if you link this example back to the story. 215 00:23:55.000 --> 00:24:04.000 We're just talking about with roller coaster tycoon, now you can do an awful lot of interesting modeling around what if I wanted to do traffic flow optimization? 216 00:24:04.000 --> 00:24:23.000 What if I were able to see data points in more detail? Can I now change which traffic lights are doing where I put other traffic lights in to increase the flow of traffic and decrease wait times for people as they're trying to get around a city? 217 00:24:23.000 --> 00:24:24.940 There's so many ways that 218 00:24:25.000 --> 00:24:37.000 something like this allows for the same kind of digital twinning that the Nvidia guys had had been working on from the, what do they call it, omniverse, I think, something like that. 219 00:24:37.000 --> 00:24:46.000 That the creation of this, and I do you remember how long it took to create this, Andy, from your right up? Don't remember. 220 00:24:45.500 --> 00:24:52.859 No, I don't. I think it was quite a long time. It was a period of, yeah, I'm going to have to go back and read it now. 221 00:24:46.000 --> 00:24:55.000 Yeah, but a while, but a while, right? 222 00:24:52.859 --> 00:24:54.299 Carry on. Yeah. 223 00:24:55.000 --> 00:25:20.000 So, you know, there's the what ifs? What if this could be done in generalistic real time? So now you can have people who are able to synthesize with their eyes and understand things and zoom in and out and what if you could rotate it like Michael, you want and kind of dial into the real buildings and what if you use this now to say, you know, I want to open a new restaurant in New York City, where will be a good place to do it, right? 224 00:25:20.000 --> 00:25:24.880 start to be able to ask good questions like that and then have this be a wonderful interface. 225 00:25:25.000 --> 00:25:38.000 to explain well if you're going to open a new restaurant that is this kind you want to put it here here here and here's why you might consider that you know that there's some really cool things that this could unlock. 226 00:25:38.000 --> 00:25:45.000 So the creativity is absolutely still there from a human perspective but there's more to it. 227 00:25:45.000 --> 00:25:55.000 All right, so exciting excitement. The next story that's in the in the list here was one from Inc and I don't spend a lot of time reading this particular. 228 00:25:55.000 --> 00:26:16.000 News source but what caught my eye here largely because I had been listening and hearing an awful lot about LVMH in the last year and change is that air maze has made a point that they're really driving unexpectedly no no. 229 00:26:16.000 --> 00:26:26.000 That they're leveraging a artist an artist to create the artwork that is starting to show. 230 00:26:25.000 --> 00:26:51.000 up in their product lines and in their advertising and even more and the thing with air maze and the way that they function with the Atteliers that are crafting the hand created products this is sitting exactly well with what they're doing from a construction on artwork and they're they're zinging where a lot of other places are zagging. 231 00:26:51.000 --> 00:26:54.640 So, I thought this was intriguing. 232 00:26:55.000 --> 00:27:07.000 Certainly, they call it fantastical styles, right, because you're seeing things of octopuses with shoes walking around that could show up on an air-made scarf and the lake. 233 00:27:07.000 --> 00:27:11.000 So, again, not surprising for the brand. 234 00:27:11.000 --> 00:27:24.000 And the further point on the article is, is that even as we're going more and more towards generation of artwork or clip art or other things, the human creativity still looms large. 235 00:27:26.000 --> 00:27:31.839 I thought it was interesting that they talked about the growing desire for hand-drawn work, 236 00:27:31.839 --> 00:27:39.599 authentic human-created work, and I was also really pleased to see that they chose to take this route. 237 00:27:41.500 --> 00:27:43.500 Yeah, I actually 238 00:27:44.660 --> 00:27:49.420 Agree with all of that what acts the angle I took looking at this story 239 00:27:50.299 --> 00:27:55.099 One is some of the artwork kind of looks like it could have been AI created 240 00:27:55.539 --> 00:28:04.720 But it's not right and then the second part is it it threw me back to a lot of the same art style from the late 60s and early 70s 241 00:28:05.319 --> 00:28:09.700 Like the keep on trucking guy with a big foot or the 242 00:28:11.500 --> 00:28:18.220 Marine art from the Beatles or, you know, even even some of the early money Python or 243 00:28:12.500 --> 00:28:20.819 hmm yeah I knew you were going there next 244 00:28:19.740 --> 00:28:21.819 And the final one would be the wall 245 00:28:22.299 --> 00:28:24.779 so there were all these really bizarre kind of 246 00:28:25.420 --> 00:28:27.420 psychedelic freaky artwork 247 00:28:28.140 --> 00:28:35.420 Examples back then especially when you get further down in the article of the two shoes kissing and the the mushroom legs 248 00:28:36.299 --> 00:28:41.460 Table as I call it those to me are like 100% correlated with 249 00:28:41.500 --> 00:28:50.700 Art style from back then and I enjoyed that it kind of was a nice throwback but still felt modern 250 00:28:51.900 --> 00:28:59.339 In another angle with the way they're doing it so very cool. I'm not an Hermes person. I don't own any. I don't buy any 251 00:28:59.980 --> 00:29:02.619 It's not my style, but this art's kind of cool 252 00:29:03.500 --> 00:29:21.500 You surprised me. Interestingly, since we are a game show as well, there was a thing with the Sims recently where they did a deal with coach, and they have got a load of coach items now in the game, which feels very five years ago in terms of brand partnerships. 253 00:29:17.000 --> 00:29:21.720 Hmm. Right. Yeah. 254 00:29:21.500 --> 00:29:31.500 It's also interesting that they have because coaches are very recognisable logo that they wanted in the game, but they don't use English in the game. 255 00:29:31.500 --> 00:29:33.539 Timlish, there's an option. 256 00:29:33.500 --> 00:29:56.500 I was watching the English version and so on, but the other thing that felt quite icky about the whole situation that I was watching clip on yesterday was that they actually sent out a load of coach gear to the influencers, the young YouTubers, to encourage them to make this content. 257 00:29:50.000 --> 00:29:52.000 Influencers? 258 00:29:56.500 --> 00:30:03.500 EA has been through this buyout with Saudi and other... 259 00:30:03.500 --> 00:30:24.500 The brand of the sims in terms of like the playfulness and overall vibe of the game. 260 00:30:20.000 --> 00:30:25.000 I have to ask, how do you say "coach" in "simlish"? 261 00:30:24.500 --> 00:30:33.500 I could not tell you, but I would just say, "Biddly, beep, boop!" or something like that. I've no idea. I've... Yeah, okay. 262 00:30:29.000 --> 00:30:32.000 That's what I thought. That's pretty close, I think. 263 00:30:32.000 --> 00:30:34.000 I think you missed pronounced that last row. 264 00:30:36.500 --> 00:30:48.500 So I'm watching the time and we're getting close but Michael you had a fun piece of art story that you want to include here that is just a really tremendous when you take a look at it so show notes look. 265 00:30:47.500 --> 00:30:51.099 - Yeah, so just 30 seconds on it. 266 00:30:51.099 --> 00:30:52.779 This is actually, it's funny. 267 00:30:52.779 --> 00:30:53.619 When I first glance at it, 268 00:30:53.619 --> 00:30:56.980 I thought they actually knitted the entire image, 269 00:30:56.980 --> 00:31:04.500 but actually it's someone looming together to go back to your usage of loom in the last story. 270 00:31:05.420 --> 00:31:11.140 The old family photos cut up into strips and then loomed into potholders. 271 00:31:11.140 --> 00:31:13.220 And it gives a really kind of a cool vibe. 272 00:31:13.220 --> 00:31:16.220 I just thought it was kind of a cool art project. 273 00:31:13.500 --> 00:31:28.859 It is a cool project I agree and the colour matching or the colour combinations that they use with the loom are to try to fit in with the vibe of the photographs is quite fun. 274 00:31:28.859 --> 00:31:43.140 The last thing I think we want to briefly touch on is this story on TechCrunch, which is talking about Netflix, which having grown up into the right as far as it can do without just continuing to hike prices is now becoming a social app. 275 00:31:43.500 --> 00:31:49.180 I can't tell you how many times back in the EBO days, 20 years ago now. 276 00:31:49.180 --> 00:31:55.380 We talked about second screen experiences and watching shows together and all that crap, 277 00:31:55.380 --> 00:31:58.460 and it's like, I don't need another social platform. 278 00:31:58.460 --> 00:32:07.980 I really don't care what you're watching, unless we're talking about it, and I don't want social platform as TV 279 00:32:00.000 --> 00:32:30.160 it. It's funny you mentioned that because the thing that I'm most excited about recently is that I can now disable YouTube shorts from showing up in search results because YouTube shorts just frustrate me because that's not the way that I consume my video content where you broadly speaking and I know that's a generational thing and lots of other things but Yeah, this seems to be where I know. 280 00:32:16.500 --> 00:32:17.839 Oh, that's a good thing. 281 00:32:28.000 --> 00:32:30.000 Get off your lawn 282 00:32:30.000 --> 00:32:37.000 This seems to be where Netflix is moving with podcasts and comments and so on. 283 00:32:33.920 --> 00:32:35.920 Yeah 284 00:32:36.519 --> 00:32:40.420 I That's 30 more seconds on get off my lawn 285 00:32:41.359 --> 00:32:49.920 Have have either of you guys gotten just totally sick of the number of companies that want some kind of survey 286 00:32:50.500 --> 00:32:52.500 Oh, yeah, now I 287 00:32:51.000 --> 00:32:55.839 I I swear the survey deluge is insane lately 288 00:32:55.859 --> 00:32:58.460 Nearly wrote a blog post on this last year. I 289 00:32:59.660 --> 00:33:08.059 Kept for a month or over a month last year I kept a folder of all of the review requests and survey requests I received from brands 290 00:33:08.819 --> 00:33:12.259 And I with the intent of writing a blog post, you know with midmetrics 291 00:33:12.859 --> 00:33:18.880 You know everything from you've just been to a physical grocery store and bought a piece of broccoli 292 00:33:19.140 --> 00:33:20.339 Could you review it? 293 00:33:20.500 --> 00:33:27.220 To, you know, to all kinds of other things, so I am also extremely frustrated. 294 00:33:21.500 --> 00:33:31.819 Yes. It was broccoli. I saw somebody on Massadon today, 295 00:33:25.000 --> 00:33:28.940 And NPS doesn't write itself though guys, then PS needs 296 00:33:31.819 --> 00:33:50.099 post a thing saying they got a basically a reminder follow-up survey thing to a marketing letter that they didn't respond to. It's like give me a frickin' break. You don't need to instrument everything. Yeah. 297 00:33:41.500 --> 00:33:51.900 Yeah, yeah Got it It's those KPIs It's those KPIs, we're incentivizing the wrong things folks 298 00:33:43.000 --> 00:33:52.460 I've received pre- I've received, yeah. 299 00:33:52.460 --> 00:33:58.200 And I've received three experience survey requests as well. 300 00:33:52.460 --> 00:33:58.700 Yep. Yes. 301 00:33:58.200 --> 00:34:03.839 So before you have the experience, I want you to note that we're going to ask you about the experience. 302 00:34:03.839 --> 00:34:10.019 And you can start your survey right now even before you've had the experience, which is like, are you kidding me? 303 00:34:07.500 --> 00:34:36.179 Yes, like us, like us, follow us on your favorite podcast or Brussels or Brussels sprouts that you got in the grocery store that you have to do an NPS survey on. 304 00:34:10.000 --> 00:34:15.780 So please send your feedback about our show that you've just listened to give us a rating of one to five 305 00:34:13.159 --> 00:34:15.239 Yes. Yeah, we'll have a bit, we'll link. 306 00:34:18.639 --> 00:34:23.880 And next week on on games at work does tech grumps will 307 00:34:24.820 --> 00:34:30.000 I won't be here because I will be in Brussels for the annual Folsdam open source conference, but I will leave you 308 00:34:34.760 --> 00:34:39.920 I will tell you my review of the excellent food that I have in Brussels 309 00:34:37.500 --> 00:34:41.820 >> Excellent. 310 00:34:40.000 --> 00:34:42.000 next week 311 00:34:40.500 --> 00:34:42.500 Yes, you could do it beforehand 312 00:34:41.820 --> 00:34:46.340 >> [LAUGH] 313 00:34:44.059 --> 00:34:46.059 See everybody 314 00:34:45.000 --> 00:34:47.579 [upbeat music] 315 00:34:45.280 --> 00:34:47.280 See ya 316 00:34:46.340 --> 00:34:46.980 >> Bye. 317 00:34:48.000 --> 00:34:53.320 You've been listening to games@work.biz, the podcast about gaming technology and play. 318 00:34:53.320 --> 00:34:56.800 We are part of the blueberry podcasting network, and we'd like to thank the band, 319 00:34:56.800 --> 00:34:59.760 Random Encounters for their song, Big Blue. 320 00:34:59.760 --> 00:35:03.599 You can follow us at our website at games@work.biz.